Re: A new, unstable system, HW problem ???
[EMAIL PROTECTED] wrote: Well I bought kingston value ram DDR400 512M * 2 for the athlon64 I built and it works. I have had good luck with kingston and crucial personally. I don't bother to save $10 per 512M anymore by getting the generic stuff. The last generic stuff I bought was a PC100 CL2 256M and it never did run stable faster than CL3 (so much for their claim). Well, just an update in my quest-for-a-stable-and-fast-debian-amd64-system :) I went and replaced the memory modules yesterday to Kingston value ram. Better, but not perfect. It's actually strange. I first booted the machine and it reported some CMOS error...I then realised that the bios (and later memtest86) reports the memory as DDR 333, although the packages in which I received the memory said DDR 400? Anyhow, I set up the DRAM timings to DDR 333 in bios, and after that it booted, ran memtest86 and even kernel compile + copying of a large directory of files + watching a video *all at the same time* without problems ...yes, the last one was a psychologically important... :) So, now I can be 100% certain that it was memory incompatibility, just as you guys said! But I wont settle for DDR 333...out of principle I want to make this perfect now that I've started the war ;) So it's back to store for me, they recomended Buffalo memory yesterday, maybe I'll give that a shot... Thanks guys, I'm one step closer... juhis Have you upgraded the BIOS? It seems Abit has made some changes to memory settings in later versions. Personally, I don't even remember what every motherboard says about the memory speeds. To me, when it works and is sufficiently fast is the time when I lay back and ignore the details. If you want, I can go check some of the settings that my AV8 systems say about memory. I do recall at least two people mentioning that in some setups, with some position of the moon etc, DDR400 memory doesn't function at more than DDR333 speed. Unfortunately, I never paid much attention to the details here. /v\ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: A new, unstable system, HW problem ???
On Wed, Apr 27, 2005 at 10:36:01PM -0700, Chris Wakefield wrote: This is a by-the-way regarding RAM modules. I have a K8V deluxe and I'm sure I read somewhere last year that you lose some functionality by filling all 3 mem slots as opposed to just the first two. It was something about the speed of the RAM being slowed somewhat for some reasonDo you recall this or know about this? As far as I am aware, if you use more than two slots you must drop to 333Mhz speed. I believe it's a limitation of the memory controller (basically running too many loads on one memory bus means it has to run slower to work reliably). Socket 939 boards have two memory controllers and can hence run 2 dimm's per bus (4 total) at full speed, although there are probably some limitations on those modules too in some cases. Servers get around this by using registered (buffered) memory which avoids the bus load issues (but also introduces more delay in accessing memory so you loose a bit of speed to gain much higher memory limits). This is why opterons can run 8 or 16 dimms per cpu. Len Sorensen -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: A new, unstable system, HW problem ???
On Thu, Apr 28, 2005 at 03:05:18PM +0300, Kyuu Eturautti wrote: Have you upgraded the BIOS? It seems Abit has made some changes to memory settings in later versions. Personally, I don't even remember what every motherboard says about the memory speeds. To me, when it works and is sufficiently fast is the time when I lay back and ignore the details. If you want, I can go check some of the settings that my AV8 systems say about memory. I do recall at least two people mentioning that in some setups, with some position of the moon etc, DDR400 memory doesn't function at more than DDR333 speed. Unfortunately, I never paid much attention to the details here. DDR333 may not be much slower than DDR400, but DDR333 would be running async from the CPU while DDR400 would run synced with the cpu as far as I know, but it may not apply with the onchip memory controller (it may run it synced to the cpu just fine even at 333). Certainly on the athlon's with the memory controller in the chipset, you wanted memory running at the same speed as the cpu bus (so DDR333 was better than DDR400 for all but the top end barton's). But it might not be the case for the athlon 64, although if you can get it running at DDR400 it would be nice I guess. How much ram do you have installed and what size are the modules? Len Sorensen -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: A new, unstable system, HW problem ???
On Thursday 28 April 2005 14:43, Lennart Sorensen wrote: On Thu, Apr 28, 2005 at 03:05:18PM +0300, Kyuu Eturautti wrote: Have you upgraded the BIOS? It seems Abit has made some changes to memory settings in later versions. Personally, I don't even remember what every motherboard says about the memory speeds. To me, when it works and is sufficiently fast is the time when I lay back and ignore the details. If you want, I can go check some of the settings that my AV8 systems say about memory. I do recall at least two people mentioning that in some setups, with some position of the moon etc, DDR400 memory doesn't function at more than DDR333 speed. Unfortunately, I never paid much attention to the details here. DDR333 may not be much slower than DDR400, but DDR333 would be running async from the CPU while DDR400 would run synced with the cpu as far as I know, but it may not apply with the onchip memory controller (it may run it synced to the cpu just fine even at 333). This argument only applies to CPUs with a frontside bus (FSB) and a separate memory controller (eg. 32-bit Athlon or P4). On these systems it's desirable to have the FSB and memory running in sync at the same speed. Amd64 cpus don't have a FSB, so this isn't a problem. Paul -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: A new, unstable system, HW problem ???
Lainaus Kyuu Eturautti [EMAIL PROTECTED]: [EMAIL PROTECTED] wrote: Hi again, This is a collective answer, many thanks to all of you guys. So, the consensus seems to be memory problems, which is something that I suspected myself. I haven't tested thoroughly yet but I will do more testing tonight (remove one of the memory units, run memtest86+..something else?). Memtest86 usually works. Unfortunately, usually always, but it's the best place to start. Often enough, memory problems can be diagnosed by certain behaviour, and what you're reported seems to drop into that category. Ok, yesterday I ran the memtest86+ utility and indeed it showed some errors on the test 5 of the test series (but the test 8 showed no errors). And this is what the memtest86 webpage says... There have been numerous reports of errors with only tests 5 and 8 on Athlon systems. Often the memory works in a different system or the vendor insists that it is good What do you guys think? Can I be relatively certain that it's infact the memory modules that are just not compatible with amd64 + VIA K8T800 cobination? I think I'm going to replace the ram since when I bought it I was going for some abit approved memory, but the seller convinced me that this is as good but cheaper :) I also ran the prime95 utility from mersenne.org and that also fails in the first test. Don't know what to make of that, but somethings definately not right. Anyhow, I want to thank you all alot for helping me out, this is what I call great support :) Thanks, juhis -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: A new, unstable system, HW problem ???
Lennart Sorensen wrote: On Wed, Apr 27, 2005 at 09:14:09AM +0300, [EMAIL PROTECTED] wrote: Ok, yesterday I ran the memtest86+ utility and indeed it showed some errors on the test 5 of the test series (but the test 8 showed no errors). And this is what the memtest86 webpage says... There have been numerous reports of errors with only tests 5 and 8 on Athlon systems. Often the memory works in a different system or the vendor insists that it is good What do you guys think? Can I be relatively certain that it's infact the memory modules that are just not compatible with amd64 + VIA K8T800 cobination? I think I'm going to replace the ram since when I bought it I was going for some abit approved memory, but the seller convinced me that this is as good but cheaper :) It may be perfectly good working memory, just not if you require the timings the athlon64 memory controller expects from DDR400 memory. it really expects the memory to perform as it says in it's SPD eeprom. If it doesn't, expect trouble. Most memory controllers in chipsets seem slightly less picky (maybe because they already have more delays involved in getting requests to and from the memory.) Of course I also read somewhere that the newer athlon64's are getting some improvements in the memory controller to make them more flexible on memory modules (I think it was supposed to make it easier to run 4 dimm's too). I second this. The memory might be just fine, not just too compatible. A future BIOS release, or certain BIOS settings may or may not help. If in any way possible, test the system with another memory. At least we've seen Kingston Value Ram working nicely. Probably a big load of other brands too. Of course, if you bought the motherboard and memory from the same place, it would seem sensible that they'd be willing to provide another memory option. With the amount of such issues coming to my attention, I'm ready to define Twinmos as a rather bad idea for Athlon64. /v\ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: A new, unstable system, HW problem ???
Hi Len. This is a by-the-way regarding RAM modules. I have a K8V deluxe and I'm sure I read somewhere last year that you lose some functionality by filling all 3 mem slots as opposed to just the first two. It was something about the speed of the RAM being slowed somewhat for some reasonDo you recall this or know about this? Thanks, Chris W. On April 27, 2005 05:41 am, Lennart Sorensen wrote: On Wed, Apr 27, 2005 at 09:14:09AM +0300, [EMAIL PROTECTED] wrote: Ok, yesterday I ran the memtest86+ utility and indeed it showed some errors on the test 5 of the test series (but the test 8 showed no errors). And this is what the memtest86 webpage says... There have been numerous reports of errors with only tests 5 and 8 on Athlon systems. Often the memory works in a different system or the vendor insists that it is good What do you guys think? Can I be relatively certain that it's infact the memory modules that are just not compatible with amd64 + VIA K8T800 cobination? I think I'm going to replace the ram since when I bought it I was going for some abit approved memory, but the seller convinced me that this is as good but cheaper :) It may be perfectly good working memory, just not if you require the timings the athlon64 memory controller expects from DDR400 memory. it really expects the memory to perform as it says in it's SPD eeprom. If it doesn't, expect trouble. Most memory controllers in chipsets seem slightly less picky (maybe because they already have more delays involved in getting requests to and from the memory.) Of course I also read somewhere that the newer athlon64's are getting some improvements in the memory controller to make them more flexible on memory modules (I think it was supposed to make it easier to run 4 dimm's too). I also ran the prime95 utility from mersenne.org and that also fails in the first test. Don't know what to make of that, but somethings definately not right. Anyhow, I want to thank you all alot for helping me out, this is what I call great support :) Well I bought kingston value ram DDR400 512M * 2 for the athlon64 I built and it works. I have had good luck with kingston and crucial personally. I don't bother to save $10 per 512M anymore by getting the generic stuff. The last generic stuff I bought was a PC100 CL2 256M and it never did run stable faster than CL3 (so much for their claim). Len Sorensen -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: A new, unstable system, HW problem ???
[EMAIL PROTECTED] wrote: Hi again, This is a collective answer, many thanks to all of you guys. So, the consensus seems to be memory problems, which is something that I suspected myself. I haven't tested thoroughly yet but I will do more testing tonight (remove one of the memory units, run memtest86+..something else?). Memtest86 usually works. Unfortunately, usually always, but it's the best place to start. Often enough, memory problems can be diagnosed by certain behaviour, and what you're reported seems to drop into that category. I've tried with the default 2.6.8.? and the most recent 2.6.11.9 kernels in the 64bit installation and the most recent 2.6.11.? kernel in the 32bit installation. And I see no difference in behavior. But maybe I could try with a pata drive if I find an empty partition somewhere :) This is quite unlikely. I get the feeling that if VIA_SATA would have had any major issues, I would have come across them by now, and I don't recall reading anyone else having major problems with it either. In the case I mentioned earlier, kernel versions seemed to matter very little. I'm not sure if some ACPI/APIC or Cool'n'Quiet settings (BIOS and kernel) had minor effects, but in all cases, it crashed sooner or later. I'd say the odds are good for the memory being faulty. Not 100%, but high anyway. A faulty motherboard is also possible, but the issues you're having seem somewhat strange for a faulty mb. Certainly a BIOS upgrade can be tried, that's one possibility too. If there's a way for you to test another memory module here temporarily, that would probably go a long way. What I'm a bit concerned of is the amount of problems with this type of memory (Twinmos) and Athlon64 systems, at least Via K8T8xx based ones. Some minor digging revealed another such case. Either it's just a major issue of incompatibility on either side, or they've just had a lot of faulty modules produced. I have yet to run into problems with Kingston Value Ram, however I haven't done overclocking since 486 times, so I have no idea on their performance, I just go for stability. I'm sure enough there are more than one quality memory brands out there, it's just all about the money. /v\ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: A new, unstable system, HW problem ???
On 4/25/05, [EMAIL PROTECTED] [EMAIL PROTECTED] wrote: I've found one oops in my syslog, but I'm not sure what caused this oops. But I was hoping someone here could maybe point out the most probable faulty component in the system. I have found the Athlon64 to be very sensitive to memory settings. Try the standard settings in the BIOS. Try disabling dual channel memory access, or using only one of the two memory channels. You can also try memtest86, but at least on my system it does not catch all problems that I would attribute to the memory. And the only parts that aren't new in the computer are power unit and video card and I've tried two of each just be sure... Don't assume that every new component is necessarily perfect. I could also be that your power unit is not designed to have the CPU load on the 12V rail. Check the rating, it should be at least 15A. And SATA is another potential problem area. At least it is on my system. Maybe it is just the cable, or maybe the controller has its issues. Thomas
Re: A new, unstable system, HW problem ???
[EMAIL PROTECTED] wrote: Hello list, This mail might be partly off-topic, since I'm pretty sure it's my hardware that's at fault and not debian-amd64, but I don't know where else to ask, so I thought I'd try here, thanks for you patience... :) So, the situation is, that I upgraded my computer on weekend, with the following components: AMD64 3200+ (Winchester core) Abit AV8 mobo Seagate 7200.8 250Gb sata harddrive 2x512Mb 64bit DDR memory (M-Tec, Twinmos MT6464400D I quess) I've got several production systems with the Abit AV8 and actually I think pretty much the exact same hard disks in at least two systems, though with Kingston Value RAM memory. I'm quite ready to give the AV8 my personal rock solid recommendation. I agree with other posts - check the memory settings. Run memtest86.com to be sure. I haven't had any SATA issues with the Via controller, in fact everything from Via has functioned perfectly with Athlon 64's, including SATA and PATA hard disks. However - and I'm not sure if this is the same case here, I had some similar crash issues earlier. It was an MSI motherboard with Via K8T800 (the AV8 has K8T800 Pro), a single SATA drive and several kernel versions from 2.6.8 to 2.6.10 I believe. Crashes occured exactly the same way, after intense hard disk use. The fault was in the memory, which was M-Tech Twinmos as in your case. However - memtest86.com did not always detect the fault. However, after the memory was replaced with a Kingston module, it did not occur anymore. So, this could have been either of the two: either the memory module was faulty, or perhaps the Twinmos memory didn't function well with the motherboard, perhaps by reporting wrong speed or something. I'm no memory expert but this was a guess from another friend. /v\ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: A new, unstable system, HW problem ???
[EMAIL PROTECTED] wrote: Hello list, This mail might be partly off-topic, since I'm pretty sure it's my hardware that's at fault and not debian-amd64, but I don't know where else to ask, so I thought I'd try here, thanks for you patience... :) So, the situation is, that I upgraded my computer on weekend, with the following components: AMD64 3200+ (Winchester core) Abit AV8 mobo Seagate 7200.8 250Gb sata harddrive 2x512Mb 64bit DDR memory (M-Tec, Twinmos MT6464400D I quess) Does your power supply have the additional ATX12V supply? That's the square 4-pin connector. According to this picture, it's right next to the usual ATX power connector on your board. http://www.overclockzone.com/newhardware/abit/av8/IMG_5220.jpg A friend of mine had built a very unstable system before someone told him to get a power supply that had an ATX12V connector. -Corey -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
