Re: dealing with deffective RAM
On Aug 19, 2004, at 11:47 PM, Laurentiu Pancescu wrote: I've tested the entire week, with some weird results. No errors are found inside any of the two modules, when I test them separately (in slot one). When I have both modules, I get a few hundreds errors. I've cleaned the slots and the contacts, but no change. The funny thing is that all the errors seem located in the 0-128M range, even if I swap the position of the two modules. I've tested multiple times, for up to 16 hours at a time, and the results are consistent: no errors with a single module, a lot of them with both inserted, no matter which is the order of the modules (original, or swapped). I assume it's the fault of the motherboard, not of the memory modules. What motherboard and chipset? We had memory problems (at least when testing the memory -- not sure if there were problems running a real OS) with the Abit KG-7 which has the AMD 760 and Via split chipset... Memory would give lots of errors in one specific range, no matter which DIMM we used, but when tested in another motherboard that had a straight AMD (not split between an AMD North? bridge and Via South? bridge) chipset there were no errors. We replaced the motherboard with something else. Chad Thanks eveybody for the answers - I think I should get a new motherboard and processor. Laurentiu --- Gary Kline [EMAIL PROTECTED] schrieb: If you're as cheap/thrify as many of us, it may seem worth the effort to test. But it's pretty likely that if one of your 128 sticks is one, the other one will soon follow. Nutshell, Kris is right. If your time is utterly free, go ahead. I've learned that it pays to bite the bullet and buy new and top-rated memory. I'd go for a 256MB stick if/when you want to upgrade. (sign me been-there) || gary ___ Gesendet von Yahoo! Mail - Jetzt mit 100MB Speicher kostenlos - Hier anmelden: http://mail.yahoo.de ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED] ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: dealing with deffective RAM
--- Chad Leigh -- Shire.Net LLC [EMAIL PROTECTED] schrieb: What motherboard and chipset? It's a Chaintech 7AJA0, based on VIA KT133. The chipset has some known problems: there was a registry patch for Windows 2000, that avoided hard hangs (the IDE controller and the processor ended up waiting for each other to free the PCI bus, for periods between a few seconds and three hours, according to the documentation). They claimed no other OS was affected by this (they had Linux drivers on the CD, so I assume other included Linux). When trying to install Windows 2000, this happened almost every time, either during hardware detection, or during the final DCOM registration. It happened just once or twice to get hardware hanging under Linux, in about 2.5 years of exclusive usage (even under heavy load). Since I use FreeBSD 4, about 6 months ago), I didn't get any hardware hangs, but some programs sigsegv-ed unexpectedly, and unreproducible. Maybe this is related to the memory, not the chipset issue. We replaced the motherboard with something else. It looks like the best thing to do, imho. I hope high-quality boards from Asus, Epox or MSI don't exhibit such problems. Laurentiu ___ Gesendet von Yahoo! Mail - Jetzt mit 100MB Speicher kostenlos - Hier anmelden: http://mail.yahoo.de ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: dealing with deffective RAM
On Aug 20, 2004, at 12:47 AM, Laurentiu Pancescu wrote: I've tested multiple times, for up to 16 hours at a time, and the results are consistent: no errors with a single module, a lot of them with both inserted, no matter which is the order of the modules (original, or swapped). I assume it's the fault of the motherboard, not of the memory modules. It could be your power supply. With age electrolytic capacitors loose capacity and detune the filters in the power supply. Noise gets thru. Errors occur. -- David Kelly N4HHE, [EMAIL PROTECTED] Top-posters will not be shown the honor of a reply. ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: dealing with deffective RAM
I've tested the entire week, with some weird results. No errors are found inside any of the two modules, when I test them separately (in slot one). When I have both modules, I get a few hundreds errors. I've cleaned the slots and the contacts, but no change. The funny thing is that all the errors seem located in the 0-128M range, even if I swap the position of the two modules. I've tested multiple times, for up to 16 hours at a time, and the results are consistent: no errors with a single module, a lot of them with both inserted, no matter which is the order of the modules (original, or swapped). I assume it's the fault of the motherboard, not of the memory modules. Thanks eveybody for the answers - I think I should get a new motherboard and processor. Laurentiu --- Gary Kline [EMAIL PROTECTED] schrieb: If you're as cheap/thrify as many of us, it may seem worth the effort to test. But it's pretty likely that if one of your 128 sticks is one, the other one will soon follow. Nutshell, Kris is right. If your time is utterly free, go ahead. I've learned that it pays to bite the bullet and buy new and top-rated memory. I'd go for a 256MB stick if/when you want to upgrade. (sign me been-there) || gary ___ Gesendet von Yahoo! Mail - Jetzt mit 100MB Speicher kostenlos - Hier anmelden: http://mail.yahoo.de ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: dealing with deffective RAM
On Sunday 15 August 2004 21:31, Laurentiu Pancescu wrote: I was afraid this is what I'll be told, you're probably right. I have two 128M modules, I'll try to find the faulty one, by running the tests just with one at a time. If it's one of them (could also be the processor, or the mainboard, right?), is it better to buy a replacement for the defective one, or just a single 256M module, to avoid mismatches between the chips? MB is a Matsonic/Chaintech 7AJA0 (I know, it's cheap and not very good - that's what Compaq decided to put inside Presario), KT833-based, hosting an Athlon 1100MHz. If 256Mb will fit in one slot instead of 2 slots with 128Mb then go for it. As others have stated, buy decent ram. Been there myself - add up your travel+time and it just isn't worth it. Fwiw, it's only once I got decent ram that I was able to diagnose defective L2 cache. Not that I want to worry you mind! Remember that gcc and bzip are excellent torture tests. Set g++ off on something heavy and have bzip2 compress/decompress all night in tandem. If it's still alive in the morning, consider yourself reliable. ;-) ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
dealing with deffective RAM
Hello! Both Memtest86 and Memtest86+ find some failures in RAM (one finds 11 faults, the other 14 - most 32-bit, but some are only 8-bit wide). How can I deal with this in FreeBSD? Buying new RAM modules is probably the best choice, but I have no guarantee that the new modules will be perfectly ok, so it might be wasted money. Under Linux, I used the badram patch (http://rick.vanrein.org/linux/badram/). Is there similar functionality in the FreeBSD kernel, or is there such a patch? I failed to find anything on Google, and boot(8) didn't help much, either. In case such a patch doesn't exist, do you have any ideas about how such a thing might be implemented? Eventually, could you point me to the right place in the kernel source? I use Linux since 1997, but I'm pretty much a FreeBSD newbie, so please be merciful... :) Best regards, Laurentiu P.S. I assume the physical pages (4k/4M) where the faults reside could just be not mapped at all in the virtual memory manager, so it should be doable, right? ___ Gesendet von Yahoo! Mail - Jetzt mit 100MB Speicher kostenlos - Hier anmelden: http://mail.yahoo.de ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: dealing with deffective RAM
In the last episode (Aug 15), Laurentiu Pancescu said: Both Memtest86 and Memtest86+ find some failures in RAM (one finds 11 faults, the other 14 - most 32-bit, but some are only 8-bit wide). How can I deal with this in FreeBSD? Buying new RAM modules is probably the best choice, but I have no guarantee that the new modules will be perfectly ok, so it might be wasted money. Under Sure you do. It's called a warranty. If it's bad, return it. -- Dan Nelson [EMAIL PROTECTED] ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: dealing with deffective RAM
On Sun, Aug 15, 2004 at 05:49:48PM +0200, Laurentiu Pancescu wrote: Hello! Both Memtest86 and Memtest86+ find some failures in RAM (one finds 11 faults, the other 14 - most 32-bit, but some are only 8-bit wide). How can I deal with this in FreeBSD? Remove the defective ones and replace them - don't waste time trying to squeeze life out of the damaged hardware. They're dead, and there are probably other faults that the memory testers didn't find. Kris pgpya9z3lEMzO.pgp Description: PGP signature
Re: dealing with deffective RAM
I was afraid this is what I'll be told, you're probably right. I have two 128M modules, I'll try to find the faulty one, by running the tests just with one at a time. If it's one of them (could also be the processor, or the mainboard, right?), is it better to buy a replacement for the defective one, or just a single 256M module, to avoid mismatches between the chips? MB is a Matsonic/Chaintech 7AJA0 (I know, it's cheap and not very good - that's what Compaq decided to put inside Presario), KT833-based, hosting an Athlon 1100MHz. Thanks, Laurentiu --- Kris Kennaway [EMAIL PROTECTED] schrieb: Remove the defective ones and replace them - don't waste time trying to squeeze life out of the damaged hardware. They're dead, and there are probably other faults that the memory testers didn't find. Kris ___ Gesendet von Yahoo! Mail - Jetzt mit 100MB Speicher kostenlos - Hier anmelden: http://mail.yahoo.de ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: dealing with deffective RAM
On Sun, Aug 15, 2004 at 10:31:41PM +0200, Laurentiu Pancescu wrote: I was afraid this is what I'll be told, you're probably right. I have two 128M modules, I'll try to find the faulty one, by running the tests just with one at a time. If it's one of them (could also be the processor, or the mainboard, right?), is it better to buy a replacement for the defective one, or just a single 256M module, to avoid mismatches between the chips? MB is a Matsonic/Chaintech 7AJA0 (I know, it's cheap and not very good - that's what Compaq decided to put inside Presario), KT833-based, hosting an Athlon 1100MHz. If you're as cheap/thrify as many of us, it may seem worth the effort to test. But it's pretty likely that if one of your 128 sticks is one, the other one will soon follow. Nutshell, Kris is right. If your time is utterly free, go ahead. I've learned that it pays to bite the bullet and buy new and top-rated memory. I'd go for a 256MB stick if/when you want to upgrade. (sign me been-there) || gary Thanks, Laurentiu --- Kris Kennaway [EMAIL PROTECTED] schrieb: Remove the defective ones and replace them - don't waste time trying to squeeze life out of the damaged hardware. They're dead, and there are probably other faults that the memory testers didn't find. Kris -- Gary Kline [EMAIL PROTECTED] www.thought.org Public service Unix ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]