Slashdot Mirror

If you just need to unlock accounts (if, say, you forgot Administrator's password), try this. There are bootable ISOs and floppy disk images.

It's not the most intuitive thing to use, but it is pretty easy if you follow the command prompts. It could probably be extended to include more tools like KNOPPIX.

Actually, something with that, KNOPPIX, and MemTest86 would be really nice and alleviate the need for 3 CDs for performing diagnostic tests on wonky PCs.

You may have been forgetting something to do with that Athlon 2100+ - its bus speed, which in the case of the 2100+ is 133MHz. At least one Athlon board I've seen defaults to 100MHz bus speed (does *not* autodetect bus speed), and those are the speeds you've been seeing. Also, many older Athlon boards have 100/133+ speeds set using a jumper, which was usually shipped as default to 100 (this has mostly gone to software now, thankfully); further bus speed configurations could be done using the BIOS.

You have to do this with Intel too whenever they bump the bus speed, unless they've changed the electrical connections on the socket or whatever as they usually do. Also, whenever you see random crashes, the first thing you think, always, is memory error; then you update board's BIOS, video, chipset drivers etc. I saw this recently on a board with a dodgy stick of RAM; causing very random crashes, despite the fact the dodgy RAM was >1GB.

You're not making adequate comparisons, anyway (despite what you said). I mean, Biostar compared to an Intel OEM board? Come on. I buy MSI for AMD builds, and I've never seen a bad one. ASUS, ABIT, Gigabyte etc. AMD boards are generally fine too, and the low end MSI KT600 is actually surprisingly good (if, as said before, you have to manually enter the bus speed, at least in software.)

In any case, Athlon 64 is a completetly different chip to AXP, so nothing from the past applies. It has a P4-style heat spreader. It's cheaper in the UK than the highest-end non-"extreme" P4 by about UKP100, it's about as fast, and according to the reviews I've read it actually produces less heat than its competitor, as have Athlon XPs since about the Thoroughbred (I've seen hot P4s c, 60C and cool Athlons c. 40C, both using the stock retail fan/compound. They're both quiet, too.) The cooling problem with AMD? No longer there, as long as you use an acceptable fan - and the retail one does come into that category.

Besides, my overclocking-freak acquaintance swears by them; if the OCers like it, it's *got* to be fine for the normal person. Wouldn't it?

Note: I do own Intel machines; my laptop is a P4-m, because at the time AMD laptops pretty much came with godawful integrated graphics. I think the situation is mostly the same today, unfortunately. My desktop is a fairly old Thunderbird Athlon 1GHz, and it works and always has worked fine. I have seen dodgy AMD machines from friends, but the main problem one was based on an ultradodgy PC Chips off-brand board that was unfortunately not the same as the Elite K7S5A, which can at least be made acceptable. It had the bus speed problem, too.

A bad CRC on a really-big gzip file isn't a very granular test. A good way of stressing a system so that it will show you wether it fails reproducibly is to compile a couple different versions of the Linux kernel, and see it if fails in the same spot (or at all). If you ran memtest on it, it'd likely have shown an error. Whenever you aren't sure, run memtest :)

I couldn't even run knoppix (let alone XP) on that computer (a SN41G2 shuttle) until I had the one stick swapped for another one.

If you get any errors on memtest86 (say after running it for 8 hours continuous) then you probably won't be able to install any OS (and/or experience random crashes).

Also, make sure you plug in a shitty pci videocard if you have a system which uses main memory for as display memory (like the nforce chipset). That segment of memory wouldn't get tested by memtest86. Setting that memory to 0 in the bios wouldn't help either. Hello? who turned off the lights ;-)

I agree completely. I run my machines for weeks with memtest86 before making them operational. And while I have found memory and motherboard problems, even with the most el-cheapo power supply it's either success or failure. This sort of reliably intermittent memory test failure that Anandtech regards as "normal" puzzles me to no end. Maybe it's just part of overclocker attitude?

Memtest86 home page

For some reason, I always want to call it memchk86, but it's memtest86, and it's indispensable. Memtest86 boots from a floppy and begins running immediately. It's got several types of test, and runs in a loop so if you have an intermittent problem, you can just boot it and leave it for a while.

The author has finally given in to popularity and set up a site for the program, MemTest86.com. I encourage anyone who's found bad ram with memtest to throw a few bucks his way. ("Professional" memory testers are nowhere near cheap!)

There's also a bootable CD .ISO for machines without a floppy drive. Some of my favorite machines don't have CDROMs either (subnotebook with a network card and nothing else), so I'd love to see a version I can download and invoke with Loadlin, but hey, there's only so much I can ask.

Well, this search returned quite a few hits on some good Redhat configuration tips and a (oudated) Linuxcare certification for a few distributions. It looks like if nothing else, Redhat 7.x is reputed to work well on it. The problem you are describing (something getting messed up after a few weeks of running it) is something I have never seen in Linux on any hardware.

If nothing else, Linux is known for being pretty rock solid stable once it is configured and running. Have you ruled out some kind of hardware problem? I had X locking up on me all the time for a while (windows ran just fine) and it turned out to be a bad memory chip. It was under warranty and after getting a new one it has worked without a problem. You might want to run memtest86 on it and see if that is what is happening to you.

Finkployd

The nForce2 is a nice chipset but their Linux drivers leave much to be desired, like working audio and firewire.

I've been using ALSA for audio. Freshrpms.net has ALSA (and other goodies) in convenient RPMs. I haven't tried using the FireWire port but everything else works. And it's trivial to run 400MHz FSB, regardless of your CPUs official spec. I'm running a 1700+ at 7.5x404MHz. MemTest86 verified the configuration.

Erm, there are actually a few programs out there called "memtest86." I promise you that if you're using this one then if that stick fails consistently (erratic RAM failure is pretty much unheard of outside of unusual operating environments) then memtest86 will find the problem.

This is the same program that Crucial uses to determine whether they should resell memory that's been sent back as bad.

Memtest86 will go a long way to test the ram. If you are going through tons of wanky ram, the issue may be your cpu or power supply however. Test the ram on a couple boxes.

As for no-name. Usually grade 'a' ram will run at a lower cas rating, where some of the generics might work at a higher (and slower) setting. Stuff that rates at PC-100 CAS 2 might only work at PC-133 CAS 3. (dang, showing my age) The good stuff tended to be able to run stable at the faster FSB and CAS settings. My time is worth more than the ~$30 bucks between solid and guesswork.

If your not pushing a system hard - cheap ram might just work. A few years back a local vendor had some dirt cheap no-name 128M sticks that ran as fast as my mushkin stuff. Go figure. You role the dice, but it matters less if you are not pushing your settings hard.

I've reclocked my Athlon 1700+ (TBred core) to 8x202MHz (404MHz DDR) on my ASUS nForce2 chipset board, using a single Corsair PC3200C2 DIMM (yes, two DIMMs would be better, but they were too expensive at the time). It's just a matter of selecting the right BIOS settings. I left the voltage levels at their defaults. MemTest86 verifies that the memory is stable at that speed. Red Hat Linux runs until I need to reboot for the usual kernel/glibc upgrades. I went this this approach because I wanted to optimize the performence/power consumption balance, what with the machine running 24x7 and all.

Of course, tweaking speeds like this is not guaranteed to work, yadda yadda, but it generally does if you built your system right.

If you want serious firepower, build a dual Athlon box, which should cost no more than the uniprocessor P4 being reviewed. time make reports a bit over 9 minutes when building Wine with MAKEFLAGS=-j2 on my dual 2400+ (not overclocked). Nice, especially when you forget the --with-nptl switch the first time around (d'oh!).

Of course, next week, the Opterons ship, starting with Opteron DP 240's and 242's. It's unclear whether there will be cheap workstation motherboards available right away or just the seriously nice (and expensive) Newisys-designed 1U rackmount servers. It appears that AMD is going to use the Opterons to slap the high-end P4's around, saving the Athlon 64 until they want a low-to-midrange 64-bit desktop platform. I'm surprised the various hardware site reviewers haven't picked up on this.

1). tomsrtbt Linux on a floppy - essential!
2). Windows 98SE boot floppy
3). Knoppix 3.2 bottable Linux on a CD.
4). Memtest86 bootable CD for testing RAM - excellent!
5). DOS freeware F-Prot and recent virus definitions
6). Norton's DOS utilities
7). Various HD setup utilities (eg: Western Digital, Seagate boot floppies)
8). Freesco Linux router/webserver on a floppy
9). Sample linux config files (eg: XFConfig-4, fstab, etc)
10). Frozen-Bubble bootable CD for times of stress

http://www.memtest86.com/ It's helped me any number of times when I was beating my head against the wall over a weird problem. It's just a diagnostic tool, though.

AMD better forget these little incremental speed bumps and switch to a whole new architecture this year if they want to remain competetive.

It's called x86-64. The Opteron ships next month.

The current architecture is like milking a deadhorse and they are already running waay too hot.

I did not need that mental image...

Current Thoroughbred and Barton core Athlons don't run all that hot. An Athlon 3000+ runs cooler than a 3GHz P4.

I reclocked my TBred core Athlon XP 1700+ to 8x202MHz (404MHz DDR) on my ASUS A7N8X Deluxe motherboard (Corsair PC3200C2 DIMM). I kept the default core voltage (1.5v). MemTest86 verified that it works reliably. Upping the FSB is mostly a matter of motherboard and memory support, not CPU support (outside of being able to adjust the clock multiplier). A few years ago I reclocked a 150MHz Pentium to 1.5x100MHz. Worked just fine.

check out memtest86. I have used it to check memory when doing troubleshooting on problem machines.

One option will generate the memory specs you can use with Linux to tell the kernel which memory spots to avoid. Although the idea of avoiding problem areas of known 'bad' memory sounds nice, I wouldn't use it in my machines.

BadRAM is a Linux kernel patch for defective RAM modules support. With defective RAM, I mean RAM which has some bits wrong at some (known) addresses.

Normally, such RAM is considered useless and thrown away; the larger RAMs get, the higher the chances of failing addresses. With ever growing RAM sizes, it would therefore be pleasant to have an alternative to discarding of defective RAM chips.

By the way, memtest86 is a free x86 memory diagnostic, which can be configured to produce BadRAM patterns

I had this on an old installation of Win98 (or maybe 95)... did a clean install and it happened again. In the end it turned out I had a dodgy SIMM... replaced that and it worked fine.

I'd suggest finding, and running, a memory tester. Try memtest86.

You should check out MemTest86 if you are having any ram trouble. It detects pretty much EVERY ram problem, including intermittent failures.

If bad ram is preventing you from booting, well, then it should be pretty obvious which stick is bad.

• use windbg.exe - it's the gui debugger
• type the ".symfix" command in to get debugging symbols
• After typing ".symfix" type ".reload" to reload the symbols.
• Type !analyze to have the debugger automatically figure out what went wrong and who to blame (if it can).

Also, memtest86.

Why? AFAIK BIOS is in ROM and memory is really cheap these days.

What I'd love to see in BIOS is a good disk partitioning tool and a memtest86 or something like it.

MemTest86
This is an memory tester that boots from a floppy. I use it frequently, and it works very well - a must-have for a testing toolkit.

Try memtest86 for diagnosing memory problems.

This is most likely a hardware failure, possibly memory. Try memtest86 before you go on a kernel debugging hunt... basically, if your server has worked great for 12 months and then craps like this it probably ain't software.

Damn. I hope these ones won't be as buggy as the 810/815. I've tried a few ones, and all I got was random halts on my penguin OS....

BTW I would like to know if any of you ran memtest86 on i810/815 boards... I did on several ones and I got some pretty nice errors... so I fall back to my oldie PII. At least this one works 7/24.

Álvaro Lopes

It no longer exists at http://Reality.SGI.com/cbrady_denver/memtest86/

Now it lives at www.MemTest86.com

Thanks Chris,
Ken Hendrickson