From Axel.Thimm at ATrpms.net Tue Mar 1 13:42:19 2005 From: Axel.Thimm at ATrpms.net (Axel Thimm) Date: Tue, 1 Mar 2005 14:42:19 +0100 Subject: Current best recommendation for Athlon 64 motherboard? Message-ID: <20050301134219.GH13835@neu.nirvana> Hi, what are currently considered good high-end motherboards for desktop use of Athlon 64s? Chipset recommendations are welcome as well! Thanks! -- Axel.Thimm at ATrpms.net -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 189 bytes Desc: not available URL: From rob.kirkbride at thales-is.com Tue Mar 1 13:52:32 2005 From: rob.kirkbride at thales-is.com (Rob Kirkbride) Date: Tue, 01 Mar 2005 13:52:32 +0000 Subject: Current best recommendation for Athlon 64 motherboard? In-Reply-To: <20050301134219.GH13835@neu.nirvana> References: <20050301134219.GH13835@neu.nirvana> Message-ID: <422473A0.6080703@thales-is.com> Axel Thimm wrote on 01/03/2005 13:42: >Hi, > >what are currently considered good high-end motherboards for desktop >use of Athlon 64s? Chipset recommendations are welcome as well! > > > I bought an Asus A8V-Deluxe and built a machine around that and it works a treat. Very stable - the onboard sound and ethernet work fine too. It does have a Wireless port on it as well but I haven't downloaded the driver from sourceforge because I haven't needed to. I've tried both FC3 and CentOS4 beta on it and it works fine. Rob From crabtrej at otc.edu Tue Mar 1 14:10:27 2005 From: crabtrej at otc.edu (Justin Crabtree) Date: Tue, 01 Mar 2005 08:10:27 -0600 Subject: Current best recommendation for Athlon 64 motherboard? In-Reply-To: <422473A0.6080703@thales-is.com> References: <20050301134219.GH13835@neu.nirvana> <422473A0.6080703@thales-is.com> Message-ID: <422477D3.4040206@otc.edu> Rob Kirkbride wrote: > Axel Thimm wrote on 01/03/2005 13:42: > >> Hi, >> >> what are currently considered good high-end motherboards for desktop >> use of Athlon 64s? Chipset recommendations are welcome as well! >> >> >> > I bought an Asus A8V-Deluxe and built a machine around that and it works > a treat. Very stable - the onboard sound and ethernet work fine too. It > does have a Wireless port on it as well but I haven't downloaded the > driver from sourceforge because I haven't needed to. > I've tried both FC3 and CentOS4 beta on it and it works fine. > > Rob > As far as the best motherboard, that depends on what you need it for. I would recommend looking at one of the new boards based on the NForce4 or K8T890 chipsets. Both of these chipsets support PCI-Express, which if you are building a new computer is the best way to go for future compatibility. Both chipsets have a just about everything you would need in a desktop machine. Check some reviews on reliable sites like AnandTech or Tom's Hardware and good luck. HTH. -- Justin Crabtree Java Programmer Ozarks Technical Community College 447-7533 From dstewart at atl.lmco.com Tue Mar 1 14:30:15 2005 From: dstewart at atl.lmco.com (Doug Stewart) Date: Tue, 01 Mar 2005 09:30:15 -0500 Subject: Current best recommendation for Athlon 64 motherboard? In-Reply-To: <20050301134219.GH13835@neu.nirvana> References: <20050301134219.GH13835@neu.nirvana> Message-ID: <42247C77.9030503@atl.lmco.com> Axel Thimm wrote: > Hi, > > what are currently considered good high-end motherboards for desktop > use of Athlon 64s? Chipset recommendations are welcome as well! > > Thanks! > Once I got my BIOS flashing heartache out of the way (thanks, list!), I've had great success with the Gigabyte K8NS Ultra-939. On-board sound works out of the box (haven't tested its 5.1 capabilities yet...) and RHEL4 installed right out of the box, which was nice. -- ---------- Doug Stewart Systems Administrator/Web Applications Developer Lockheed Martin Advanced Technology Labs dstewart at atl.lmco.com From timothy.leblanc at gmail.com Tue Mar 1 14:46:04 2005 From: timothy.leblanc at gmail.com (Timothy LeBlanc) Date: Tue, 1 Mar 2005 09:46:04 -0500 Subject: Current best recommendation for Athlon 64 motherboard? In-Reply-To: <20050301134219.GH13835@neu.nirvana> References: <20050301134219.GH13835@neu.nirvana> Message-ID: <35d6160405030106465e39c3e6@mail.gmail.com> I agree with Rob. I also bought the Asus A8V-Deluxe and I'm very happy with it. I installed FC3. The only issue I ran into was the board has built in RAID and I wanted to have a mirrored drive set. FC3 see's the drives fine but not as a mirrored set as set up in the BIOS. I have looked around for drivers for my Raid controller but have not found any yet. I can always use the software RAID in FC3 to create my mirror set but for now I'm happy and will continue to look for a driver. Tim On Tue, 1 Mar 2005 14:42:19 +0100, Axel Thimm wrote: > Hi, > > what are currently considered good high-end motherboards for desktop > use of Athlon 64s? Chipset recommendations are welcome as well! > > Thanks! > -- > Axel.Thimm at ATrpms.net > > > -- > amd64-list mailing list > amd64-list at redhat.com > https://www.redhat.com/mailman/listinfo/amd64-list > > > -- My new primary mail account has changed to timothy.leblanc at gmail.com Please make any necessary changes to your address book. From b.j.smith at ieee.org Tue Mar 1 14:50:21 2005 From: b.j.smith at ieee.org (Bryan J. Smith) Date: Tue, 1 Mar 2005 06:50:21 -0800 Subject: Current best recommendation for Athlon 64 motherboard? Message-ID: <2.1-23129203-407-A-OEWW@smtpauth.earthlink.net> Axel Thimm wrote: > what are currently considered good high-end motherboards for desktop > use of Athlon 64s? Chipset recommendations are welcome as well! First off, understand the term "chipset" is rather useless in describing whether an A64/Opteron is a "good" or "poor" design. Mainbaord manufacturers can cut corners regardless of chipset. For more of what I'm talking about, see my 2004 November article in Sys Admin magazine (http://www.samagm.com). Secondly, how high-end? Tyan just released the S2895 ($450) which uses the nVidia Professional 2200 (CPU0) + nVidia Professional 2050 (CPU1) + AMD 8131 (CPU0) chips. It has about 3x the I/O throughput of the normal nForce4 - especially the PCI-X (that is PCI-X - *not* PCIe aka PCI-Express) slots (8x the PCI throughput). There is a darth of PCIe storage controllers so if you go with a 3Ware storage device, you don't want to put it in a lowly, shared PCI slot like most nForce4/KT890 PCIe mainbaords have or it will saturate it's legacy PCI bus. You'll want a mainbaord with multiple PCI-X channels c/o an AMD 8131 or 8132. If you haven't notice by now, you can "mix'n match" different HyperTransport chips in A64/Opteron - and there is no "fixed" concept of a chipset. Otherwise, most of the nForce4 and KT890 mainboards coming out with PCIe are nice and now $100-150. PCIe delivers *dedicated* serial PCI channels - even x1 is 0.25GBps compared to the "shared" 0.125GBps that *all* PCI cards share (except those with PCI-X c/o an added AMD8131/8132 chips -- which offer 8-16x the throughput over 2 channels). The downside is that PCIe is not compatible with PCI, which means it doesn't help segment legacy PCI cards (again, that's PCI-X). But PCIe is much cheaper to implement and provide 4x+ the I/O previously found in PCI-only mainbaords (without the cost of PCI-X -- e.g., mainbaords costing $100-150 instead of $350-500+). Which is where PCIe really matters (and not so much AGP x8 v. PCIe x16 for video - Tom's Hardware showed that PCIe x4 is within 2-3 percent of max performance for today's cards). Can't vouche for the KT890 (ViA always seems to change ATA logic way too much), but other than the NIC, the nForce4 mainbaords have GPL drivers supported by nVidia (and even the NIC has a GPL driver, forcedeth, but it's not supported by nVidia and it seems to be flaky at GbE speeds - hence the proprietary nvnet driver). Most nForce4 mainboards are coming with a 2nd NIC off-nForce that is connected to the shared PCI (or PCI-X in the case of high-end baords), as well as using an off-nForce audio controller in the ALC850 or ALC860 which is supported by ALSA (although the nForce MCP-integrated audio is supported as well, most mainbaord vendors are not opting for it). SATA and other components are supported in *legacy* ATA mode (i.e., turn off AHCI support which is still forthcoming). -- Bryan J. Smith mailto:b.j.smith at ieee.org Currently Mobile From b.j.smith at ieee.org Tue Mar 1 15:03:51 2005 From: b.j.smith at ieee.org (Bryan J. Smith) Date: Tue, 1 Mar 2005 07:03:51 -0800 Subject: Current best recommendation for Athlon 64 motherboard? Message-ID: <2.1-23193357-407-B-OEWW@smtpauth.earthlink.net> Timothy LeBlanc wrote: > The only issue I ran into was the board has built in RAID and I wanted > to have a mirrored drive set. FC3 see's the drives fine but not as a > mirrored set as set up in the BIOS. That's because it is what I call "FRAID" (Fake/Free RAID). There is *0* hardware, it is a standard ATA controller with a trick BIOS. Once the OS boots, it needs a driver with all the *software* RAID logic. Because this logic contains all the RAID IP (typically licensed), there will *never* be a GPL driver. If you're lucky, you can find a binary-only driver for a specific kernel. Now there is a generic GPL RAID logic called "ataraid" and various FRAID interface modules (hptraid, pdcraid and silraid) but they typically don't work - and I've seen people trash their boot. Again, the problem stems from the fact that these FRAID approaches have 0 hardware - they are 100 percent software other then the boot-time 16-bit Int13h disk services. > I have looked around for drivers > for my Raid controller but have not found any yet. I can always use > the software RAID in FC3 to create my mirror set but for now I'm happy > and will continue to look for a driver. LVM/MD in Linux will always *outperform* a FRAID driver. Same deal with LDM in NT5 (200x/XP). For more on LVM v. FRAID v. microcontroller v. ASIC-driven RAID, see my 2004 April Sys Admin article. Unfortunately it is not freely available on-line, but FYI if you get Sys Admin - they include a CD with all past articles if you have a subscription. -- Bryan J. Smith mailto:b.j.smith at ieee.org Currently Mobile From angelus at sangreal.demon.nl Tue Mar 1 22:27:47 2005 From: angelus at sangreal.demon.nl (Angelo Machils) Date: Tue, 01 Mar 2005 23:27:47 +0100 Subject: Plextor DVD problems Message-ID: <4224EC63.7070206@sangreal.demon.nl> Hello there! I have exchanged my Sony DVD reader with a Plextor PX-116A and also installed a Plextor PX-716A DVD writer. When I put a burned DVD into either one of these drives they show up on the desktop, but when I open them, they are seen as empty (blank) disks. I can however mount them manually as my normal user and go to the mountpoint in a terminal. Only there I can see the contents of the disk. Other times I can't even mount the disk at all. All this only happens with DVD's, not CD's. My /etc/fstab entries are: /dev/hdc /media/cdrecorder auto pamconsole,exec,noauto,fscontext=system_u:object_r:removable_t,managed 0 0 /dev/hdb /media/cdrom auto pamconsole,exec,noauto,fscontext=system_u:object_r:removable_t,managed 0 0 I'm running Gnome on FC3 x86-64, fully updated. Kind regards, Angelo From angelus at sangreal.demon.nl Tue Mar 1 22:33:01 2005 From: angelus at sangreal.demon.nl (Angelo Machils) Date: Tue, 01 Mar 2005 23:33:01 +0100 Subject: long wait for log out window Message-ID: <4224ED9D.4030107@sangreal.demon.nl> Angelo Machils wrote: >> >> >> Message: 17 Date: Sun, 27 Feb 2005 15:31:11 -0600 From: Jonathan >> Berry Subject: Re: long wait for log out window >> To: For users of Fedora Core releases >> Message-ID: <8767947e050227133133616881 at mail.gmail.com> Content-Type: >> text/plain; charset=US-ASCII On Sun, 27 Feb 2005 22:12:53 +0100, >> Angelo Machils wrote: >> >>>> Hello there! >>>> >>>> Since a while (since 2.6.10-1.766_FC3???) I have noticed that when I >>>> choose 'Log Out' from the window in Gnome (running on FC3 x86-64, >>>> fully >>>> updated) it takes a very long time before the actual window (where you >>>> can choose either to logout, halt or restart the system) appears. I'm >>>> talking several minutes here, and during that time I can also choose >>>> nothing from the menu, but for the rest the system works without a >>>> fault. Top says the system is pretty much idle, so it's not a >>>> performance problem. >>>> Does anyone else have this problem? Is it a known bug? >>>> >>>> Kind regards, Angelo >>> >>> >>> >> >> >> >> Hi Angelo, >> No, I have not seen anything like this. In fact, I just checked and >> with updatedb banging on the disk, it only took very little time >> (almost no delay) to bring up the logout window in Gnome for me >> (though it took a considerably longer time for the menu to actually >> pop up to click on logout :)). I'm running 766_FC3 on x86_64 as well. >> You can also bring up this dialog with the command: >> gnome-session-save --kill >> You might try doing an strace on this to see if you can find where it >> is hanging: >> strace gnome-session-save --kill >> This should scroll by a lot of stuff. Take note of any places that it >> hangs. For me, stuff scrolls by until the dialog comes up. You can >> then run it again and direct the output to a file: >> strace gnome-session-save --kill 2> logout.txt >> which will put everything in logout.txt so you can post info to the >> list or see more of what is going on, just remember where you noticed >> it hang before. In another terminal, you can also run "tail -f >> logout.txt" to see what is being printed to the file so you could >> perhaps run it only once. If nothing looks obvious, send more info to >> the list. >> >> Jonathan >> > > Thank you for the reply. Weird thing, when I use the command, there is > no waiting for the dialog to popup and also the response to the choice > I take in the dialog is immidiate..... Only when I choose it from the > menu I experience this delay. > > Kind regards, Angelo > Seems I spoke too soon :-) Update: Sometimes it's immidiate, and sometimes it takes as long as from the menu. The strace shows that is stops at: connect(16, {sa_family=AF_INET, sin_port=htons(16001), sin_addr=inet_addr("127.0.0.1")}, 16 When the dialog *does* show up, the same line says: connect(16, {sa_family=AF_INET, sin_port=htons(16001), sin_addr=inet_addr("127.0.0.1")}, 16) = -1 ETIMEDOUT (Connection timed out) I also have an Athlon XP with FC3 (also fully updated) and this one doesn't show this behaviour.... Kind regards, Angelo From cochranb at speakeasy.net Tue Mar 1 23:32:56 2005 From: cochranb at speakeasy.net (Robert L Cochran) Date: Tue, 01 Mar 2005 18:32:56 -0500 Subject: Current best recommendation for Athlon 64 motherboard? In-Reply-To: <2.1-23193357-407-B-OEWW@smtpauth.earthlink.net> References: <2.1-23193357-407-B-OEWW@smtpauth.earthlink.net> Message-ID: <4224FBA8.1080107@speakeasy.net> Bryan J. Smith wrote: > Timothy LeBlanc wrote: > >> The only issue I ran into was the board has built in RAID and I wanted >> to have a mirrored drive set. FC3 see's the drives fine but not as a >> mirrored set as set up in the BIOS. > > > That's because it is what I call "FRAID" (Fake/Free RAID). > There is *0* hardware, it is a standard ATA controller with a trick BIOS. > Once the OS boots, it needs a driver with all the *software* RAID logic. > Because this logic contains all the RAID IP (typically licensed), > there will *never* be a GPL driver. > If you're lucky, you can find a binary-only driver for a specific kernel. > > Now there is a generic GPL RAID logic called "ataraid" and various > FRAID interface modules (hptraid, pdcraid and silraid) but they > typically don't work - and I've seen people trash their boot. > Again, the problem stems from the fact that these FRAID approaches > have 0 hardware - they are 100 percent software other then the > boot-time 16-bit Int13h disk services. > >> I have looked around for drivers >> for my Raid controller but have not found any yet. I can always use >> the software RAID in FC3 to create my mirror set but for now I'm >> happy and will continue to look for a driver. > > > LVM/MD in Linux will always *outperform* a FRAID driver. > Same deal with LDM in NT5 (200x/XP). > > For more on LVM v. FRAID v. microcontroller v. ASIC-driven RAID, see > my 2004 April Sys Admin article. > Unfortunately it is not freely available on-line, but FYI if you get > Sys Admin - they include a CD with all past articles if you have a > subscription. > -- > Bryan J. Smith mailto:b.j.smith at ieee.org > Currently Mobile > I agree with Bryan....I've been there before and had the same RAID troubles. Solution was to change my motherboard from Asus to Gigabyte and use Linux software RAID. It wasn't that hard getting RAID working and it was fun for a while, but then I coveted the mirror drive for other things and said the heck with RAID and went back to a single drive configuration (no RAID.) But my big problem at that time was I foolishly invested in (very expensive) RDRAM, and did so just as RDRAM was going downhill with avalanche speed. I had been very influenced by Maximim PC magazine when making my buying choices. Now I'm considerably wiser. Bob Cochran From b.j.smith at ieee.org Wed Mar 2 00:24:59 2005 From: b.j.smith at ieee.org (Bryan J. Smith) Date: Tue, 1 Mar 2005 16:24:59 -0800 Subject: Current best recommendation for Athlon 64 motherboard? Message-ID: <2.1-24275035-410-A-OEWW@smtpauth.earthlink.net> Robert L Cochran wrote: > I agree with Bryan....I've been there before and had the same RAID > troubles. Solution was to change my motherboard from Asus to Gigabyte > and use Linux software RAID. It wasn't that hard getting RAID working > and it was fun for a while, but then I coveted the mirror drive for > other things and said the heck with RAID and went back to a single > drive configuration (no RAID.) > > But my big problem at that time was I foolishly invested in (very > expensive) RDRAM, and did so just as RDRAM was going downhill with > avalanche speed. I had been very influenced by Maximim PC magazine > when making my buying choices. Now I'm considerably wiser. 3Ware Escalade 7006-2 and 8006-2 cards are $100. Now you can find older Escalade 6200 and 6410 on eBay for under $50. The 6.9 firmware even updates them with 48-bit addressing (greater than 137GB/128GiB support). But the newer 3DM2 (3Ware Disk Manager version 2) doesn't work with those cards. And the older 3DM interfaces are deprecated in the new 2.6 kernel. -- Bryan J. Smith mailto:b.j.smith at ieee.org Currently Mobile From eugen at leitl.org Wed Mar 2 18:28:21 2005 From: eugen at leitl.org (Eugen Leitl) Date: Wed, 2 Mar 2005 19:28:21 +0100 Subject: Current best recommendation for Athlon 64 motherboard? In-Reply-To: <2.1-24275035-410-A-OEWW@smtpauth.earthlink.net> References: <2.1-24275035-410-A-OEWW@smtpauth.earthlink.net> Message-ID: <20050302182821.GG13336@leitl.org> On Tue, Mar 01, 2005 at 04:24:59PM -0800, Bryan J. Smith wrote: > Robert L Cochran wrote: > >I agree with Bryan....I've been there before and had the same RAID > >troubles. Solution was to change my motherboard from Asus to Gigabyte > >and use Linux software RAID. It wasn't that hard getting RAID working > >and it was fun for a while, but then I coveted the mirror drive for > >other things and said the heck with RAID and went back to a single > >drive configuration (no RAID.) > > > >But my big problem at that time was I foolishly invested in (very > >expensive) RDRAM, and did so just as RDRAM was going downhill with > >avalanche speed. I had been very influenced by Maximim PC magazine > >when making my buying choices. Now I'm considerably wiser. > > 3Ware Escalade 7006-2 and 8006-2 cards are $100. They will perform very poorly in comparison to native /dev/md0, and you'll have to buy a couple, to have one as spare. No such issues with soft RAID, which can be read with any vanilla motherboard. Sure, an appliance with acoustic alarm, hotplug and automatic rebuild saves some major hassles. > Now you can find older Escalade 6200 and 6410 on eBay for under $50. > The 6.9 firmware even updates them with 48-bit addressing (greater than > 137GB/128GiB support). > But the newer 3DM2 (3Ware Disk Manager version 2) doesn't work with > those cards. > And the older 3DM interfaces are deprecated in the new 2.6 kernel. I don't see any value in cheap hardware RAID. Too bad Lustre is not yet ready for large-scale RAIC. -- Eugen* Leitl leitl ______________________________________________________________ ICBM: 48.07078, 11.61144 http://www.leitl.org 8B29F6BE: 099D 78BA 2FD3 B014 B08A 7779 75B0 2443 8B29 F6BE http://moleculardevices.org http://nanomachines.net -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 198 bytes Desc: not available URL: From ken.snider at datawire.net Wed Mar 2 18:55:46 2005 From: ken.snider at datawire.net (Ken Snider) Date: Wed, 02 Mar 2005 13:55:46 -0500 Subject: Current best recommendation for Athlon 64 motherboard? In-Reply-To: <20050302182821.GG13336@leitl.org> References: <2.1-24275035-410-A-OEWW@smtpauth.earthlink.net> <20050302182821.GG13336@leitl.org> Message-ID: <42260C32.9030006@datawire.net> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Eugen Leitl wrote: | They will perform very poorly in comparison to native /dev/md0, and you'll | have to buy a couple, to have one as spare. No such issues with soft RAID, | which can be read with any vanilla motherboard. | | Sure, an appliance with acoustic alarm, hotplug and automatic rebuild saves | some major hassles. Some other considerations. A drive blowing on a hardware RAID rarely locks up the RAID controller - the same can not be said of most onboard ATA controllers, and some SCSI controllers as well. RAID controllers that have battery-backed-up NVRAM can complete writes much more reliably in failure scenarios - and ensure that even in a power-loss situation, or OS lockup, data will get flushed to disk, though this may or may not matter to you given your usage and filesystem. lastly, many RAID controllers offer far more ports than you could hope to get on a mainboard - the 3ware 8 and 12 drive controllers actually have 8 or 12 on-board channels, I've never seen a motherboard with more than 5 channels (you wouldn't want to use master/slave configurations on these controllers, in fact you don't even have that option with the 3ware cards). I have software RAID deployed on over 50 boxes here, but on the ones I *really* care about, I use, at minimum, hardware mirroring, to offload the rebuild process in the case of a bad disk, as well as to reduce the risk that a bad disk may hose the underlying IDE driver. - -- Ken Snider -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.3 (GNU/Linux) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org iD8DBQFCJgwy409OqHEpU1kRAgmgAKCWIPRn/lkz9Zm+q9l5ay/NS23p0gCfdfcA M/TNRCq5m5bvulrHTRIEjFU= =pkZw -----END PGP SIGNATURE----- From b.j.smith at ieee.org Wed Mar 2 20:35:59 2005 From: b.j.smith at ieee.org (Bryan J. Smith) Date: Wed, 2 Mar 2005 12:35:59 -0800 Subject: Current best recommendation for Athlon 64 motherboard? Message-ID: <2.1-24435512-407-A-OEWW@smtpauth.earthlink.net> Ken Snider wrote: > Some other considerations. > A drive blowing on a hardware RAID rarely locks up the RAID controller > - the same can not be said of most onboard ATA controllers, Yes, and that issue with hotplug includes "FRAID" cards. FRAID cards often lack hotplug, despite marketing. In a nutshell, you need some *intelligence* between the rest of the system and the drive controllers. That intelligence then handles hotplug. > and some SCSI controllers as well. Unlike ATA, SCSI *always* comes with intelligence, so its more of an issue with bus disconnect. But regardless of whether ATA or SCSI drive controllers are used, if there is an intelligence on the card, then the rest of the system doesn't see the hardware, and hotplug becomes easy. This is not like FRAID which still allows the system to have direct access to the drives, using the FRAID driver to "hide" them. > RAID controllers that have battery-backed-up NVRAM can complete writes much > more reliably in failure scenarios - and ensure that even in a > power-loss situation, or OS lockup, data will get flushed to disk, > though this may or may not matter to you given your usage and > filesystem. Battery-backed RAM is only necessary for DRAM, not SRAM. Hence why 3Ware Escalade 9000 series have batteries (because they use DRAM as well as SRAM) whereas earlier Escalades do not (SRAM-only). Of course, you should *never* cut the power to the system for more than a few seconds, otherwise the SRAM loses their states. But SRAM doesn't need to be refreshed like DRAM, so only a tiny amount of power needs to be applied (which is easily done). > lastly, many RAID controllers offer far more ports than you could hope > to get on a mainboard - the 3ware 8 and 12 drive controllers actually > have 8 or 12 on-board channels, I've never seen a motherboard with > more than 5 > channels (you wouldn't want to use master/slave configurations on > these controllers, in fact you don't even have that option with the > 3ware cards). Actually, the nVidia Pro 2200+2050 has 8 SATA ports. And the new AHCI firmware allows up to 32 ports to be intelligently controlled in software (including using queuing on the drive). But even AHCI is still not intelligence, and that means 2-12x the data load on your interconnects. It's not the XORs that kill, but passing all the data through the CPU over its interconnects and back. > I have software RAID deployed on over 50 boxes here, but on the ones I > *really* care about, I use, at minimum, hardware mirroring, to offload > the > rebuild process in the case of a bad disk, as well as to reduce the risk > that a bad disk may hose the underlying IDE driver. Commodity drives are not designed for 24x7 operation, but only 8 hours x 50,000 restarts. Hitachi and many manufacturers rate their drives at 14x5 maximum weekly usage. Again, that is for *commodity* drives. Enterprise drives are rated differently. The interface has nothing to do with drive quality, although typically most ATA models are commodity design/tested. But some SCSI drives do come off the same lines as ATA/SATA. And a few, select ATA/SATA drives come of enterprise SCSI lines too. -- Bryan J. Smith mailto:b.j.smith at ieee.org Currently Mobile From maurice at harddata.com Thu Mar 3 02:51:08 2005 From: maurice at harddata.com (Maurice Hilarius) Date: Wed, 02 Mar 2005 19:51:08 -0700 Subject: Current best recommendation for Athlon 64 motherboard? In-Reply-To: <422477D3.4040206@otc.edu> References: <20050301134219.GH13835@neu.nirvana> <422473A0.6080703@thales-is.com> <422477D3.4040206@otc.edu> Message-ID: <42267B9B.40000@harddata.com> Justin Crabtree wrote: > Rob Kirkbride wrote: > >> Axel Thimm wrote on 01/03/2005 13:42: >> >>> Hi, >>> >>> what are currently considered good high-end motherboards for desktop >>> use of Athlon 64s? Chipset recommendations are welcome as well! >>> >>> >>> >> I bought an Asus A8V-Deluxe and built a machine around that and it >> works a treat. Very stable - the onboard sound and ethernet work fine >> too. It does have a Wireless port on it as well but I haven't >> downloaded the driver from sourceforge because I haven't needed to. >> I've tried both FC3 and CentOS4 beta on it and it works fine. >> >> Rob >> > > > As far as the best motherboard, that depends on what you need it for. > I would recommend looking at one of the new boards based on the > NForce4 or K8T890 chipsets. Both of these chipsets support > PCI-Express, which if you are building a new computer is the best way > to go for future compatibility. Both chipsets have a just about > everything you would need in a desktop machine. Check some reviews on > reliable sites like AnandTech or Tom's Hardware and good luck. HTH. > I second this. The nForce4 is the current leader, mainly because the VIA is so darned new. I think the ASUS and MSI SLI and non SLI versions are the current performance and reliability leaders. The Gigabyte, ECS, and so on are cheaper, but I think at this time you "get what you pay for". With our best regards, Maurice W. Hilarius Telephone: 01-780-456-9771 Hard Data Ltd. FAX: 01-780-456-9772 11060 - 166 Avenue email:maurice at harddata.com Edmonton, AB, Canada http://www.harddata.com/ T5X 1Y3 From Axel.Thimm at ATrpms.net Thu Mar 3 03:22:42 2005 From: Axel.Thimm at ATrpms.net (Axel Thimm) Date: Thu, 3 Mar 2005 04:22:42 +0100 Subject: Current best recommendation for Athlon 64 motherboard? In-Reply-To: <42267B9B.40000@harddata.com> References: <20050301134219.GH13835@neu.nirvana> <422473A0.6080703@thales-is.com> <422477D3.4040206@otc.edu> <42267B9B.40000@harddata.com> Message-ID: <20050303032242.GD5701@neu.nirvana> On Wed, Mar 02, 2005 at 07:51:08PM -0700, Maurice Hilarius wrote: > Justin Crabtree wrote: > >Rob Kirkbride wrote: > >>Axel Thimm wrote on 01/03/2005 13:42: > >>>what are currently considered good high-end motherboards for desktop > >>>use of Athlon 64s? Chipset recommendations are welcome as well! > >>I bought an Asus A8V-Deluxe and built a machine around that and it > >>works a treat. Very stable - the onboard sound and ethernet work > >>fine too. It does have a Wireless port on it as well but I haven't > >>downloaded the driver from sourceforge because I haven't needed > >>to. I've tried both FC3 and CentOS4 beta on it and it works fine. > >As far as the best motherboard, that depends on what you need it > >for. The purpose of these machines are to become Linux desktops and/or number crunchers for scientific staff. Ethernet and IDE are important, graphics and sound is not (although some staff members may think otherwise :) Also stability and projected long life (e.g. good manufacturing quality, low dissipation etc.) to keep the support from the IT staff minimal. In a nutshell: o needs to run on Linux ;) o needs to run o needs to run o needs to run o (optional: good performance in disk/network/memory IO) > >I would recommend looking at one of the new boards based on the > >NForce4 or K8T890 chipsets. Both of these chipsets support > >PCI-Express, which if you are building a new computer is the best > >way to go for future compatibility. Both chipsets have a just > >about everything you would need in a desktop machine. Check some > >reviews on reliable sites like AnandTech or Tom's Hardware and good > >luck. HTH. > I second this. > The nForce4 is the current leader, mainly because the VIA is so > darned new. I think the ASUS and MSI SLI and non SLI versions are > the current performance and reliability leaders. > The Gigabyte, ECS, and so on are cheaper, but I think at this time > you "get what you pay for". Thanks for all suggestions. I remember (from the dark ages) that the nforce Linux drivers were at first non-existent, then closed source. I'd prefer a motherboard that runs out of the box on a recent kernel, or one that has a chance to do so in a couple of kernel releases (patching kernel sources or building open source kernel modules externally wouldn't break my leg). But if nforce4 is the best recommendation, we'd even bite the bullet and package up closed source nforce4 drivers. Thanks! -- Axel.Thimm at ATrpms.net -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 189 bytes Desc: not available URL: From bill at cse.ucdavis.edu Thu Mar 3 05:45:30 2005 From: bill at cse.ucdavis.edu (Bill Broadley) Date: Wed, 2 Mar 2005 21:45:30 -0800 Subject: Current best recommendation for Athlon 64 motherboard? In-Reply-To: <20050303032242.GD5701@neu.nirvana> References: <20050301134219.GH13835@neu.nirvana> <422473A0.6080703@thales-is.com> <422477D3.4040206@otc.edu> <42267B9B.40000@harddata.com> <20050303032242.GD5701@neu.nirvana> Message-ID: <20050303054530.GA8669@cse.ucdavis.edu> > I remember (from the dark ages) that the nforce Linux drivers were at > first non-existent, then closed source. I'd prefer a motherboard that > runs out of the box on a recent kernel, or one that has a chance to do > so in a couple of kernel releases (patching kernel sources or building > open source kernel modules externally wouldn't break my leg). > > But if nforce4 is the best recommendation, we'd even bite the bullet > and package up closed source nforce4 drivers. Nforce4 drivers are neither closed source or non-existant. I got a new MSI nforce4 board and installed FC3 x86-64. Audio works, SATA works, GigE works. I haven't tried firewire yet. -- Bill Broadley Computational Science and Engineering UC Davis From jlabadie at acm.org Fri Mar 4 03:44:43 2005 From: jlabadie at acm.org (Jon LaBadie) Date: Thu, 03 Mar 2005 22:44:43 -0500 Subject: managing PERC raid Message-ID: Hardware is Dell 670n workstation with Adaptec 79xx SCSI controller and Dell's PERC 320/DC raid controller. OS is RHEL 3 WS running in x86_64 mode. What tools, if any, are available for managing the raid system? As supplied, the 4 drives are in RAID 10, with just swap and a single, very large root partition. I'd like to split it into several, more manageable partitions. Figure I would approach it as split the mirror, reconfigure one side, copy things from the running side, make sure the split side will boot, then work on the other side. But I don't see any tools to work with the hardware raid. From maurice at harddata.com Fri Mar 4 06:21:02 2005 From: maurice at harddata.com (Maurice Hilarius) Date: Thu, 03 Mar 2005 23:21:02 -0700 Subject: managing PERC raid In-Reply-To: References: Message-ID: <4227FE4E.7010008@harddata.com> Jon LaBadie wrote: > Hardware is Dell 670n workstation with Adaptec 79xx SCSI controller > and Dell's PERC 320/DC raid controller. > > OS is RHEL 3 WS running in x86_64 mode. > > What tools, if any, are available for managing the raid system? > > As supplied, the 4 drives are in RAID 10, with just swap and > a single, very large root partition. > > I'd like to split it into several, more manageable partitions. > Figure I would approach it as split the mirror, reconfigure > one side, copy things from the running side, make sure the > split side will boot, then work on the other side. > > But I don't see any tools to work with the hardware raid. > Instead of that I suggest software RAID. Use mdadm. With our best regards, Maurice W. Hilarius Telephone: 01-780-456-9771 Hard Data Ltd. FAX: 01-780-456-9772 11060 - 166 Avenue email:maurice at harddata.com Edmonton, AB, Canada http://www.harddata.com/ T5X 1Y3 From eugen at leitl.org Fri Mar 4 08:15:03 2005 From: eugen at leitl.org (Eugen Leitl) Date: Fri, 4 Mar 2005 09:15:03 +0100 Subject: adding storage to v20z Message-ID: <20050304081502.GZ13336@leitl.org> I'm trying to add storage to a v20z (a 1U dual-Opteron system running SLES8) on the cheap, for storage and some minor ORACLEing. Both drive slots are full, and just swapping the larger drive would take two days worth of installation, so that's not so good. I tried external USB drives yesterday, but the single port seems to be USB1.1 (?) There's no external SCSI, so I presume I'd have to get a PCI-X U320 SCSI host (FC is way out of price range) and run a few drives in an external enclosure. Any other options? NFS mount (can Oracle do this?), iSCSI, SATA, iSATA? Firewire? (remember, I only have PCI-X). Any suggestions are appreciated. -- Eugen* Leitl leitl ______________________________________________________________ ICBM: 48.07078, 11.61144 http://www.leitl.org 8B29F6BE: 099D 78BA 2FD3 B014 B08A 7779 75B0 2443 8B29 F6BE http://moleculardevices.org http://nanomachines.net -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 198 bytes Desc: not available URL: From justin.conover at gmail.com Fri Mar 4 18:59:33 2005 From: justin.conover at gmail.com (Justin Conover) Date: Fri, 4 Mar 2005 12:59:33 -0600 Subject: Current best recommendation for Athlon 64 motherboard? In-Reply-To: <20050303054530.GA8669@cse.ucdavis.edu> References: <20050301134219.GH13835@neu.nirvana> <422473A0.6080703@thales-is.com> <422477D3.4040206@otc.edu> <42267B9B.40000@harddata.com> <20050303032242.GD5701@neu.nirvana> <20050303054530.GA8669@cse.ucdavis.edu> Message-ID: Asus A8V-Deluxe/amd64 3000 CentOS 4rc1 x86_64 Running great MSI K8N Neo2 Platinum, AMD64 3200+ Fedora Core 3 x86_64 Running great On Wed, 2 Mar 2005 21:45:30 -0800, Bill Broadley wrote: > > I remember (from the dark ages) that the nforce Linux drivers were at > > first non-existent, then closed source. I'd prefer a motherboard that > > runs out of the box on a recent kernel, or one that has a chance to do > > so in a couple of kernel releases (patching kernel sources or building > > open source kernel modules externally wouldn't break my leg). > > > > But if nforce4 is the best recommendation, we'd even bite the bullet > > and package up closed source nforce4 drivers. > > Nforce4 drivers are neither closed source or non-existant. I got a new > MSI nforce4 board and installed FC3 x86-64. Audio works, SATA works, > GigE works. > > I haven't tried firewire yet. > > -- > Bill Broadley > Computational Science and Engineering > UC Davis > > -- > amd64-list mailing list > amd64-list at redhat.com > https://www.redhat.com/mailman/listinfo/amd64-list > From forrest at ece.ucsb.edu Fri Mar 4 19:47:32 2005 From: forrest at ece.ucsb.edu (Forrest Brewer) Date: Fri, 04 Mar 2005 11:47:32 -0800 Subject: AMD-64 large memory... Message-ID: <1109965652.8262.2.camel@canopus.ece.ucsb.edu> I am interested in the large memory possible in this processor design. Does anyone have AMD-64 cards with more than 2GB memory running linux? Any issues? -- Forrest Brewer From tjb at unh.edu Fri Mar 4 20:50:45 2005 From: tjb at unh.edu (Thomas J. Baker) Date: Fri, 04 Mar 2005 15:50:45 -0500 Subject: Current best recommendation for Athlon 64 motherboard? In-Reply-To: References: <20050301134219.GH13835@neu.nirvana> <422473A0.6080703@thales-is.com> <422477D3.4040206@otc.edu> <42267B9B.40000@harddata.com> <20050303032242.GD5701@neu.nirvana> <20050303054530.GA8669@cse.ucdavis.edu> Message-ID: <1109969445.27233.32.camel@wintermute.sr.unh.edu> On Fri, 2005-03-04 at 12:59 -0600, Justin Conover wrote: > Asus A8V-Deluxe/amd64 3000 CentOS 4rc1 x86_64 Running great > MSI K8N Neo2 Platinum, AMD64 3200+ Fedora Core 3 x86_64 Running great > I also have an MSI K8N Neo2 Platinum, AMD64 3500+, FC3 x86_64 working well. My only problem is that grub refuses to boot WinXP. It complains about "unsupported architecture" or something. I've tried everything I could find on the net and it just won't work. tjb --- ======================================================================= | Thomas Baker email: tjb at unh.edu | | Systems Programmer | | Research Computing Center voice: (603) 862-4490 | | University of New Hampshire fax: (603) 862-1761 | | 332 Morse Hall | | Durham, NH 03824 USA http://wintermute.sr.unh.edu/~tjb | ======================================================================= From joshua at iwsp.com Fri Mar 4 23:24:40 2005 From: joshua at iwsp.com (Joshua Jensen) Date: Fri, 4 Mar 2005 18:24:40 -0500 Subject: AMD-64 large memory... In-Reply-To: <1109965652.8262.2.camel@canopus.ece.ucsb.edu> References: <1109965652.8262.2.camel@canopus.ece.ucsb.edu> Message-ID: <20050304232440.GA3903@iwsp.com> We have a few machines with 32 gigs of memory... I've not heard of any problems with them, no. Joshua On Fri, Mar 04, 2005 at 11:47:32AM -0800, Forrest Brewer wrote: > I am interested in the large memory possible in this processor design. > Does anyone have AMD-64 cards with more than 2GB memory running linux? > Any issues? > -- > Forrest Brewer > > -- > amd64-list mailing list > amd64-list at redhat.com > https://www.redhat.com/mailman/listinfo/amd64-list -- Joshua Jensen joshua at iwsp.com "If God didn't want us to eat animals, why did he make them out of meat?" From cochranb at speakeasy.net Fri Mar 4 23:31:21 2005 From: cochranb at speakeasy.net (Robert L Cochran) Date: Fri, 04 Mar 2005 18:31:21 -0500 Subject: AMD-64 large memory... In-Reply-To: <20050304232440.GA3903@iwsp.com> References: <1109965652.8262.2.camel@canopus.ece.ucsb.edu> <20050304232440.GA3903@iwsp.com> Message-ID: <4228EFC9.5090501@speakeasy.net> I realize this is going to sound stupid, but what motherboard, exactly, can physically take 32 Gb of memory? What type of memory and where do you get it at a reasonable price? It must cost a fortune for that much. My wife has some applications that need memory beyond 4 Gb so I'm interested in what you have to say. Thanks Bob Cochran Joshua Jensen wrote: >We have a few machines with 32 gigs of memory... I've not heard of any >problems with them, no. > >Joshua > > > >On Fri, Mar 04, 2005 at 11:47:32AM -0800, Forrest Brewer wrote: > > >>I am interested in the large memory possible in this processor design. >>Does anyone have AMD-64 cards with more than 2GB memory running linux? >>Any issues? >>-- >>Forrest Brewer >> >>-- >>amd64-list mailing list >>amd64-list at redhat.com >>https://www.redhat.com/mailman/listinfo/amd64-list >> >> > > > -------------- next part -------------- A non-text attachment was scrubbed... Name: smime.p7s Type: application/x-pkcs7-signature Size: 3260 bytes Desc: S/MIME Cryptographic Signature URL: From dsavage at peaknet.net Sat Mar 5 00:11:16 2005 From: dsavage at peaknet.net (dsavage at peaknet.net) Date: Fri, 4 Mar 2005 18:11:16 -0600 (CST) Subject: AMD-64 large memory... Message-ID: <32999.206.80.72.54.1109981476.squirrel@www.peaknet.net> On Fri, 2005-03-04 at 18:31 -0500, Robert L Cochran wrote: > I realize this is going to sound stupid, but what motherboard, exactly, > can physically take 32 Gb of memory? What type of memory and where do > you get it at a reasonable price? It must cost a fortune for that much. > My wife has some applications that need memory beyond 4 Gb so I'm > interested in what you have to say. Bob, AFAIK, with ten memory slots the Tyan quad-Opteron board (http://www.tyan.com/products/html/thunderk8qspro.html) is the current king of the hill. You'd need 4G memory parts ($$$!!!) to get to 32G on this or any of the many boards with eight memory slots. But that's all theoretical because when maxed out this mobo is a true yacht. You know...if you have to ask how much it costs, you can't afford it. -- Doc Robert G. (Doc) Savage, BSE(EE), CISSP, RHCE | Fairview Heights, IL RHEL4/ES on Tyan S2468UGN w/3G, dual Athlon MP 2800+, 1.1T RAID5 "Perfection is the enemy of good enough." -- Admiral of the Fleet Sergei G. Gorshkov From maurice at harddata.com Sat Mar 5 05:31:37 2005 From: maurice at harddata.com (Maurice Hilarius) Date: Fri, 04 Mar 2005 22:31:37 -0700 Subject: AMD-64 large memory... In-Reply-To: <1109965652.8262.2.camel@canopus.ece.ucsb.edu> References: <1109965652.8262.2.camel@canopus.ece.ucsb.edu> Message-ID: <42294439.9010605@harddata.com> Forrest Brewer wrote: >I am interested in the large memory possible in this processor design. >Does anyone have AMD-64 cards with more than 2GB memory running linux? >Any issues? > > For example 32GB? No problems. Used FC, Tyan S4882 board (AMD chipset) Something more modest? 16GB on S2892 AMD chipset) 16GB on S2891 (nVidia nForce4 chipset) 4GB on Tyan S2850, S2875, ASUS and MSI socket 939 board with VIA and nForce chipsets. No issues. With our best regards, Maurice W. Hilarius Telephone: 01-780-456-9771 Hard Data Ltd. FAX: 01-780-456-9772 11060 - 166 Avenue email:maurice at harddata.com Edmonton, AB, Canada http://www.harddata.com/ T5X 1Y3 From maurice at harddata.com Sat Mar 5 05:38:22 2005 From: maurice at harddata.com (Maurice Hilarius) Date: Fri, 04 Mar 2005 22:38:22 -0700 Subject: AMD-64 large memory... In-Reply-To: <4228EFC9.5090501@speakeasy.net> References: <1109965652.8262.2.camel@canopus.ece.ucsb.edu> <20050304232440.GA3903@iwsp.com> <4228EFC9.5090501@speakeasy.net> Message-ID: <422945CE.9030006@harddata.com> Robert L Cochran wrote: > I realize this is going to sound stupid, but what motherboard, > exactly, can physically take 32 Gb of memory? What type of memory and > where do you get it at a reasonable price? It must cost a fortune for > that much. My wife has some applications that need memory beyond 4 Gb > so I'm interested in what you have to say. > > Thanks > > Bob Cochran Most dual CPU boards have 8 sockets, and we commonly put 16GB in them. Using 2GB modules/ We build 4 CPU machines, and these boards have 16 slots. What is "reasonable price" ? We currently sell ECC Registered DDR400 low profile 1GB DIMMs (Corsair CM72SD1024RLP-3200/S) for $205 2GB are $475 For Socket939 board you do not need the ECC modules, and good quality branded parts are 10% to 15% less. And you can always find cheaper stuff, but it might be less compatible or reliable.. With our best regards, Maurice W. Hilarius Telephone: 01-780-456-9771 Hard Data Ltd. FAX: 01-780-456-9772 11060 - 166 Avenue email:maurice at harddata.com Edmonton, AB, Canada http://www.harddata.com/ T5X 1Y3 From eugen at leitl.org Sat Mar 5 07:50:56 2005 From: eugen at leitl.org (Eugen Leitl) Date: Sat, 5 Mar 2005 08:50:56 +0100 Subject: AMD-64 large memory... In-Reply-To: <4228EFC9.5090501@speakeasy.net> References: <1109965652.8262.2.camel@canopus.ece.ucsb.edu> <20050304232440.GA3903@iwsp.com> <4228EFC9.5090501@speakeasy.net> Message-ID: <20050305075056.GF13336@leitl.org> On Fri, Mar 04, 2005 at 06:31:21PM -0500, Robert L Cochran wrote: > I realize this is going to sound stupid, but what motherboard, exactly, > can physically take 32 Gb of memory? What type of memory and where do We have a SunFire v20z ( Newisys 2100 http://www.aceshardware.com/read.jsp?id=55000254 ) which can take up to 16 GBytes DDR-333 ECC (four memory slots for each CPU). It runs with 6 GBytes with no problems (Oracle, custom search engine working out of memory). I think the hardware is certified both for SLES8SP3 and RHEL3, but do your own research before buying. Don't buy it from Sun, unless you have special developer discounts. There are alternative sources for it. The hardware is stable, performant, and relatively cheap (there are cheaper systems, but these have reliability problems, see Beowulf archives). > you get it at a reasonable price? It must cost a fortune for that much. > My wife has some applications that need memory beyond 4 Gb so I'm > interested in what you have to say. If you need cheap memory, look on pricewatch.com. > Thanks > > Bob Cochran > > Joshua Jensen wrote: > > >We have a few machines with 32 gigs of memory... I've not heard of any > >problems with them, no. > > > >Joshua > > > > > > > >On Fri, Mar 04, 2005 at 11:47:32AM -0800, Forrest Brewer wrote: > > > > > >>I am interested in the large memory possible in this processor design. > >>Does anyone have AMD-64 cards with more than 2GB memory running linux? > >>Any issues? > >>-- > >>Forrest Brewer > >> > >>-- > >>amd64-list mailing list > >>amd64-list at redhat.com > >>https://www.redhat.com/mailman/listinfo/amd64-list > >> > >> > > > > > > > > -- > amd64-list mailing list > amd64-list at redhat.com > https://www.redhat.com/mailman/listinfo/amd64-list -- Eugen* Leitl leitl ______________________________________________________________ ICBM: 48.07078, 11.61144 http://www.leitl.org 8B29F6BE: 099D 78BA 2FD3 B014 B08A 7779 75B0 2443 8B29 F6BE http://moleculardevices.org http://nanomachines.net -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 198 bytes Desc: not available URL: From eugen at leitl.org Sat Mar 5 10:41:48 2005 From: eugen at leitl.org (Eugen Leitl) Date: Sat, 5 Mar 2005 11:41:48 +0100 Subject: AMD-64 large memory... In-Reply-To: <42294439.9010605@harddata.com> References: <1109965652.8262.2.camel@canopus.ece.ucsb.edu> <42294439.9010605@harddata.com> Message-ID: <20050305104148.GM13336@leitl.org> On Fri, Mar 04, 2005 at 10:31:37PM -0700, Maurice Hilarius wrote: > Forrest Brewer wrote: > > >I am interested in the large memory possible in this processor design. > >Does anyone have AMD-64 cards with more than 2GB memory running linux? > >Any issues? > > > > > For example 32GB? > No problems. > Used FC, Tyan S4882 board (AMD chipset) Careful with Tyan: http://www.beowulf.org/archive/2004-October/010956.html and other threads. > Something more modest? > > 16GB on S2892 AMD chipset) > 16GB on S2891 (nVidia nForce4 chipset) > > 4GB on Tyan S2850, S2875, ASUS and MSI socket 939 board with VIA and > nForce chipsets. > > No issues. -- Eugen* Leitl leitl ______________________________________________________________ ICBM: 48.07078, 11.61144 http://www.leitl.org 8B29F6BE: 099D 78BA 2FD3 B014 B08A 7779 75B0 2443 8B29 F6BE http://moleculardevices.org http://nanomachines.net -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 198 bytes Desc: not available URL: From justin.conover at gmail.com Sat Mar 5 14:34:06 2005 From: justin.conover at gmail.com (Justin Conover) Date: Sat, 5 Mar 2005 08:34:06 -0600 Subject: AMD-64 large memory... In-Reply-To: <20050305104148.GM13336@leitl.org> References: <1109965652.8262.2.camel@canopus.ece.ucsb.edu> <42294439.9010605@harddata.com> <20050305104148.GM13336@leitl.org> Message-ID: http://www.linuxjournal.com/article/7563 With Fedora Core release 1.92 (FC2 Test 3) installed, the Ultimate Linux Box put up good numbers on the benchmarks, as might be expected. Yes, with this much RAM we took the opportunity to build a kernel in a tmpfs partition. 2.6.4 with all defaults set completed in 1 minute 41 seconds. More detailed benchmark results follow. Motherboard/chassis: Celestica A8440 (AMD-8131 Chipset) Memory: 16x PC2700 2048MB ECC REG (32GB) RAID: Adaptec ASR2200S Storage: Seagate ST336607LC 36GB U320 SCSI HDD x 4 You can read more from the link.. Here are some "workstations" you can look into also http://www.pogolinux.com/systems/workstations/index.html On Sat, 5 Mar 2005 11:41:48 +0100, Eugen Leitl wrote: > On Fri, Mar 04, 2005 at 10:31:37PM -0700, Maurice Hilarius wrote: > > Forrest Brewer wrote: > > > > >I am interested in the large memory possible in this processor design. > > >Does anyone have AMD-64 cards with more than 2GB memory running linux? > > >Any issues? > > > > > > > > For example 32GB? > > No problems. > > Used FC, Tyan S4882 board (AMD chipset) > > Careful with Tyan: http://www.beowulf.org/archive/2004-October/010956.html > and other threads. > > > Something more modest? > > > > 16GB on S2892 AMD chipset) > > 16GB on S2891 (nVidia nForce4 chipset) > > > > 4GB on Tyan S2850, S2875, ASUS and MSI socket 939 board with VIA and > > nForce chipsets. > > > > No issues. > > -- > Eugen* Leitl leitl > ______________________________________________________________ > ICBM: 48.07078, 11.61144 http://www.leitl.org > 8B29F6BE: 099D 78BA 2FD3 B014 B08A 7779 75B0 2443 8B29 F6BE > http://moleculardevices.org http://nanomachines.net > > > -- > amd64-list mailing list > amd64-list at redhat.com > https://www.redhat.com/mailman/listinfo/amd64-list > > > From maurice at harddata.com Sat Mar 5 16:46:33 2005 From: maurice at harddata.com (Maurice Hilarius) Date: Sat, 05 Mar 2005 09:46:33 -0700 Subject: AMD-64 large memory... In-Reply-To: <20050305104148.GM13336@leitl.org> References: <1109965652.8262.2.camel@canopus.ece.ucsb.edu> <42294439.9010605@harddata.com> <20050305104148.GM13336@leitl.org> Message-ID: <4229E269.9060609@harddata.com> Eugen Leitl wrote: >>For example 32GB? >>No problems. >>Used FC, Tyan S4882 board (AMD chipset) >> >> > >Careful with Tyan: http://www.beowulf.org/archive/2004-October/010956.html >and other threads. > > > Absolutely. Highlights why it is a perilous path to go the DIY approach. The facts stated are quite correct. DOA rates on Tyan motherboards are indeed 10% to 15% As a result we routinely bring in an overage, and all boards get tested within 48 hours of receipt. 24 hour burnin HAS to be done before ANY board can be considered a candidate to build a system. Before we think that makes Tyan POS, we see similar DOA rates on Intel, Supermicro, etc.. And regarding the earlier comment about sourcing RAM via "Pricewatch" the same comments apply in spades. If you buy RAM by that route I can nearly guarantee you will have grief. With our best regards, Maurice W. Hilarius Telephone: 01-780-456-9771 Hard Data Ltd. FAX: 01-780-456-9772 11060 - 166 Avenue email:maurice at harddata.com Edmonton, AB, Canada http://www.harddata.com/ T5X 1Y3 This email, message, and content, should be considered confidential, and is the copyrighted property of Hard Data Ltd., unless stated otherwise. -------------- next part -------------- An HTML attachment was scrubbed... URL: From cochranb at speakeasy.net Sun Mar 6 00:48:50 2005 From: cochranb at speakeasy.net (Robert L Cochran) Date: Sat, 05 Mar 2005 19:48:50 -0500 Subject: AMD-64 large memory... In-Reply-To: <32999.206.80.72.54.1109981476.squirrel@www.peaknet.net> References: <32999.206.80.72.54.1109981476.squirrel@www.peaknet.net> Message-ID: <422A5372.3090606@speakeasy.net> dsavage at peaknet.net wrote: >On Fri, 2005-03-04 at 18:31 -0500, Robert L Cochran wrote: > > >>I realize this is going to sound stupid, but what motherboard, exactly, >>can physically take 32 Gb of memory? What type of memory and where do >>you get it at a reasonable price? It must cost a fortune for that much. >>My wife has some applications that need memory beyond 4 Gb so I'm >>interested in what you have to say. >> >> > >Bob, > >AFAIK, with ten memory slots the Tyan quad-Opteron board >(http://www.tyan.com/products/html/thunderk8qspro.html) is the current >king of the hill. You'd need 4G memory parts ($$$!!!) to get to 32G on >this or any of the many boards with eight memory slots. > >But that's all theoretical because when maxed out this mobo is a true >yacht. You know...if you have to ask how much it costs, you can't afford >it. > >-- Doc >Robert G. (Doc) Savage, BSE(EE), CISSP, RHCE | Fairview Heights, IL >RHEL4/ES on Tyan S2468UGN w/3G, dual Athlon MP 2800+, 1.1T RAID5 >"Perfection is the enemy of good enough." > -- Admiral of the Fleet Sergei G. Gorshkov > > > > It's quite true I can't afford one of these babies. But it still leads me to another question. Where could I go to be shown how to build, configure, and test these systems? Or are they pretty much the same as consumer systems -- I've built 3 single processor systems for home use. Would these be the same, except one installs more CPUs, heat sinks, and memory. Or does it take specialized knowledge? Thanks Bob From c00jsh00 at nchc.org.tw Sun Mar 6 02:11:23 2005 From: c00jsh00 at nchc.org.tw (Jyh-Shyong Ho) Date: Sun, 06 Mar 2005 10:11:23 +0800 Subject: AMD-64 large memory... In-Reply-To: <422A5372.3090606@speakeasy.net> References: <32999.206.80.72.54.1109981476.squirrel@www.peaknet.net> <422A5372.3090606@speakeasy.net> Message-ID: <422A66CA.8090405@nchc.org.tw> Robert L Cochran wrote: > dsavage at peaknet.net wrote: > >> On Fri, 2005-03-04 at 18:31 -0500, Robert L Cochran wrote: >> >> >>> I realize this is going to sound stupid, but what motherboard, exactly, >>> can physically take 32 Gb of memory? What type of memory and where do >>> you get it at a reasonable price? It must cost a fortune for that much. >>> My wife has some applications that need memory beyond 4 Gb so I'm >>> interested in what you have to say. >>> >> >> >> Bob, >> >> AFAIK, with ten memory slots the Tyan quad-Opteron board >> (http://www.tyan.com/products/html/thunderk8qspro.html) is the current >> king of the hill. You'd need 4G memory parts ($$$!!!) to get to 32G on >> this or any of the many boards with eight memory slots. >> >> But that's all theoretical because when maxed out this mobo is a true >> yacht. You know...if you have to ask how much it costs, you can't afford >> it. >> >> -- Doc >> Robert G. (Doc) Savage, BSE(EE), CISSP, RHCE | Fairview Heights, IL >> RHEL4/ES on Tyan S2468UGN w/3G, dual Athlon MP 2800+, 1.1T RAID5 >> "Perfection is the enemy of good enough." >> -- Admiral of the Fleet Sergei G. Gorshkov >> >> >> >> > It's quite true I can't afford one of these babies. But it still leads > me to another question. Where could I go to be shown how to build, > configure, and test these systems? Or are they pretty much the same as > consumer systems -- I've built 3 single processor systems for home > use. Would these be the same, except one installs more CPUs, heat > sinks, and memory. Or does it take specialized knowledge? > > Thanks > > Bob > > Hi, We built a 4-way Opteron server with TYAN mother board and 32GB RAM last September, we used SuSE 9.1 for AMD64, it costed about 30,000 US$. The procedure is the same as for installing sngle and dual Opteron system, no special knowledge is needed (at least in this case, or perhaps we were just lucky). When all components are properly installed, SuSE installer picked up the right kernel and drivers for this 4-way system, the system has been running continuously since then.. Jyh-Shyong Ho, Ph.D. Research Scientist National Center for High Performance Computing Hsinchu, Taiwan, ROC -------------- next part -------------- An HTML attachment was scrubbed... URL: From b.j.smith at ieee.org Sat Mar 5 21:23:39 2005 From: b.j.smith at ieee.org (Bryan J. Smith) Date: Sat, 05 Mar 2005 16:23:39 -0500 Subject: AMD-64 large memory... In-Reply-To: <4229E269.9060609@harddata.com> References: <1109965652.8262.2.camel@canopus.ece.ucsb.edu> <42294439.9010605@harddata.com> <20050305104148.GM13336@leitl.org> <4229E269.9060609@harddata.com> Message-ID: <1110057820.5626.10.camel@localhost.localdomain> On Sat, 2005-03-05 at 09:46 -0700, Maurice Hilarius wrote: > Absolutely. > Highlights why it is a perilous path to go the DIY approach. > The facts stated are quite correct. > DOA rates on Tyan motherboards are indeed 10% to 15% > As a result we routinely bring in an overage, and all boards get > tested within 48 hours of receipt. > 24 hour burnin HAS to be done before ANY board can be considered a > candidate to build a system. > Before we think that makes Tyan POS, we see similar DOA rates on > Intel, Supermicro, etc.. > And regarding the earlier comment about sourcing RAM via "Pricewatch" > the same comments apply in spades. > If you buy RAM by that route I can nearly guarantee you will have > grief. 10% of mainboards will fail within 30 days, even after passing a 24-48 hour burn-in. If they last past 30 days, then they will last for years. -- Bryan J. Smith mailto:b.j.smith at ieee.org ---------------------------------------------------------- Beware of those who define their preferences in terms of hate of another option, than on the merits of their choice From b.j.smith at ieee.org Sat Mar 5 19:33:21 2005 From: b.j.smith at ieee.org (Bryan J. Smith) Date: Sat, 05 Mar 2005 14:33:21 -0500 Subject: AMD-64 large memory... In-Reply-To: <42294439.9010605@harddata.com> References: <1109965652.8262.2.camel@canopus.ece.ucsb.edu> <42294439.9010605@harddata.com> Message-ID: <1110051202.5626.8.camel@localhost.localdomain> On Fri, 2005-03-04 at 22:31 -0700, Maurice Hilarius wrote: > Forrest Brewer wrote: > For example 32GB? > No problems. > Used FC, Tyan S4882 board (AMD chipset) > Something more modest? > 16GB on S2892 AMD chipset) > 16GB on S2891 (nVidia nForce4 chipset) > 4GB on Tyan S2850, S2875, ASUS and MSI socket 939 board with VIA and > nForce chipsets. > No issues. If you go beyond 4GB, you want to be running on an EV6 platform (Athlon [32] with 40-bit support in the BIOS, or any Athlon64/Opteron), and _not_ any Intel GTL platform (not even AGTL+ LGA-775). There is some major interconnect limitations with GTL. -- Bryan J. Smith mailto:b.j.smith at ieee.org ---------------------------------------------------------- Beware of those who define their preferences in terms of hate of another option, than on the merits of their choice From b.j.smith at ieee.org Sun Mar 6 04:04:07 2005 From: b.j.smith at ieee.org (Bryan J. Smith) Date: Sat, 05 Mar 2005 23:04:07 -0500 Subject: [OT] Tyan S2895 K8WE and EPV12V/SSI 3.51 power supply Message-ID: <1110081847.5445.24.camel@localhost.localdomain> I'm looking at going with one of these for a workstation when they become available mid-month. My main reason is that I use 64-bit*66MHz PCI 3Ware Escalade storage controllers (and am playing around with the 64-bit*66MHz PCI NetCell controller as well). Putting these suckers on a 32-bit*33MHz bus in the nForce4 mainboards is not my idea of a good time, especially as the audio will compete for the same PCI bus (at least the chipset NIC is segmented on a PCIe x1 channel). Especially since I already have 64- bit*66MHz PCI slots in even my older Athlon MP mainboard (even if the 32*33MHz slots are bridged from it). The _lack_ of _quality_ PCIe x1/x4/x8 storage controllers is really killing me (the only one I've seen is the IOP332 XScale-based LSI SerialATA 8-channel PCIe x8 (bridged to PCI-X in the IOP332) and it is over $500 -- and still isn't quite as high performing as the 3Ware Escalade 9000 series at RAID-5, or even the cheaper 8000 series at RAID-0+1). I was comparing the specifications of a $200 Athlon64 3500+ 2GHz/512KB 90nm and the $300 Opteron 246 2GHz/1MB 130nm ($500 if I want to go 90nm/55W) and the Opteron edges out the newer Athlon64 in performance (except for maybe a few things). So I figured I could go with the S2895 mainboard now, which also gives me the option to upgrade with a 2nd CPU later on (or maybe dual-core * 2, or maybe just a single-core when the dual-cores come out and the single-cores drop in price). The manual says single processor is okay, as long as you understand the nVidia 2050Pro chip won't be usable (which removes the use of the 2nd PCIe x16 slot and 2nd 10/100/1000 NIC). [ Side Note: I also already have a bunch (4 DIMMs) of CL2.5 PC2100 Registered ECC SDRAM which should work fine (it already works with my Athlon MP, which is basically the same EV6 memory interconnect, only segmented in the NUMA/EV6 Opteron), saving some cost there. I'll worry about upgrading that when I add a 2nd CPU later. ] The only kicker is the new EPS12V/SSI 3.51 power supply requirement. >From the looks of it, any EPS12V should do for a single CPU/video card. But I'm considering buying a solid power supply that will also be usable when I upgrade. That means a _true_ 24-pin + 8-pin + 6-pin "workstation" connector (_not_ a 6-pin AUX connector). I have _not_ seen a _single_ power supply claiming SSI 3.51. There are many that have both 24-pin and either 8-pin or 6-pin, but not both. The closest I've come to verifying support is the older S2885 recommendations here: http://tyan.com/l_german/support/html/r_s2885.html It also looks like the $190-250 Enermax EPS12V 550/660W power supplies _might_ have all 3 as well (they call it a EPS12V 2.0). Especially given the fact that they have *4* +12V lines (18/18A, 18/16A, 14/14A, 14/14A -- for the 660/550, respectively): http://www.enermax.com.tw/products_page.php? Tid=1&gon=236&Gid=26&Gid2=35 They also have a new $150-175 "All-in-one" 600W series, although I couldn't verify it has a true 24-pin + 8-pin + 6-pin connector (kinda looked like it did have the 6-pin, but maybe not the 24-pin + 8-pin completely in addition, just an option?): http://www.enermax.com.tw/products_page.php?Tid=1&gon=261&Gid=18&Gid2=45 If _anyone_ has any info on what power supplies will do, let me know. It seems there are plenty of 24+8 or 24+8/6WS and even 24+8+6AUX but _not_ the _full_ 24+8+6WS. The Tyan S2895 isn't on the shelf yet (expected mid-March), so I guess the power supplies will be more "well-known" then? But then again, I don't think the issue is any different than for the S2885 either, and true 24+8+6 seem to be hard to find. It might have not been such a big deal with the S2885's single AGP Pro slot, but now we're talking two (2) PCIe x16 slots with a lot of power requirements. Not that I have to worry, I'm only going a single Opteron 246 and GeForce 6800GT to start. But I'd still like to buy one power supply now, and not have to upgrade in 6-9 months when I had the 2nd CPU and 2nd 6800GT. Which also makes me wonder, if I have the full 24+8+6WS power, which should deliver adequate power to the card itself (the PS has 4 +12V inputs!), do I need to use the 4-pin Molex on the 6800GT video card? Curious minds want to know. -- Bryan J. Smith mailto:b.j.smith at ieee.org ---------------------------------------------------------- Beware of those who define their preferences in terms of hate of another option, than on the merits of their choice From b.j.smith at ieee.org Sun Mar 6 04:13:29 2005 From: b.j.smith at ieee.org (Bryan J. Smith) Date: Sat, 05 Mar 2005 23:13:29 -0500 Subject: [OT] Tyan S2895 K8WE and EPV12V/SSI 3.51 power supply (correction/additions) In-Reply-To: <1110081847.5445.24.camel@localhost.localdomain> References: <1110081847.5445.24.camel@localhost.localdomain> Message-ID: <1110082409.5445.30.camel@localhost.localdomain> On Sat, 2005-03-05 at 23:04 -0500, Bryan J. Smith wrote: > I was comparing the specifications of a $200 Athlon64 3500+ 2GHz/512KB > 90nm and the $300 Opteron 246 2GHz/1MB 130nm ($500 if I want to go > 90nm/55W) and the Opteron edges out the newer Athlon64 in performance > (except for maybe a few things). Ack, let me correct something. I mean the $200 Athlon64 _3200+_. The 3500+ is $300 and 2.2GHz/512KB. In any case, the Opteron 246 2GHz/1MB performs very well against. So it's really just a matter of additional 2x (+$300) in mainboard cost (I'm playing for the 2x in PCIe channels, plus PCI-X channels which I really want) over an SLI mainboard. I like these options. I already have an Li-Lian PC-1200V that cools very, very well, while being very quiet (at least considering it has 2 x 120mm fans), and keeps the 5 hard drives cool. The only "extra negative" of the S2895 v. most nForce4 mainboards is that Tyan choose a _cheap_ AD1981B 2-channel audio, whereas the latter typically comes with a 6-channel ALC850/860 which _does_ give you 6- channel with the i8x0 ALSA driver. So I've gotta waste the 32-bit PCI slot for a SB Audigy2 (or similar). -- Bryan J. Smith mailto:b.j.smith at ieee.org ---------------------------------------------------------- Beware of those who define their preferences in terms of hate of another option, than on the merits of their choice From dsavage at peaknet.net Sun Mar 6 04:17:11 2005 From: dsavage at peaknet.net (dsavage at peaknet.net) Date: Sat, 5 Mar 2005 22:17:11 -0600 (CST) Subject: AMD-64 large memory... Message-ID: <34757.206.80.72.54.1110082631.squirrel@www.peaknet.net> On Sun, 2005-03-06 at 10:11 +0800, Jyh-Shyong Ho wrote: > Robert L Cochran wrote: > > It's quite true I can't afford one of these babies. But it still > > leads me to another question. Where could I go to be shown how to > > build, configure, and test these systems? Or are they pretty much > > the same as consumer systems -- I've built 3 single processor > > systems for home use. Would these be the same, except one installs > > more CPUs, heat sinks, and memory. Or does it take specialized > > knowledge? > > > > Thanks > > > > Bob > > Hi, > > We built a 4-way Opteron server with TYAN mother board and 32GB RAM > last September, we used SuSE 9.1 for AMD64, it costed about 30,000 > US$. The procedure is the same as for installing sngle and dual > Opteron system, no special knowledge is needed (at least in this > case, or perhaps we were just lucky). When all components are > properly installed, SuSE installer picked up the right kernel and > drivers for this 4-way system, the system has been > running continuously since then.. Dr. Ho has probably trumped all of us here by actually having done this. In theory it shouldn't be all that different from building any other high end system, except it would cost up to ten times as much. I would expect the biggest challenges would be hardware: (1) finding a suitable power supply (supplies?), (2) finding an enclosure that could channel enough air across the board to keep it cool, and (3) finding the 4G memory parts. I'll be waiting a few more months until AMD releases dual-core Opterons before I try upgrading to a more affordable quad system with two sockets. -- Doc Robert G. (Doc) Savage, BSE(EE), CISSP, RHCE | Fairview Heights, IL RHEL4/ES on Tyan S2468UGN w/3G, dual Athlon MP 2800+, 1.1T RAID5 "Perfection is the enemy of good enough." -- Admiral of the Fleet Sergei G. Gorshkov From b.j.smith at ieee.org Sun Mar 6 04:52:41 2005 From: b.j.smith at ieee.org (Bryan J. Smith) Date: Sat, 05 Mar 2005 23:52:41 -0500 Subject: [OT] Tyan S2895 K8WE and EPV12V/SSI 3.51 power supply In-Reply-To: <422A8797.4080403@speakeasy.net> References: <1110081847.5445.24.camel@localhost.localdomain> <422A8797.4080403@speakeasy.net> Message-ID: <1110084761.5445.64.camel@localhost.localdomain> On Sat, 2005-03-05 at 23:31 -0500, Robert L Cochran wrote: > Would this baby work for you? I don't know when it will be available. If > I need a power supply that has to work right and not fail, then I shell > out the dough for PC Power & Cooling supplies. > http://www.pcpowercooling.com/about/index_whatnew.htm Which product? Most of those are ATX, not EPS12V/SSI. I was already looking at the 510AG (note the "AG") here: http://www.pcpowercooling.com/products/power_supplies/maxperformance/turbocools/index_hp_atx-2.htm Now that had dual-12V, and was $199, comparable in price to the quad-12V of the Enermax 660/550W units, but not features. But the last one on that page is a quad-12V 850W (is that was you meant? you passed the original frame, not the 850W link ;-): http://www.pcpowercooling.com/about/whatnew_850_ETX.htm I can only imagine this thing is $350+ -- especially seeing *2* 6-pin WS connectors (wow!). The Tyan S2895 only has a connector for 1. > I think Antec is second best for consumer machines and will use an > Antec supply if I need decent quality at an acceptable price point. It varies. I have an Antec TruePower ATX 1.x 550W in my Athlon MP system. And I like it, especially since the S2460/2466 had such issues due to the fact that way too much current was coming from the stock 20- pin ATX connector. But Enermax has seemingly had more options, especially on EPS12V. Antec doesn't offer an EPS12V with the 24+8+6WS option -- and, again, the Tyan page notes that specifically: http://tyan.com/l_german/support/html/r_s2885.html I even looked it up: http://www.antec.com/specs/true550EPS12V_spe.html Yet another 6-pin "AUX", not a 6-pin "Workstation" connector. -- Bryan J. Smith mailto:b.j.smith at ieee.org ---------------------------------------------------------- Beware of those who define their preferences in terms of hate of another option, than on the merits of their choice From b.j.smith at ieee.org Sun Mar 6 05:02:38 2005 From: b.j.smith at ieee.org (Bryan J. Smith) Date: Sun, 06 Mar 2005 00:02:38 -0500 Subject: AMD-64 large memory... -- 4-way v. 2-way dual-core ... In-Reply-To: <34757.206.80.72.54.1110082631.squirrel@www.peaknet.net> References: <34757.206.80.72.54.1110082631.squirrel@www.peaknet.net> Message-ID: <1110085358.5445.75.camel@localhost.localdomain> On Sat, 2005-03-05 at 22:17 -0600, dsavage at peaknet.net wrote: > I'll be waiting a few more months until AMD releases dual-core Opterons > before I try upgrading to a more affordable quad system with two sockets. I don't think you'll be saving any money or price/performance. Why? 1) You'll only have (4) DDR channels in (2) Socket-940 versus (8) DDR channels in (4) Socket-940 2) The dual-core will be "EE" (55W low-power) logic (110W total) and command a hefty premium, and aren't nearly the fastest. E.g., the current 240/840EE (1.4GHz) and 246/846EE (2.0GHz) versions are currently 70%+ extra, and still only the lower or mid-speed compared to the availability of the 252/852 (2.6GHz) at about the same price. It wouldn't surprise me if (4) 8xx series CPUs of the same speed are the same prices as (2) 2xx series CPUs with dual-core at the same speed -- including after the mainboard costs. ;-> Then factor in the reality that most 4-way systems (like the AMD reference design) have a 2nd AMD8131/8132 chip for I/O than the 2-way systems and 4-way is typically better. I could be wrong, especially if the dual-cores sell so the "economies of scale" can take over. But I think that will _only_ happen for the Socket-939 versions, and not the Socket-940 versions. They will command a premium. So you're losing DDR channels while paying a lot more for what you actually get at 4-way. At least for a server role when you need the I/O -- especially for lots of threaded servers with lots of simultaneous I/O. Now if you just want CPU power, that's different. Although if it's just CPU power, then you're better off buying a 1 or 2 CPU system with an HTX slot and setting up an Infiniband network for less cost. ;-> Heck, I think the first mainboard manufacturer to introduce a Socket-939 mainboard with an HTX slot won't be able to meet orders -- especially after the dual-core Socket-939s come out. -- Bryan J. Smith mailto:b.j.smith at ieee.org ---------------------------------------------------------- Beware of those who define their preferences in terms of hate of another option, than on the merits of their choice From c00jsh00 at nchc.org.tw Sun Mar 6 06:10:34 2005 From: c00jsh00 at nchc.org.tw (Jyh-Shyong Ho) Date: Sun, 06 Mar 2005 14:10:34 +0800 Subject: AMD-64 large memory... In-Reply-To: <34757.206.80.72.54.1110082631.squirrel@www.peaknet.net> References: <34757.206.80.72.54.1110082631.squirrel@www.peaknet.net> Message-ID: <422A9EDA.4020400@nchc.org.tw> dsavage at peaknet.net wrote: >On Sun, 2005-03-06 at 10:11 +0800, Jyh-Shyong Ho wrote: > > >>Robert L Cochran wrote: >> >> >>>It's quite true I can't afford one of these babies. But it still >>>leads me to another question. Where could I go to be shown how to >>>build, configure, and test these systems? Or are they pretty much >>>the same as consumer systems -- I've built 3 single processor >>>systems for home use. Would these be the same, except one installs >>>more CPUs, heat sinks, and memory. Or does it take specialized >>>knowledge? >>> >>>Thanks >>> >>>Bob >>> >>> >>Hi, >> >>We built a 4-way Opteron server with TYAN mother board and 32GB RAM >>last September, we used SuSE 9.1 for AMD64, it costed about 30,000 >>US$. The procedure is the same as for installing sngle and dual >>Opteron system, no special knowledge is needed (at least in this >>case, or perhaps we were just lucky). When all components are >>properly installed, SuSE installer picked up the right kernel and >>drivers for this 4-way system, the system has been >>running continuously since then.. >> >> > >Dr. Ho has probably trumped all of us here by actually having done this. >In theory it shouldn't be all that different from building any other high >end system, except it would cost up to ten times as much. I would expect >the biggest challenges would be hardware: (1) finding a suitable power >supply (supplies?), (2) finding an enclosure that could channel enough air >across the board to keep it cool, and (3) finding the 4G memory parts. > >I'll be waiting a few more months until AMD releases dual-core Opterons >before I try upgrading to a more affordable quad system with two sockets. > >-- Doc >Robert G. (Doc) Savage, BSE(EE), CISSP, RHCE | Fairview Heights, IL >RHEL4/ES on Tyan S2468UGN w/3G, dual Athlon MP 2800+, 1.1T RAID5 >"Perfection is the enemy of good enough." > -- Admiral of the Fleet Sergei G. Gorshkov > > > Hi, Next week, a local SI will provide me a price quote for a 8-way Opteron server based on IWill's motherboard, IWill's 8-way Opteron server consists two connected 4-way Opteron boards in a 5U chassis, there are 32 DIMMs available on the server. If one can find 4GB RAM, the server can accomodate 128GB RAM! If we can get the budget this year, we will see what the performance of this 8-way server would be. As a comparison, we have a HP Superdome with 64 IA64 CPUs and 128GB RAM in a 42U rack, which costs about 1,300,000 US$ two years ago. I estimate that the 5U 8-way Opteron server with 128GB RAM would cost about 70,000US$ today. If your applications requires very large memory, IWill's H8501 would be a very good choice (http://www.iwill.com.tw/product_2.asp?p_id=90&sp=Y). 8-way Opteron server should be able to realize the full performance potential of Opteron's design. Best regards Jyh-Shyong Ho, Ph.D. Research Scientist National Center for High Performance Computing Hsinchu, Taiwan, ROC -------------- next part -------------- An HTML attachment was scrubbed... URL: From eugen at leitl.org Sun Mar 6 11:01:12 2005 From: eugen at leitl.org (Eugen Leitl) Date: Sun, 6 Mar 2005 12:01:12 +0100 Subject: AMD-64 large memory... In-Reply-To: <422A5372.3090606@speakeasy.net> References: <32999.206.80.72.54.1109981476.squirrel@www.peaknet.net> <422A5372.3090606@speakeasy.net> Message-ID: <20050306110112.GW13336@leitl.org> On Sat, Mar 05, 2005 at 07:48:50PM -0500, Robert L Cochran wrote: > It's quite true I can't afford one of these babies. But it still leads If you need some >4 GByte memory this either means large modules, or ECC/registered, which is not cheap. With 16 GByte memory costs dominate in comparison to a vanilla dual-Opteron. > me to another question. Where could I go to be shown how to build, > configure, and test these systems? Or are they pretty much the same as > consumer systems -- I've built 3 single processor systems for home use. Is this a rackmount, or a standalone server? How many rack units high? Do you need remote management? Is this a mission-critical system? It all depends on answers to these questions. > Would these be the same, except one installs more CPUs, heat sinks, and > memory. Or does it take specialized knowledge? Cooling 1U systems is nontrivial. I doubt there are dual-Opteron kits for 1U. You should have no issues with a 4U case, or a standalone server. -- Eugen* Leitl leitl ______________________________________________________________ ICBM: 48.07078, 11.61144 http://www.leitl.org 8B29F6BE: 099D 78BA 2FD3 B014 B08A 7779 75B0 2443 8B29 F6BE http://moleculardevices.org http://nanomachines.net -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 198 bytes Desc: not available URL: From b.j.smith at ieee.org Sun Mar 6 11:05:52 2005 From: b.j.smith at ieee.org (Bryan J. Smith) Date: Sun, 06 Mar 2005 06:05:52 -0500 Subject: AMD-64 large memory... In-Reply-To: <20050306110112.GW13336@leitl.org> References: <32999.206.80.72.54.1109981476.squirrel@www.peaknet.net> <422A5372.3090606@speakeasy.net> <20050306110112.GW13336@leitl.org> Message-ID: <1110107152.5445.223.camel@localhost.localdomain> On Sun, 2005-03-06 at 12:01 +0100, Eugen Leitl wrote: > Cooling 1U systems is nontrivial. I doubt there are dual-Opteron kits for 1U. There are. Even the original Opteron 2xx's that could run over 100W. But now there are the 30W Opteron 240EE (1.4GHz) and 55W Opteron 246EE (2.0GHz) versions. Definitely fine for dual-1U. > You should have no issues with a 4U case, or a standalone > server. Plenty of 1U and 2U 2-way, and even some 3U 4-ways (although typically with a 4-way you're more interested in I/O expansion). -- Bryan J. Smith mailto:b.j.smith at ieee.org ---------------------------------------------------------- Beware of those who define their preferences in terms of hate of another option, than on the merits of their choice From cochranb at speakeasy.net Sun Mar 6 23:39:08 2005 From: cochranb at speakeasy.net (Robert L Cochran) Date: Sun, 06 Mar 2005 18:39:08 -0500 Subject: [OT] Tyan S2895 K8WE and EPV12V/SSI 3.51 power supply In-Reply-To: <1110084761.5445.64.camel@localhost.localdomain> References: <1110081847.5445.24.camel@localhost.localdomain> <422A8797.4080403@speakeasy.net> <1110084761.5445.64.camel@localhost.localdomain> Message-ID: <422B949C.9000103@speakeasy.net> Bryan J. Smith wrote: > <>On Sat, 2005-03-05 at 23:31 -0500, Robert L Cochran wrote: > Which product? Most of those are ATX, not EPS12V/SSI. > >But the last one on that page is a quad-12V 850W (is that was you meant? >you passed the original frame, not the 850W link ;-): >http://www.pcpowercooling.com/about/whatnew_850_ETX.htm > >I can only imagine this thing is $350+ -- especially seeing *2* 6-pin WS >connectors (wow!). The Tyan S2895 only has a connector for 1. > > > Well, I wasn't sure if the 850W supply would fit your needs, but since it is a brand new offering from PC Power & Cooling, I thought I'd pass it on. Yes it is expensive -- I've read in Maximum PC magazine that it will run about USD $430. But if it meets your needs and can do big mean motherboards then it is worth it, hmnm? Some day I'll build a multiprocessor machine. If the various parts are available on the market and one only needs to be handy with a phillips screwdriver plus take care with antistatic grounding, then I'll give it a whirl. It's true I need to get somewhat richer than I am now, but we all need goals.... Bob From b.j.smith at ieee.org Mon Mar 7 01:13:08 2005 From: b.j.smith at ieee.org (Bryan J. Smith) Date: Sun, 06 Mar 2005 20:13:08 -0500 Subject: [OT] Tyan S2895 K8WE and EPV12V/SSI 3.51 power supply In-Reply-To: <422B949C.9000103@speakeasy.net> References: <1110081847.5445.24.camel@localhost.localdomain> <422A8797.4080403@speakeasy.net> <1110084761.5445.64.camel@localhost.localdomain> <422B949C.9000103@speakeasy.net> Message-ID: <1110157988.5445.325.camel@localhost.localdomain> On Sun, 2005-03-06 at 18:39 -0500, Robert L Cochran wrote: > Well, I wasn't sure if the 850W supply would fit your needs, but since > it is a brand new offering from PC Power & Cooling, I thought I'd pass > it on. Yes it is expensive -- I've read in Maximum PC magazine that it > will run about USD $430. But if it meets your needs and can do big mean > motherboards then it is worth it, hmnm? It's overkill for the S2895. But thanx anyway. > Some day I'll build a multiprocessor machine. If the various parts are > available on the market and one only needs to be handy with a phillips > screwdriver plus take care with antistatic grounding, then I'll give it > a whirl. It's true I need to get somewhat richer than I am now, but we > all need goals.... I am _not_ building it because it's multi-CPU. Multi-CPU matters little. I'm building it because it has about 3x the I/O of even a new generation PCIe system, let alone 10x the I/O of yester-year's PC generation. So many people spend a _lot_ of money on CPU and other details, then put all of that I/O on a _single_, _shared_ 32-bit at 33MHz PCI bus. One that is wholely _inadequate_ for today's disk+NIC+audio. -- Bryan J. Smith mailto:b.j.smith at ieee.org ---------------------------------------------------------- Beware of those who define their preferences in terms of hate of another option, than on the merits of their choice From agshew at gmail.com Mon Mar 7 01:25:11 2005 From: agshew at gmail.com (Andrew Shewmaker) Date: Sun, 6 Mar 2005 18:25:11 -0700 Subject: AMD-64 large memory... -- 4-way v. 2-way dual-core ... In-Reply-To: <1110085358.5445.75.camel@localhost.localdomain> References: <34757.206.80.72.54.1110082631.squirrel@www.peaknet.net> <1110085358.5445.75.camel@localhost.localdomain> Message-ID: On Sun, 06 Mar 2005 00:02:38 -0500, Bryan J. Smith wrote: > 2) The dual-core will be "EE" (55W low-power) logic (110W total) and > command a hefty premium, and aren't nearly the fastest. E.g., the > current 240/840EE (1.4GHz) and 246/846EE (2.0GHz) versions are currently > 70%+ extra, and still only the lower or mid-speed compared to the > availability of the 252/852 (2.6GHz) at about the same price. It > wouldn't surprise me if (4) 8xx series CPUs of the same speed are the > same prices as (2) 2xx series CPUs with dual-core at the same speed -- > including after the mainboard costs. ;-> Up until a month ago I had heard the same (that the dual cores would be considerably slower), but recently there have been people claiming that AMD's dual cores will becoming out at 2.4 GHz. http://www.theinquirer.net/?article=21531 http://www.informationweek.com/story/showArticle.jhtml?articleID=60402975 http://www.varbusiness.com/sections/news/breakingnews.jhtml%3Bjsessionid=URCWE4KKBJDGSQSNDBCCKH0CJUMEKJVN?articleId=60402926 -- Andrew Shewmaker From b.j.smith at ieee.org Mon Mar 7 01:30:44 2005 From: b.j.smith at ieee.org (Bryan J. Smith) Date: Sun, 06 Mar 2005 20:30:44 -0500 Subject: AMD-64 large memory... -- 4-way v. 2-way dual-core ... In-Reply-To: References: <34757.206.80.72.54.1110082631.squirrel@www.peaknet.net> <1110085358.5445.75.camel@localhost.localdomain> Message-ID: <1110159044.5445.328.camel@localhost.localdomain> On Sun, 2005-03-06 at 18:25 -0700, Andrew Shewmaker wrote: > Up until a month ago I had heard the same (that the dual cores would be > considerably slower), but recently there have been people claiming that > AMD's dual cores will becoming out at 2.4 GHz. Considering they are coming out Q3, after AMD has released it's planned 2.8 and 3.0GHz single core parts, 2.4GHz sounds normal for dual-core. In a nutshell, due to sheer transistor count, something has got to give -- cache, clock, etc... for thermal considerations, compared to single core. -- Bryan J. Smith mailto:b.j.smith at ieee.org ---------------------------------------------------------- Beware of those who define their preferences in terms of hate of another option, than on the merits of their choice From jay at scherrer.com Tue Mar 8 18:31:07 2005 From: jay at scherrer.com (Jay Scherrer) Date: Tue, 8 Mar 2005 10:31:07 -0800 Subject: New install vs upgrade? Message-ID: <200503081031.07948.jay@scherrer.com> I tried installing Fedora core 3 version x86_64 on a new Laptop: Asus Z8100 with a mobile AMD 64 memory is 1 gig. The install went ok but after logging in the system it seemed very slow. I hadn't yumed it yet untill I try the install for FC3 x386 The install of FC3 x386 went smooth and everything runs faster. When I config cpan it detects my system as a dual proccesor which should be normal for running a dual layer proccesor. Now I am thinking of yum updating FC3 x386 then ugrade to FC3 x86_64 because during the initial install there were several files that remained compiled for the x386. Has anyone else tried upgrading toFC3 x86_64 from FC3 x386? Jay Scherrer From lamont at gurulabs.com Tue Mar 8 18:51:40 2005 From: lamont at gurulabs.com (Lamont R. Peterson) Date: Tue, 08 Mar 2005 11:51:40 -0700 Subject: AMD-64 large memory... In-Reply-To: <4228EFC9.5090501@speakeasy.net> References: <1109965652.8262.2.camel@canopus.ece.ucsb.edu> <20050304232440.GA3903@iwsp.com> <4228EFC9.5090501@speakeasy.net> Message-ID: <1110307900.5418.12.camel@corsair.lrp.advansoft.us> On Fri, 2005-03-04 at 18:31 -0500, Robert L Cochran wrote: > I realize this is going to sound stupid, but what motherboard, exactly, > can physically take 32 Gb of memory? What type of memory and where do > you get it at a reasonable price? It must cost a fortune for that much. > My wife has some applications that need memory beyond 4 Gb so I'm > interested in what you have to say. Check out pricing on pricewatch.com (as suggested already on the list), but do got go too cheap. Buy stuff from a company with a real warranty. For RAM, the measuring stick of a "real" warranty is Crucial [http://www.crucial.com/], which is lifetime. Personally, for any system I care about or would ever spend that kind of money on, I *always* buy my RAM from Crucial. It just is not worth the stupid problems that come up with flaky RAM. -- Lamont R. Peterson Senior Instructor Guru Labs, L.C. http://www.GuruLabs.com/ -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 189 bytes Desc: This is a digitally signed message part URL: From joshua at iwsp.com Wed Mar 9 16:51:25 2005 From: joshua at iwsp.com (Joshua Jensen) Date: Wed, 9 Mar 2005 11:51:25 -0500 Subject: AMD-64 large memory... In-Reply-To: <1110307900.5418.12.camel@corsair.lrp.advansoft.us> References: <1109965652.8262.2.camel@canopus.ece.ucsb.edu> <20050304232440.GA3903@iwsp.com> <4228EFC9.5090501@speakeasy.net> <1110307900.5418.12.camel@corsair.lrp.advansoft.us> Message-ID: <20050309165124.GB2786@iwsp.com> Agreed. Of course many vendors won't support your machine if they aren't using memory that they have approved.... so while crucial is a good source, look to your box vendor for guidance. Joshua On Tue, Mar 08, 2005 at 11:51:40AM -0700, Lamont R. Peterson wrote: > On Fri, 2005-03-04 at 18:31 -0500, Robert L Cochran wrote: > > I realize this is going to sound stupid, but what motherboard, exactly, > > can physically take 32 Gb of memory? What type of memory and where do > > you get it at a reasonable price? It must cost a fortune for that much. > > My wife has some applications that need memory beyond 4 Gb so I'm > > interested in what you have to say. > > Check out pricing on pricewatch.com (as suggested already on the list), > but do got go too cheap. Buy stuff from a company with a real warranty. > For RAM, the measuring stick of a "real" warranty is Crucial > [http://www.crucial.com/], which is lifetime. > > Personally, for any system I care about or would ever spend that kind of > money on, I *always* buy my RAM from Crucial. It just is not worth the > stupid problems that come up with flaky RAM. > -- > Lamont R. Peterson > Senior Instructor > Guru Labs, L.C. http://www.GuruLabs.com/ > -- > amd64-list mailing list > amd64-list at redhat.com > https://www.redhat.com/mailman/listinfo/amd64-list -- Joshua Jensen joshua at iwsp.com "If God didn't want us to eat animals, why did he make them out of meat?" From jlabadie at acm.org Fri Mar 11 06:15:21 2005 From: jlabadie at acm.org (Jon LaBadie) Date: Fri, 11 Mar 2005 01:15:21 -0500 Subject: parted and raid 10 Message-ID: As supplied by dell, the system has a single / ext3 filesystem of 136 GB (RHEL 3, update 2, x86_64). It is the last fdisk logical partition, sda6, on a hardware raid 10 (4 drives). I've not used parted. Is it feasible/safe to use parted to shrink the fs and partition (about 40 GB of data in the 136GB space) then create several new partitions and fs's? From angelus at sangreal.demon.nl Fri Mar 11 21:50:30 2005 From: angelus at sangreal.demon.nl (Angelo Machils) Date: Fri, 11 Mar 2005 22:50:30 +0100 Subject: BIOS update ECS mobo Message-ID: <423212A6.8080701@sangreal.demon.nl> Hello there! Has anyone done a bios-update of a ECS Elitegroups (755-A2) motherboard under Linux? I need to update my bios to (perhaps) solve my SATA problem, but I can only find a win util to load the update and the box is running FC3... I have written to ECS, but no answer till now (probably the last ECS mobo I will ever buy) Thanks in advance, Angelo From jay at scherrer.com Wed Mar 16 16:56:25 2005 From: jay at scherrer.com (Jay Scherrer) Date: Wed, 16 Mar 2005 08:56:25 -0800 Subject: SiS162 wireless/modem Message-ID: <200503160856.25241.jay@scherrer.com> Has anyone had any luck configuring a SiS162 wireless/modem. I think it is an off shoot of the WinModem which has always had problems with Linux. Jay Scherrer From jay at scherrer.com Sun Mar 20 03:22:20 2005 From: jay at scherrer.com (Jay Scherrer) Date: Sat, 19 Mar 2005 19:22:20 -0800 Subject: gftp Message-ID: <1111288940.5523.6.camel@gimly> Having trouble with gftp on a Fedora Core 3 x86_64 system. I can connect between systems but once I try to copy files the gftp crashes. Jay Scherrer From rchiodin at bellsouth.net Sun Mar 20 13:52:11 2005 From: rchiodin at bellsouth.net (Bob Chiodini) Date: Sun, 20 Mar 2005 08:52:11 -0500 Subject: gftp In-Reply-To: <1111288940.5523.6.camel@gimly> References: <1111288940.5523.6.camel@gimly> Message-ID: <1111326731.9570.27.camel@littlenail.homelinux.net> On Sat, 2005-03-19 at 19:22 -0800, Jay Scherrer wrote: > Having trouble with gftp on a Fedora Core 3 x86_64 system. > I can connect between systems but once I try to copy files the gftp > crashes. > > Jay Scherrer > Jay, https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=140567 Bob... From jmorris at beau.org Tue Mar 22 18:12:53 2005 From: jmorris at beau.org (John Morris) Date: Tue, 22 Mar 2005 12:12:53 -0600 Subject: gftp In-Reply-To: <1111288940.5523.6.camel@gimly> References: <1111288940.5523.6.camel@gimly> Message-ID: <1111515173.2894.7.camel@mjolnir> On Sat, 2005-03-19 at 21:22, Jay Scherrer wrote: > Having trouble with gftp on a Fedora Core 3 x86_64 system. > I can connect between systems but once I try to copy files the gftp > crashes. Not a good sign that RH thinks x86_64 has a place on the desktop yet. I entered that one in bugzilla for FC2 as #124650 on 2004-05-28, it was later consolidated into #140567 so someone actually read it and a fix is in the followup posts, but it still lives in FC3? -- John M. http://www.beau.org/~jmorris This post is 100% M$Free! Geekcode 3.1:GCS C+++ UL++++$ P++ L+++ W++ w--- Y++ b++ 5+++ R tv- e* r From devnull at adc.idt.com Tue Mar 29 21:33:20 2005 From: devnull at adc.idt.com (devnull at adc.idt.com) Date: Tue, 29 Mar 2005 16:33:20 -0500 (EST) Subject: OT: acroread on AMD64 Message-ID: Has anyone give acroread 7 on Linux a whirl on their AMD64 machines. Everytime I try and open a document, acroread crashes with (acroread:3808): GdkPixbuf-WARNING **: Error loading XPM image loader: Unable to load image-loading module: /usr/lib64/gtk-2.0/2.4.0/loaders/libpixbufloader-xpm.so: /usr/lib64/gtk-2.0/2.4.0/loaders/libpixbufloader-xpm.so: cannot open shared object file: No such file or directory I guess it should be using the libs under /usr/lib and not /usr/lib64 Any ideas ? /dev/null devnull at adc.idt.com From rchiodin at bellsouth.net Wed Mar 30 14:27:17 2005 From: rchiodin at bellsouth.net (Bob Chiodini) Date: Wed, 30 Mar 2005 09:27:17 -0500 Subject: OT: acroread on AMD64 In-Reply-To: References: Message-ID: <1112192837.8399.148.camel@tweedy.ksc.nasa.gov> On Tue, 2005-03-29 at 16:33 -0500, devnull at adc.idt.com wrote: > Has anyone give acroread 7 on Linux a whirl on their AMD64 machines. > > Everytime I try and open a document, acroread crashes with > > (acroread:3808): GdkPixbuf-WARNING **: Error loading XPM image loader: > Unable to load image-loading module: > /usr/lib64/gtk-2.0/2.4.0/loaders/libpixbufloader-xpm.so: > /usr/lib64/gtk-2.0/2.4.0/loaders/libpixbufloader-xpm.so: cannot open shared object file: No such file or directory > > I guess it should be using the libs under /usr/lib and not /usr/lib64 > > Any ideas ? > > > /dev/null > > devnull at adc.idt.com > > Is acroread a 32 or 64 bit app? Bob... From berryja at gmail.com Wed Mar 30 17:31:09 2005 From: berryja at gmail.com (Jonathan Berry) Date: Wed, 30 Mar 2005 11:31:09 -0600 Subject: OT: acroread on AMD64 In-Reply-To: <1112192837.8399.148.camel@tweedy.ksc.nasa.gov> References: <1112192837.8399.148.camel@tweedy.ksc.nasa.gov> Message-ID: <8767947e0503300931481de971@mail.gmail.com> On Wed, 30 Mar 2005 09:27:17 -0500, Bob Chiodini wrote: > On Tue, 2005-03-29 at 16:33 -0500, devnull at adc.idt.com wrote: > > Has anyone give acroread 7 on Linux a whirl on their AMD64 machines. > > > > Everytime I try and open a document, acroread crashes with > > > > (acroread:3808): GdkPixbuf-WARNING **: Error loading XPM image loader: > > Unable to load image-loading module: > > /usr/lib64/gtk-2.0/2.4.0/loaders/libpixbufloader-xpm.so: > > /usr/lib64/gtk-2.0/2.4.0/loaders/libpixbufloader-xpm.so: cannot open shared object file: No such file or directory > > > > I guess it should be using the libs under /usr/lib and not /usr/lib64 > > > > Any ideas ? > > > > > > /dev/null > > > > devnull at adc.idt.com > > > > Is acroread a 32 or 64 bit app? > > Bob... It's a 32-bit app, and still not a full release version (for version 7). It seems to work fine for me. Does the file mentioned indeed exist on your system? What about the 32-bit equivalent? I'm running FC3 x86_64. Jonathan From devnull at adc.idt.com Wed Mar 30 19:33:26 2005 From: devnull at adc.idt.com (devnull at adc.idt.com) Date: Wed, 30 Mar 2005 14:33:26 -0500 (EST) Subject: OT: acroread on AMD64 In-Reply-To: <8767947e0503300931481de971@mail.gmail.com> References: <1112192837.8399.148.camel@tweedy.ksc.nasa.gov> <8767947e0503300931481de971@mail.gmail.com> Message-ID: >> Is acroread a 32 or 64 bit app? >> >> Bob... > > It's a 32-bit app, and still not a full release version (for version > 7). It seems to work fine for me. Does the file mentioned indeed > exist on your system? What about the 32-bit equivalent? I'm running > FC3 x86_64. Good to hear that it works well. Is it an upgrade worth doing ? The mentioned file exists(obviously cant be used), the 32-bit equivalents exist too. Can you send me your ld.so.conf from under /etc Thanks. From berryja at gmail.com Wed Mar 30 19:50:28 2005 From: berryja at gmail.com (Jonathan Berry) Date: Wed, 30 Mar 2005 13:50:28 -0600 Subject: OT: acroread on AMD64 In-Reply-To: References: <1112192837.8399.148.camel@tweedy.ksc.nasa.gov> <8767947e0503300931481de971@mail.gmail.com> Message-ID: <8767947e050330115031c64def@mail.gmail.com> On Wed, 30 Mar 2005 14:33:26 -0500 (EST), devnull at adc.idt.com wrote: > >> Is acroread a 32 or 64 bit app? > >> > >> Bob... > > > > It's a 32-bit app, and still not a full release version (for version > > 7). It seems to work fine for me. Does the file mentioned indeed > > exist on your system? What about the 32-bit equivalent? I'm running > > FC3 x86_64. > > Good to hear that it works well. Is it an upgrade worth doing ? Is an upgrade of what worth doing? You mean to version 7? Yes, version 7 is much better than the archaic version 5 that is currently available. It looks a lot nicer and works better, like the scroll wheel actually works. > The mentioned file exists(obviously cant be used), the 32-bit equivalents > exist too. > > Can you send me your ld.so.conf from under /etc > > Thanks. Sure. What does this file do? Looks like there is not much there. $ cat /etc/ld.so.conf include ld.so.conf.d/*.conf /usr/X11R6/lib /usr/X11R6/lib64 /usr/lib/mysql /usr/lib64/mysql $ ls /etc/ld.so.conf.d/ qt-i386.conf qt-x86_64.conf xorg-x11-i386.conf xorg-x11-x86_64.conf $ more /etc/ld.so.conf.d/* :::::::::::::: /etc/ld.so.conf.d/qt-i386.conf :::::::::::::: /usr/lib/qt-3.3/lib :::::::::::::: /etc/ld.so.conf.d/qt-x86_64.conf :::::::::::::: /usr/lib64/qt-3.3/lib :::::::::::::: /etc/ld.so.conf.d/xorg-x11-i386.conf :::::::::::::: /usr/X11R6/lib :::::::::::::: /etc/ld.so.conf.d/xorg-x11-x86_64.conf :::::::::::::: /usr/X11R6/lib64 Hope this helps. Jonathan From eugen at leitl.org Thu Mar 31 07:23:43 2005 From: eugen at leitl.org (Eugen Leitl) Date: Thu, 31 Mar 2005 09:23:43 +0200 Subject: OT: acroread on AMD64 In-Reply-To: <8767947e0503300931481de971@mail.gmail.com> References: <1112192837.8399.148.camel@tweedy.ksc.nasa.gov> <8767947e0503300931481de971@mail.gmail.com> Message-ID: <20050331072342.GO24702@leitl.org> On Wed, Mar 30, 2005 at 11:31:09AM -0600, Jonathan Berry wrote: > It's a 32-bit app, and still not a full release version (for version It's a beta, prematurely released due to upcoming tax deadline in U.K., or something. > 7). It seems to work fine for me. Does the file mentioned indeed > exist on your system? What about the 32-bit equivalent? I'm running > FC3 x86_64. Ditto here: [kiki at helium ~]$ acroread => JavaScript Console: Acrobat EScript Built-in Functions Version 7.0 => JavaScript Console: Acrobat Annotations / Collaboration Built-in Functions Ve rsion 7.0 => JavaScript Console: Acrobat Annotations / Collaboration Built-in Wizard Funct ions Version 7.0 => JavaScript Console: Acrobat EScript Initial Functions Version 7.0 => JavaScript Console: Acrobat SOAP 7.0 -- Eugen* Leitl leitl ______________________________________________________________ ICBM: 48.07078, 11.61144 http://www.leitl.org 8B29F6BE: 099D 78BA 2FD3 B014 B08A 7779 75B0 2443 8B29 F6BE http://moleculardevices.org http://nanomachines.net -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 198 bytes Desc: not available URL: