From gene at czarc.net Mon Jul 4 16:57:56 2005 From: gene at czarc.net (Gene Czarcinski) Date: Mon, 4 Jul 2005 12:57:56 -0400 Subject: Athlon64 X2 Message-ID: <200507041257.56592.gene@czarc.net> Any experience with the Athlon64 X2 processors and their supporting motherboards yet? Gene From cochranb at speakeasy.net Mon Jul 4 17:40:07 2005 From: cochranb at speakeasy.net (Robert L Cochran) Date: Mon, 04 Jul 2005 13:40:07 -0400 Subject: Athlon64 X2 In-Reply-To: <200507041257.56592.gene@czarc.net> References: <200507041257.56592.gene@czarc.net> Message-ID: <42C97477.5050407@speakeasy.net> I'm interested in this same thing. I haved an MSI K8N Neo2 motherboard with an Athlon 64 3500+, and I want to upgrade to the X2 processors. I also want to start programming specifically for dual cores -- I think for that your code needs to start multiple threads. I need to start looking for documentation that explains how to do this. MSI has not yet released an update to the current 1.8 BIOS that will support dual core CPUs. There are people who are playing with beta bios'es that do support the X2 and are having results from them though. Bob Gene Czarcinski wrote: >Any experience with the Athlon64 X2 processors and their supporting >motherboards yet? > >Gene > > > From cochranb at speakeasy.net Tue Jul 5 22:48:09 2005 From: cochranb at speakeasy.net (Robert L Cochran) Date: Tue, 05 Jul 2005 18:48:09 -0400 Subject: Athlon64 X2 In-Reply-To: References: Message-ID: <42CB0E29.6010201@speakeasy.net> Mark Hahn wrote: >>with an Athlon 64 3500+, and I want to upgrade to the X2 processors. I >>also want to start programming specifically for dual cores -- I think >> >> > >you can practice by getting a dual-socket board with single-core chips; >there's nothing new, at least at user-level, to dual-core. > > That would be sort of expensive for me. hopefully MSI will release a BIOS update real soon now. > > >>for that your code needs to start multiple threads. I need to start >>looking for documentation that explains how to do this. >> >> > >yes, you need >1 thread or process. but that can be had using many different >programming interfaces: fork, threads, openmp, mpi, etc. > > Speaking as a beginner in C where can I find good sample code that does this? How can I programmatically check for multiple CPUs or a multicore CPU? How can I programmatically send one thread to a specific CPU or CPU core? Or do I just create multiple threads and let the OS decide where to send them? >regards, mark hahn. > > > > > From jch at scalix.com Wed Jul 6 07:45:06 2005 From: jch at scalix.com (John Haxby) Date: Wed, 6 Jul 2005 08:45:06 +0100 Subject: Athlon64 X2 In-Reply-To: <42CB0E29.6010201@speakeasy.net> References: Message-ID: <42CB8C02.8030702@scalix.com> Robert L Cochran wrote: > Speaking as a beginner in C where can I find good sample code that > does this? How can I programmatically check for multiple CPUs or a > multicore CPU? How can I programmatically send one thread to a > specific CPU or CPU core? Or do I just create multiple threads and let > the OS decide where to send them? The very nature of Linux means that the vast majority of programs don't need to know or care about the number of CPUs in a system, whether physically separate, dual core or hyperthreaded. My relatively idle machine has, at the moment, 127 processors and if it had more than one processor the kernel would be cheerfully scheduling processes to all the processors. In many cases the only multi-threaded programming you need is a collection of cooperating processes and all multiple CPUs give you is a greater likelihood of race conditions. If the programs you're writing would benefit from concurrency or parallelism then you need to be concerned about multiple threads. If you're wondering what those terms mean then you need a beginning book on threads (there are quite a lot out there for a variety of programming languages, pick one you like the look of). Once you're over the beginning then you need to be concerned more with the pattens used for multi-threaded programs: my personal recommendation would be for Doug Lea's Java threads book -- I wish I could remember the title, but I don't think he's written many books about thread patterns. Yes, it's Java, but the concepts are generic. The last thing you were asking about is processor affinity. If you're hacking the kernel, you'll know about this. Chances are you'll never need it at a user level; although the pthreads interface does allow setting CPU affinity for an individual thread. There are supposed to be a few pathological cases where assigning a thread to a specific processor (or pool of processors) is the right thing to do. jch From jay at scherrer.com Wed Jul 6 14:17:10 2005 From: jay at scherrer.com (Jay Scherrer) Date: Wed, 06 Jul 2005 07:17:10 -0700 Subject: Athlon64 X2 In-Reply-To: <42CB8C02.8030702@scalix.com> References: <42CB8C02.8030702@scalix.com> Message-ID: <1120659430.3404.9.camel@gimly.scherco.local> On Wed, 2005-07-06 at 08:45 +0100, John Haxby wrote: > Robert L Cochran wrote: > > > Speaking as a beginner in C where can I find good sample code that > > does this? How can I programmatically check for multiple CPUs or a > > multicore CPU? How can I programmatically send one thread to a > > specific CPU or CPU core? Or do I just create multiple threads and let > > the OS decide where to send them? > > The very nature of Linux means that the vast majority of programs don't > need to know or care about the number of CPUs in a system, whether > physically separate, dual core or hyperthreaded. My relatively idle > machine has, at the moment, 127 processors and if it had more than one > processor the kernel would be cheerfully scheduling processes to all the > processors. In many cases the only multi-threaded programming you need > is a collection of cooperating processes and all multiple CPUs give you > is a greater likelihood of race conditions. > > If the programs you're writing would benefit from concurrency or > parallelism then you need to be concerned about multiple threads. If > you're wondering what those terms mean then you need a beginning book on > threads (there are quite a lot out there for a variety of programming > languages, pick one you like the look of). Once you're over the > beginning then you need to be concerned more with the pattens used for > multi-threaded programs: my personal recommendation would be for Doug > Lea's Java threads book -- I wish I could remember the title, but I > don't think he's written many books about thread patterns. Yes, it's > Java, but the concepts are generic. > > The last thing you were asking about is processor affinity. If you're > hacking the kernel, you'll know about this. Chances are you'll never > need it at a user level; although the pthreads interface does allow > setting CPU affinity for an individual thread. There are supposed to > be a few pathological cases where assigning a thread to a specific > processor (or pool of processors) is the right thing to do. > > jch > Robert, Try this link to AMD 64 development . I found it very interesting and responsive. Jay Scherrer From tweeks at rackspace.com Wed Jul 6 15:42:41 2005 From: tweeks at rackspace.com (tweeks) Date: Wed, 6 Jul 2005 10:42:41 -0500 Subject: Athlon64 X2 In-Reply-To: <42CB0E29.6010201@speakeasy.net> References: <42CB0E29.6010201@speakeasy.net> Message-ID: <200507061042.41810.tweeks@rackspace.com> On Tuesday 05 July 2005 17:48, Robert L Cochran wrote: > >you can practice by getting a dual-socket board with single-core chips; > >there's nothing new, at least at user-level, to dual-core. > > That would be sort of expensive for me. hopefully MSI will release a > BIOS update real soon now. More expensive.. but faster. Two single core w/hyper-xport is faster than one dual core with shared hyper xport (On one die). Tweeks -- Thomas Weeks, Lead Sys. Engineer The Managed Hosting Specialist(TM) Rackspace Managed Hosting http://www.rackspace.com/ 9725 Datapoint Dr, #100 Email: San Antonio, TX 78229 (w)210.447.4451 (f)210.447.4041 From jay at scherrer.com Wed Jul 6 23:08:36 2005 From: jay at scherrer.com (Jay Scherrer) Date: Wed, 06 Jul 2005 16:08:36 -0700 Subject: unsigned rpms Message-ID: <1120691316.3056.9.camel@gimly.scherco.local> List, Is it normal that the default FC4 up2date repository have so many unsigned rpm packages? Or is this their way of distinguishing between actual commits? Jay Scherrer From jlabadie at acm.org Wed Jul 13 06:15:55 2005 From: jlabadie at acm.org (Jon LaBadie) Date: Wed, 13 Jul 2005 02:15:55 -0400 Subject: swap partition limits Message-ID: Has the 2GB limit per swap partition been raised in 64-bit versions of RHEL? From arjanv at redhat.com Wed Jul 13 07:01:30 2005 From: arjanv at redhat.com (Arjan van de Ven) Date: Wed, 13 Jul 2005 09:01:30 +0200 Subject: swap partition limits In-Reply-To: References: Message-ID: <1121238091.3959.2.camel@laptopd505.fenrus.org> On Wed, 2005-07-13 at 02:15 -0400, Jon LaBadie wrote: > Has the 2GB limit per swap partition been raised in > 64-bit versions of RHEL? the 2Gb swap partition limit doesn't exist in rhel3 or rhel4, no matter which architecture. (and in rhel 2.1 it's only due to the mkswap program) -------------- 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 gene at czarc.net Sat Jul 16 19:45:27 2005 From: gene at czarc.net (Gene Czarcinski) Date: Sat, 16 Jul 2005 15:45:27 -0400 Subject: X2 motherboards Message-ID: <200507161545.27895.gene@czarc.net> I just got an Abit AN8 SLI motherboard with an Athlon64 X2 (4400+) with an ATI Radeon X300 PCI-E video card ... looking real good so far. Both i386 and x86_64 versions of FC4 install and running with all hardware recognized. Gene From fedora at lyxx.de Sat Jul 16 20:26:51 2005 From: fedora at lyxx.de (Benjamin Duehring) Date: Sat, 16 Jul 2005 22:26:51 +0200 Subject: X2 motherboards In-Reply-To: <200507161545.27895.gene@czarc.net> References: <200507161545.27895.gene@czarc.net> Message-ID: <200507162226.51715.fedora@lyxx.de> On Saturday 16 July 2005 21:45, Gene Czarcinski wrote: > I just got an Abit AN8 SLI motherboard with an Athlon64 X2 (4400+) with an ATI > Radeon X300 PCI-E video card ... looking real good so far. Both i386 and > x86_64 versions of FC4 install and running with all hardware recognized. > > Gene > Could you please post /proc/cpuinfo ? I would be interested what it says there. __________________________ Benjamin Duehring, FORWISS| Universitaet Passau | Priv. e-mail : duehring at forwiss.uni-passau.de D-94030 Passau | WWW : http://www.lyxx.de/card From gene at czarc.net Sun Jul 17 17:08:04 2005 From: gene at czarc.net (Gene Czarcinski) Date: Sun, 17 Jul 2005 13:08:04 -0400 Subject: X2 motherboards In-Reply-To: <200507162226.51715.fedora@lyxx.de> References: <200507161545.27895.gene@czarc.net> <200507162226.51715.fedora@lyxx.de> Message-ID: <200507171308.04827.gene@czarc.net> On Saturday 16 July 2005 16:26, Benjamin Duehring wrote: > On Saturday 16 July 2005 21:45, Gene Czarcinski wrote: > > I just got an Abit AN8 SLI motherboard with an Athlon64 X2 (4400+) with > > an ATI Radeon X300 PCI-E video card ... looking real good so far. ?Both > > i386 and x86_64 versions of FC4 install and running with all hardware > > recognized. > > > > Gene > > Could you please post ?/proc/cpuinfo ? > I would be interested what it says there. processor : 0 vendor_id : AuthenticAMD cpu family : 15 model : 35 model name : AMD Athlon(tm) 64 X2 Dual Core Processor 4400+ stepping : 2 cpu MHz : 2210.221 cache size : 1024 KB physical id : 0 siblings : 2 core id : 0 cpu cores : 2 fpu : yes fpu_exception : yes cpuid level : 1 wp : yes flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt lm 3dnowext 3dnow pni lahf_lm cmp_legacy bogomips : 4374.52 TLB size : 1024 4K pages clflush size : 64 cache_alignment : 64 address sizes : 40 bits physical, 48 bits virtual power management: ts fid vid ttp processor : 1 vendor_id : AuthenticAMD cpu family : 15 model : 35 model name : AMD Athlon(tm) 64 X2 Dual Core Processor 4400+ stepping : 2 cpu MHz : 2210.221 cache size : 1024 KB physical id : 0 siblings : 2 core id : 1 cpu cores : 2 fpu : yes fpu_exception : yes cpuid level : 1 wp : yes flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt lm 3dnowext 3dnow pni lahf_lm cmp_legacy bogomips : 4407.29 TLB size : 1024 4K pages clflush size : 64 cache_alignment : 64 address sizes : 40 bits physical, 48 bits virtual power management: ts fid vid ttp From vamsee_k at students.iiit.net Sun Jul 17 17:26:36 2005 From: vamsee_k at students.iiit.net (G. Vamsee Krishna) Date: Sun, 17 Jul 2005 22:56:36 +0530 (IST) Subject: X2 motherboards In-Reply-To: <200507171308.04827.gene@czarc.net> References: <200507161545.27895.gene@czarc.net> <200507162226.51715.fedora@lyxx.de> <200507171308.04827.gene@czarc.net> Message-ID: On Sun, 17 Jul 2005, Gene Czarcinski wrote: > > processor : 1 > vendor_id : AuthenticAMD > cpu family : 15 > model : 35 > model name : AMD Athlon(tm) 64 X2 Dual Core Processor 4400+ > stepping : 2 > cpu MHz : 2210.221 > cache size : 1024 KB > physical id : 0 > siblings : 2 > core id : 1 > cpu cores : 2 > fpu : yes > fpu_exception : yes > cpuid level : 1 > wp : yes > flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca > cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt lm > 3dnowext 3dnow pni lahf_lm cmp_legacy > bogomips : 4407.29 > TLB size : 1024 4K pages > clflush size : 64 > cache_alignment : 64 > address sizes : 40 bits physical, 48 bits virtual > power management: ts fid vid ttp > That's some real specs. Impressive. And I got a question. On a single-core processor, we can run only ONE operating system at a time. That OS takes care of all the management behind the scenes. So, is it really possible to run 2 OSs at a time on a dual-core processor? Regards, Vamsee -- Playfully doing something difficult, whether useful or not, that is hacking. -- Richard Stallman From webmaster at margo.bijoux.nom.br Sun Jul 17 20:01:27 2005 From: webmaster at margo.bijoux.nom.br (Pedro Fernandes Macedo) Date: Sun, 17 Jul 2005 17:01:27 -0300 Subject: X2 motherboards In-Reply-To: References: <200507161545.27895.gene@czarc.net> <200507162226.51715.fedora@lyxx.de> <200507171308.04827.gene@czarc.net> Message-ID: <42DAB917.6020308@margo.bijoux.nom.br> G. Vamsee Krishna wrote: > That's some real specs. Impressive. > > And I got a question. On a single-core processor, we can run only ONE > operating system at a time. That OS takes care of all the management > behind the scenes. So, is it really possible to run 2 OSs at a time on > a dual-core processor? AFAIK , on current dual core CPUs , no. There's some technologies being worked on (I think one of them is called Vanderpool by Intel ) that theoretically will allow virtualization on the CPU level... Probably it'll be something like Xen on steroids... -- Pedro Macedo From eugen at leitl.org Sun Jul 17 20:37:34 2005 From: eugen at leitl.org (Eugen Leitl) Date: Sun, 17 Jul 2005 22:37:34 +0200 Subject: X2 motherboards In-Reply-To: <42DAB917.6020308@margo.bijoux.nom.br> References: <200507161545.27895.gene@czarc.net> <200507162226.51715.fedora@lyxx.de> <200507171308.04827.gene@czarc.net> <42DAB917.6020308@margo.bijoux.nom.br> Message-ID: <20050717203734.GW4317@leitl.org> On Sun, Jul 17, 2005 at 05:01:27PM -0300, Pedro Fernandes Macedo wrote: > G. Vamsee Krishna wrote: > > >That's some real specs. Impressive. > > > >And I got a question. On a single-core processor, we can run only ONE > >operating system at a time. That OS takes care of all the management > >behind the scenes. So, is it really possible to run 2 OSs at a time on > >a dual-core processor? > > AFAIK , on current dual core CPUs , no. There's some technologies being > worked on (I think one of them is called Vanderpool by Intel ) that > theoretically will allow virtualization on the CPU level... Probably > it'll be something like Xen on steroids... Most people who run Xen actually need the VServer patch, they just haven't heard. -- Eugen* Leitl leitl ______________________________________________________________ ICBM: 48.07100, 11.36820 http://www.leitl.org 8B29F6BE: 099D 78BA 2FD3 B014 B08A 7779 75B0 2443 8B29 F6BE -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 189 bytes Desc: Digital signature URL: From ericwort at uiuc.edu Thu Jul 28 21:45:41 2005 From: ericwort at uiuc.edu (Eric Wort) Date: Thu, 28 Jul 2005 16:45:41 -0500 Subject: HDAMA Onboard SATA issues with installed RAM > 4GB Message-ID: <42E95205.9060509@uiuc.edu> First off, I have never been able to get this machine to boot any release of Fedora Core with more than 4GB of RAM installed, as it starts having SATA seek errors. Configuration is as follows: Dual Opteron 248's in a RioWorks HDAMA board 4 or 8 GB worth of 1GB Corsair registered DDR400 sticks 2 SATA 74gig Raptors (for the OS) 2 SATA 200gig Seagates 3 PATA 200gig Seagates SATA drives are hooked to the Promise PDC20319 onboard controller Upon trying to upgrade to FC4 from FC2 with 4GB of RAM in the box, the machine would crash just after detecting the videocard in anaconda. For some reason I decided to randomly toss in the other 4 sticks(bringing the total to 8GB) and try the install again, which worked perfectly during the entire install. Upon rebooting the fresh FC4 install, it had the same SATA seek errors as normal. Taking the box back down to 4GB allowed it to boot right up. Does anyone know why the install kernel would have zero problems with an hour long install, while the stock FC4smp kernel can't even boot without seek errors? I imagine it may be due to the install kernel being single processor, but I am not sure. Thanks, Eric Wort From harpreet at dhillonz.net Fri Jul 29 04:05:27 2005 From: harpreet at dhillonz.net (Harpreet Dhillon) Date: Thu, 28 Jul 2005 21:05:27 -0700 Subject: HDAMA Onboard SATA issues with installed RAM > 4GB In-Reply-To: <42E95205.9060509@uiuc.edu> References: <42E95205.9060509@uiuc.edu> Message-ID: <42E9AB07.10201@dhillonz.net> Does it boot fine when you select to boot from UP kernel? Try passing 'acpi=off' to kernel while booting and see if that helps. Some drivers still have issues with ACPI. - harpreet Eric Wort wrote: > First off, I have never been able to get this machine to boot any > release of Fedora Core with more than 4GB of RAM installed, as it > starts having SATA seek errors. > > Configuration is as follows: > Dual Opteron 248's in a RioWorks HDAMA board > 4 or 8 GB worth of 1GB Corsair registered DDR400 sticks > 2 SATA 74gig Raptors (for the OS) > 2 SATA 200gig Seagates > 3 PATA 200gig Seagates > SATA drives are hooked to the Promise PDC20319 onboard controller > > Upon trying to upgrade to FC4 from FC2 with 4GB of RAM in the box, the > machine would crash just after detecting the videocard in anaconda. > For some reason I decided to randomly toss in the other 4 > sticks(bringing the total to 8GB) and try the install again, which > worked perfectly during the entire install. Upon rebooting the fresh > FC4 install, it had the same SATA seek errors as normal. Taking the > box back down to 4GB allowed it to boot right up. Does anyone know > why the install kernel would have zero problems with an hour long > install, while the stock FC4smp kernel can't even boot without seek > errors? I imagine it may be due to the install kernel being single > processor, but I am not sure. > > Thanks, > Eric Wort > From gene at czarc.net Sat Jul 30 16:41:30 2005 From: gene at czarc.net (Gene Czarcinski) Date: Sat, 30 Jul 2005 12:41:30 -0400 Subject: x86_64 SMP kernels Message-ID: <200507301241.30517.gene@czarc.net> For systems running the x86_64 SMP kernel, you need to look at https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=164247 I have tested the patch and it appears to fix the OOPS which was causing me a lot of headaches on my Athlon64 X2 4400+ Gene From ericwort at uiuc.edu Sun Jul 31 19:05:02 2005 From: ericwort at uiuc.edu (Eric Wort) Date: Sun, 31 Jul 2005 14:05:02 -0500 Subject: HDAMA Onboard SATA issues with installed RAM > 4GB In-Reply-To: <42E9AB07.10201@dhillonz.net> References: <42E95205.9060509@uiuc.edu> <42E9AB07.10201@dhillonz.net> Message-ID: <42ED20DE.4050603@uiuc.edu> Thanks all for the assistance, but I found a Tyan proprietary sata_sil driver for SuSE 9.3 that contains the following text: 6. Known Restriction The driver could only work on a platform that has less than 4 giga memory. It looks like I may be out of luck with this SATA controller, I guess I'll go back to trying to get my HPT374 driver to not lock up with drives attached on boot. Eric Wort > Eric Wort wrote: > >> First off, I have never been able to get this machine to boot any >> release of Fedora Core with more than 4GB of RAM installed, as it >> starts having SATA seek errors. >> >> Configuration is as follows: >> Dual Opteron 248's in a RioWorks HDAMA board >> 4 or 8 GB worth of 1GB Corsair registered DDR400 sticks >> 2 SATA 74gig Raptors (for the OS) >> 2 SATA 200gig Seagates >> 3 PATA 200gig Seagates >> SATA drives are hooked to the Promise PDC20319 onboard controller >> >> Upon trying to upgrade to FC4 from FC2 with 4GB of RAM in the box, >> the machine would crash just after detecting the videocard in >> anaconda. For some reason I decided to randomly toss in the other 4 >> sticks(bringing the total to 8GB) and try the install again, which >> worked perfectly during the entire install. Upon rebooting the fresh >> FC4 install, it had the same SATA seek errors as normal. Taking the >> box back down to 4GB allowed it to boot right up. Does anyone know >> why the install kernel would have zero problems with an hour long >> install, while the stock FC4smp kernel can't even boot without seek >> errors? I imagine it may be due to the install kernel being single >> processor, but I am not sure. >> >> Thanks, >> Eric Wort >> >