Help Interpreting Prime95 and Memtest86+ Tests

[The J]

The Problem Within the past two weeks, I've begun to have problems with crashing in my games: FarCry, Doom3, UT2004, and The Elder Scrolls 3: Morrowind. The game will crash to the desktop in the middle of it at seemingly random intervals. It does not happen all the time, but happens quite often (twice today with FarCry in fact). Other times, the game will just freeze for about 10 seconds and then continue on as though nothing has happened--no hard drive activity or anything, just freezing. I've played FarCry the most lately so I've obviously noticed it there the most often; however, it does occur in my other games as well. Like I said above, the problem started within the last 2 weeks.

The Specs My computer specs are in my sig. I have 2 sticks of 512MB PC2700 RAM modules. I also have a Powman "450W" PSU (maybe that's the problem??). Both sticks are by Apacer and both have Infineon chips and the same SPD data according to Everest and timings of 2.5-3-3-7.

And Now, the Tests...
Event Log

Whenever the game crashes to the desktop, I get two of the following errors in the Event Viewer:

Event Type: Error
Event Source:
Event Category: None
Event ID: 0
Date: 10/22/2004
Time: 9:56:10 PM
User: N/A
Computer: J
Description:
The description for Event ID ( 0 ) in Source ( ) cannot be found. The local computer may not have the necessary registry information or message DLL files to display messages from a remote computer. You may be able to use the /AUXSOURCE= flag to retrieve this description; see Help and Support for details. The following information is part of the event: 7.
Data:
0000: 71 67 88 72 bb b9 ca 75 qg?r»¹Êu
0008: 6e 6f 81 7a b3 b1 c2 7d no?z³±Â}
0010: 76 77 99 62 ab a9 da 65 vw?b«©Úe
0018: 7e 7f 91 6a a3 a1 d2 6d ~?j£¡Òm
0020: 46 47 a9 52 9b 99 ea 55 FG©R??êU
0028: 4e 4f a1 5a 93 91 e2 5d NO¡Z??â]
0030: 56 57 b9 42 8b 89 fa 45 VW¹B??úE
0038: 5e 5f b1 4a 83 81 f2 4d ^_±J??òM
0040: 26 27 c9 32 fb f9 8a 35 &'É2ûù?5
0048: 2e 2f c1 3a f3 f1 82 3d ./Á:óñ?=
0050: 36 37 d9 22 eb e9 9a 25 67Ù"ëé?%
0058: 3e 3f d1 2a e3 e1 92 2d >?Ñ*ãá?-
0060: 06 07 e9 12 db d9 aa 15 ..é.ÛÙª.
0068: 0e 0f e1 1a d3 d1 a2 1d ..á.ÓÑ¢.
0070: 16 17 f9 02 cb c9 ba 05 ..ù.Ëɺ.
0078: 1e 1f f1 0a c3 c1 b2 0d ..ñ.ÃÁ².
0080: e6 e7 09 f2 3b 39 4a f5 æç.ò;9Jõ
0088: ee ef 01 fa 33 31 42 fd îï.ú31Bý
0090: cd f7 19 e2 08 29 5a e5 Í÷.â.)Zå
0098: dd ff 11 ea 22 21 52 ed Ýÿ.ê"!Rí
00a0: 39 38 d6 2d db bf 6b e0 98Ö-Û¿kà
00a8: 56 62 3f e2 df dd ae 11 Vb?âßÝ®.
00b0: d7 d7 39 c2 ab f1 76 c4 ××9«ñvÄ
00b8: cc c7 35 f2 18 01 72 cd ÌÇ5ò..rÍ
00c0: 24 a5 49 b2 13 94 18 b5 $¥I².?.µ
00c8: 8d af 41 ba 71 a5 b9 4b ?¯Aºq¥¹K
00d0: b6 4a 3d c4 69 bd a1 53 ¶J=Äi½¡S
00d8: be 42 35 cc 61 b5 a9 5b ¾B5Ìaµ©[
00e0: 86 7a 0d f4 59 8d 91 63 ?z.ôY??c
00e8: 8e 72 05 fc 51 85 99 6b ?r.üQ??k
00f0: 96 6a 1d e4 49 9d 81 73 ?j.äI??s
00f8: 9e 62 15 ec 41 95 89 7b ?b.ìA??{
0100: 66 9a ed 14 b9 6d 71 83 f?í.¹mq?
0108: 6e 92 e5 1c b1 65 79 8b n?å.±ey?
0110: 76 8a fd 04 a9 7d 61 93 v?ý.©}a?
0118: 7e 82 f5 0c a1 75 69 9b ~?õ.¡ui?
0120: 46 ba cd 34 99 4d 51 a3 FºÍ4?MQ£
0128: 4e b2 c5 3c 91 45 59 ab N²Å<?EY«
0130: 56 aa dd 24 89 5d 41 b3 VªÝ$?]A³
0138: 5e a2 d5 2c 81 55 49 bb ^¢Õ,?UI»
0140: 26 da ad 54 f9 2d 31 c3 &amp;Ú­Tù-1Ã
0148: 2e d2 a5 5c f1 25 39 cb .Ò¥\ñ%9Ë
0150: 36 ca bd 44 e9 3d 21 d3 6ʽDé=!Ó
0158: 3e c2 b5 4c e1 35 29 db >µLá5)Û
0160: 06 fa 8d 74 d9 0d 11 e3 .ú?tÙ..ã
0168: 0e f2 85 7c d1 05 19 eb .ò?|Ñ..ë
0170: 16 ea 9d 64 c9 1d 01 f3 .ê?dÉ..ó
0178: 1e e2 95 6c c1 15 09 fb .â?lÁ..û
0180: e6 1a 6d 94 39 ed f1 03 æ.m?9íñ.
0188: ee 12 65 9c 31 e5 f9 0b î.e?1åù.
0190: f6 0a 7d 84 29 fd e1 13 ö.}?)ýá.
0198: fe 02 75 8c 21 f5 e9 1b þ.u?!õé.
01a0: c6 3a 4d b4 19 cd d1 23 Æ:M´.ÍÑ#
01a8: ce 32 45 bc 11 c5 d9 2b Î2E¼.ÅÙ+
01b0: d6 2a 5d a4 09 dd c1 33 Ö*]¤.ÝÁ3
01b8: de 22 55 ac 01 d5 c9 3b Þ"U¬.ÕÉ;
01c0: a6 5a 2d d4 79 ad b1 43 ¦Z-Ôy­±C
01c8: ae 52 25 dc 71 a5 b9 4b ®R%Üq¥¹K
01d0: b6 4a 3d c4 69 bd a1 53 ¶J=Äi½¡S
01d8: be 42 35 cc 61 b5 a9 5b ¾B5Ìaµ©[
01e0: 86 7a 0d f4 59 8d 91 63 ?z.ôY??c
01e8: 8e 72 05 fc 51 85 99 6b ?r.üQ??k
01f0: 96 6a 1d e4 49 9d 81 73 ?j.äI??s
01f8: 9e 62 15 ec 41 95 89 7b ?b.ìA??{
0200: 66 9a ed 14 b9 6d 71 83 f?í.¹mq?
0208: 6e 92 e5 1c b1 65 79 8b n?å.±ey?
0210: 76 8a fd 04 a9 7d 61 93 v?ý.©}a?
0218: 7e 82 f5 0c a1 75 69 9b ~?õ.¡ui?
0220: 46 ba cd 34 99 4d 51 a3 FºÍ4?MQ£
0228: ee da b4 73 4c eb eb af îÚ´sLëë¯
0230: 54 83 02 b4 8b 74 9e 23 T?.´?t?#
0238: 5c 8b 0a bc 83 7c 96 2b \?.¼?|?+
0240: 24 f3 72 c4 fb 04 ee 53 $órÄû.îS
0248: 2c fb 7a cc f3 0c e6 5b ,ûzÌó.æ[
0250: 34 e3 62 d4 eb 14 fe 43 4ãbÔë.þC
0258: 3c eb 6a dc e3 1c f6 4b <ëjÜã.öK
0260: 04 d3 52 e4 db 24 ce 73 .ÓRäÛ$Îs
0268: 0c db 5a ec d3 2c c6 7b .ÛZìÓ,Æ{
0270: 14 c3 42 f4 cb 34 de 63 .ÃBôË4Þc
0278: 1c cb 4a fc c3 3c d6 6b .ËJüÃ<Ök
0280: e4 33 b2 04 3b c4 2e 93 ä3².;Ä.?
0288: ec 3b ba 0c 33 cc 26 9b ì;º.3Ì&amp;?
0290: f4 23 a2 14 2b d4 3e 83 ô#¢.+Ô>?
0298: fc 2b aa 1c 23 dc 36 8b ü+ª.#Ü6?
02a0: c4 13 92 24 1b e4 0e b3 Ä.?$.ä.³
02a8: cc 1b 9a 2c 13 ec 06 bb Ì.?,.ì.»
02b0: d4 03 82 34 0b f4 1e a3 Ô.?4.ô.£
02b8: dc 0b 8a 3c 03 fc 16 ab Ü.?<.ü.«
02c0: a4 73 f2 44 7b 84 6e d3 ¤sòD{?nÓ
02c8: ac 7b fa 4c ¬{úL


Because of the length of this post and the amount of testing done, I will continue with another post.

Specs are below.

 

[The J]

...continuing from the first post:

Prime 95

My first thought was to run Prime95 since something quite obviously isn't stable. With both sticks of RAM in the motherboard, the test will either fail within the first five minutes or last for about 4 hours, after which it'll still fail. I removed one of the RAM modules and ran the test again. It lasted over six hours without failing. I removed that module and put in the other one and got the same thing. Basically the test will fail if both sticks are in, but seem to run fine if either one of the sticks is removed. I've also tried moving the RAM to different slots in the motherboard (three slots in total). The same results as above occurred. I tried to relax the timings in my BIOS to 3-3-3-7 (the loosest it'll allow) and it still fails with both sticks in and works fine with only one in.

I should note here that FarCry, Doom3, Aquamark 3, and 3DMark2001SE work fine with the one module. The benchmarks have never crashed, however, even with both modules in. I ran 3DMark2001 as that seems to be more CPU intensive than the other 3DMarks and so should crash if the CPU had problems, but it didn't.

Memtest86 and 86+

I ran memtest86 at first, but found that 86+ was newer; however, it didn't seem to matter as both returned the same results. I let the tests run through one pass. The standard tests would not find any problems with the RAM modules, so I ran the full tests (14 of them). With both modules, memtest86 (either version) would report about 110,000 errors. No that's not a typo: one-hundred ten thousand errors!! I tried moving the modules to different slot locations and would get the same result. I relaxed the timings as I had with Prime95 with both modules in and got 500,000 errors. Once again, I don't get any errors with only one module in. It doesn't matter which module it is or even what slot it is in. This coincides with what I got in Prime95.

The errors seemed to start in Test 9 [Moving inv, 8 bit pat, no cache]. According to what was shown on the screen (I could only see 12 errors onscreen and I couldn't get them to scroll), the RAM would lose a digit in the string of data for all of the errors onscreen. For example, if the pattern memtest86 was using is 40404040, it might return with 40004040--leaving out a 4. There were many errors I could not see, however, so obviously not all of them are necessarily the same problem.


*With all of the tests above, I tried to record everything I've tried. Unfortunately, I may have missed some things. Ask me and I can tell you if I tried it, but forgot to post it.*

Steps Taken to Try to Fix the Problems

Now I certainly wouldn't go through all the trouble of typing this out without trying to rectify the situation. My first step was to check for viruses, spyware, and any other malicious processes. I used McAfee Virus Scan 2003, Lavasoft Ad-Aware SE, and Hijack This! to look for these things. McAfee and Ad-Aware found nothing and my Hijack This! log is clean. When I originally had built this computer last September, I had only one module and it gave me problems similar to above, but not as frequent.

I updated my BIOS and all was well. I got a new BIOS from Shuttle's website and flashed it using the recommended flashing utility: Award WinFlash, which did it from Windows XP. I got the newest BIOS available (I was a few versions behind), flashed it, and get the same problems still. However, now at the device screen (the one right before Windows begins to load), it says "Invalid Flash Type" even though the BIOS flashed fine or at least seemed to.

I checked my PSU (Powmax 450W) voltages in Everest as I do not have nor know how to use a mulitmeter. I get the following:
CPU core: 1.62V - 1.65V (it varies)
+3.3V: 3.28V
+5V: 5.03V - 5.05V
+12V: 12.06V - 12.40V
+5VSB: 4.78V
VBAT: 3.22V
AGP: 1.49V

I also check my CPU and system temp in Everest. They are never higher than 46C and 38C at any given time. My 6800GT will vary from 55C to about 61C. The side of my case (also Powmax; feel free to laugh) is open now and was during most of the tests.

I am running the latest drivers for all my components, except for my 6800GT, which is using 66.32 beta drivers. They were there several weeks before my games crashing so I don't think that's it. I have all the Windows Updates as well. Service Pack 2 is installed and running fine.

Questions and Conclusion

After seeing this mountain of data, it's not hard to see why I need help. I really don't know where to look in terms of replacing parts or software. I thought that the RAM was bad, but they work fine alone. They just don't like each other. Loosening the timings I thought would have stopped latency issues between the modules. A new BIOS did not help, though I can use a Sempron now. My case isn't very dusty and it doesn't overheat. What about my PSU? Motherboard? CPU? Any ideas are welcome. If you need any more info, then please let me know. This issue has been bugging me to no end, especially since I'm one of those people who wants his computer working correctly. I can't stand these intermittent errors. This is made worse in FarCry, with its checkpoint system, and having to do the same area over and over again because Jack Carver had a stroke and kicked me back out to my desktop. Once again, any ideas are greatly appreciated. I will also try to help anyone with more info on a specific area if I am able.

Thank you in advance,

The J

 

[The J]

Update:

I ran Prime95 in Safe Mode at the suggestion of my roommate. It failed in 1 minute.


I realize there is a lot of stuff to wade through, which is why I'm having a lot of difficulty figuring out exactly what's going on. If anyone wants to throw anything out there at all, go ahead.

 

[Navid]

Don't waste your time with prime95. You need to pass memtest overnight first. Only then, should you start running prime95.

A single stick of Ram is more stable than 2 if you activate dual-channel performance. Do you have dual-channel active?

You need to reduce RAM frequency or relax the RAM timing or increase the RAM voltage by a notch and repeat memtest.

After you pass memtest, you start prime95 and try to pass that overnight. Don't touch the memory settings anymore now. If it has stopped running the next day, you need to reduce the CPU frequency or improve its cooling or increase VCore by a notch.

After you pass memtest and prime95 both, if you have crashed to desktop, it means that your graphics card is getting too hot. Then, you will need to reduce graphics card overclock rate or improve its cooling.

 

[The J]

I have a VIA KT333CF chipset, which does not have Dual-Channel available.

I can reduce the RAM frequency to 133MHz, but I have an Athlon XP 2600+ Barton, which runs at 166MHz.

The RAM was not overclocked and was set to 166MHz in the BIOS at the time of these tests.

I cannot change the voltages in the BIOS, only the timings.

I relaxed the timings as much as I could (3-3-3-7), but ended up getting 500,000 errors in memtest after one pass of full (14) tests. That would be 400,000 more than if I had just left the BIOS at the "Auto by SPD" setting. I usually have it set to Auto. (This make the timings 2.5-3-3-7)

I cannot raise the CPU voltage either.

I have a Volcano 9 heatsink which, while not great, is adequate. My temps are never above 48C and are usually around 45C at load.

I have This Case w/ PSU and am using the included Powmax PSU. I've heard that Powmax has a terrible reputation when it comes to stability on these boards. Do you think that poses a problem?


I will try running memtest86 overnight as you suggested. I appreciate your responses and would try all of them if I could, but my BIOS will not let me change voltages and has very little adjustibility in the timings. Please look at my system specs in my sig (Compusaurus). That may give you more info and help you figure this out. Thank you for what you've done so far.

 

[eydolic]

It seems to obviously be RAM, but I doubt that since the sticks test fine seperately. So, my next step would be to replace the PowMan PSU, because they're not known for giving out clean, steady power. I've seen them out-fluctuate complete generics before. Maybe you should pull a PSU from somebody else's system before investing in a new one. Don't relax the Cas Latency, leave that at 2.5 for the most stability. The other numbers are a little easier to deal with. Set everything to 2.5-4-4-8, and see what happens.

 

[The J]

I will try another PSU if I can find one. My roommate has an Antec 400W he may let me use. I cannot set my timing to 2.5-4-4-8. My BIOS will not let me. All of the most relaxed settings will give me only 3-3-3-7. I don't have a very flexible BIOS as you can tell. My RAM runs at 2.5-3-3-7, according to Everest, so I really can't change anything at all. Thanks for the replies.

EDIT: Something I did before, but forgot to add, was using MSconfig to set my RAM limit to 512MB, but keeping both sticks in. I ran Prime95 for 3 or 4 hours and got nothing. That may be helpful, though, since I didn't run Prime95 for a really long time.

 

[eydolic]

WTF kind of board are you using, first of all?

 

[The J]

Shuttle AK38/N <-- That's my motherboard.

Last night, I ran memtest86 with my drives (2 HDD, 2 optical) unplugged. I hoped that this would relive stress on my PSU when running the test. I did find a PSU I should be able to borrow for another round of testing. After letting memtest86 run overnight, I got over 4 million errors.

Below are my specs copied from my Anandtech profile:

CPU: AMD Athlon XP 2600+ Barton 1916MHz
Motherboard: Shuttle AK38/N
Memory: 1024 MB of Apacer PC2700
Video Card: eVGA GeForce 6800GT
Hard Drive: Western Digital WD800JB 80.0 GB @ 7200 RPMS
Additional Hard Drives: Samsung SP1213N 120GB 7200RPM 8MB buffer
Monitor: Dell P780 17" Trinitron flat
DVD Manufacturer &amp; Model: Lite-On JLMS XJ-HD166S 16xDVD/48xCD
CDR/CDRW Manufacturer &amp; Model: CyberDrive CW078D 40x16x48x
Storage Interface: IDE
Sound Card: Creative Sound Blaster Audigy 2 ZS
Speakers: Cambridge Megaworks 550 THX
Internet Connection: College Network T1
Operating System: Windows XP Home SP2

 

[The J]

Also, does anyone have a clue as to why I get the message "Invalid Flash Type" at boot? I've been getting that ever since I updated my BIOS to try to correct my problems. The computer doesn't run any different other than that message when booting up. Also, I still haven't been able to get ahold of another power supply to try it, but I should be able to during this week. I may just get a new case and PSU anyway just because I don't like my case (I've had it for 2 months, extremely flimsy and dented in several places when I got it) and Powmax scares me, with their terrible reputation and fluctuating 12V line (I don't know if the fluctuating is in ATX spec, though) and all.

 

[The J]

I have nothing interesting to say now. I'm just bumping this since I've been posting updates to this late at night. The mid-day crowd derserves a chance to take a stab at this pickle of a problem.

 

[IvanAndreevich]

No memory volts options? You can try LOWERING the timings, especially the CAS. That can actually sometimes improve stability.

Otherwise, you have no choice but to go to FS/T and trade you ram for another stick 😀

 

[The J]

I didn't think of lowering the timings because this isn't super-RAM or anything, just basic CL2.5 RAM made by Apacer. I'll try it anyway. So you think that the sticks are bad? They work fine alone, just not together. Thanks for the reply, maybe I can just find somone else with RAM rather than having to sell it.

 

[The J]

Okay, I ran for about 2 hours with the CAS Latency set to 2 (2-3-3-7) and it got to the middle of test 8. By then, I had over 700,000 errors as opposed to 400,000 by that time at normal settings. I didn't think letting it finish would do much good at this time just because there are already tons of errors so finishing would still tell me the same thing: something's still up.

 

[The J]

Okay, I took my roommate's 400W Antec PSU and used it overnight. When I took out my Powmax PSU I realized just how light it was. It feels as though it's a metal box with nothing inside of it. The Antec, however, felt *far* heavier than the Powmax does. If weight is a measure of quality, then I should get a new power supply anyway.

It got to the beginning of Test 11 and still had about 120,000 errors. So it must not be the PSU causing the problems. I still can't think of why, even though they are the same brand, speed, chipset, and type, they will not work correctly together but will run fine alone. Any more ideas before I let this finally die?