Prime95 version 27.6 - Page 3

Prime95 · 2012-04-27, 15:14

Quote:

Originally Posted by BrianOC

Hi. My problem:

Bios reset stock all.
i7 2600, 8GB RAM, GTX 480

This bug?

Looks like the 3rd core of your quad-core is a bit flaky.

BrianOC · 2012-04-27, 15:37

Quote:

Originally Posted by LaurV

No bug.
Use the last version 27.6
Use only 4 workers (i7-2600 has only 4 physical cores, the rest are logical - hyperthreaded - cores, using 8 cores produce additional heat, without additional performance, well, somehow).
Also, the versions with AVX will stress that particular CPU more then the versions with SSE (AVX will use the CPU more intensely). I have 2600k and I can run 4 workers, each with helper HT on v26.6 at 4.5GHz, but when I run v27.6 I have either to reduce the clock or either kill the helpers, otherwise I get blue screens. Of course, 27.6 gets better performance, despite of reduced clock and missing helper threads.

Ok tested 4 worker.

Quote:

Originally Posted by Stef42

The core voltage seems very low, it should be more like 1.2+ instead of 1.1V
Even so, because you're running stock it should be working normally. Have you tried other stress programs?

Yes i use linx(AVX) max memory 2 hour no errors and prime95 27.4 6 hour Blend and FTT. Stock clock and voltage max 1.92v.1.2.
Error in stock :S 0 Minutes 27.4 work 6 hour? How?

flashjh · 2012-04-27, 15:37

Quote:

Originally Posted by BrianOC

Hi. My problem:

27.6: http://i50.tinypic.com/345fp1j.png

27.4:http://i45.tinypic.com/2ylps9s.png

Bios reset stock all.
i7 2600, 8GB RAM, GTX 480

This bug? Sorry my english bad. Thanks!

Brian, I have to agree with the hardware problem. As has been discussed in the forum many times, when P95 detects a problem, it's *most likely* a hardware issue. I would run the stress tests until you figure out the issue. If you're OCing, that would be the first thing to eliminate, then work with the hardware.

Dubslow · 2012-04-27, 16:11

Guys, what's he's saying is that he gets errors with 27.6, but NOT with 27.4. In theory, if he gets errors with one version, he should get errors with all versions.

Edit: More specifically, I think he said he ran 27.4 for 6 hours, all three torture test styles, no errors, but 27.6 fails in 5 minutes. There's a bug with one or the other. BrianOC, which test did you run that fails on 27.6? I also have a 2600K which I'm absolutely sure is stable and try and reproduce your problem.

BrianOC · 2012-04-27, 16:36

Quote:

Originally Posted by Dubslow

Guys, what's he's saying is that he gets errors with 27.6, but NOT with 27.4. In theory, if he gets errors with one version, he should get errors with all versions.

Edit: More specifically, I think he said he ran 27.4 for 6 hours, all three torture test styles, no errors, but 27.6 fails in 5 minutes. There's a bug with one or the other. BrianOC, which test did you run that fails on 27.6? I also have a 2600K which I'm absolutely sure is stable and try and reproduce your problem.

Yes 27.4 running 6hour all test. Linx(AVX) 1.30h maximum problem size.

Downclock my RAM and CPU. CPU 3GHZ 1.25v RAM 800MHz
and: http://i45.tinypic.com/2jax8nr.png

Error

PC: i7 2600 non K
ASUS P8P67 BIOS 2303 (update 3207 now)
8GB Kingston memory 1333MHz CL9
MSI GTX 480
XFX CORE EDITION PRO 450W power supply
3 HDD RAID0

Edit: 4 worker stopped. HT problem?

Dubslow · 2012-04-27, 16:47

What's "Linx"? And my question was, which torture test are you using: Blend, In-place, or the other one? (Or does it fail with all three?)

To humor us, what are the temperatures you see when running the torture test? I'd recommend HWMonitor, it's by the same people as CPUZ.

Edit: Definitely a bug somewhere. Again, I'm absolutely sure of stability: I've turned in nothing but good DCs for the last 4 months, and temps run less than 65C, typically <60C.

Code:

Your choice: 15

Number of torture test threads to run (4): 3
Choose a type of torture test to run.
1 = Small FFTs (maximum FPU stress, data fits in L2 cache, RAM not tested 
much).
2 = In-place large FFTs (maximum heat and power consumption, some RAM 
tested).
3 = Blend (tests some of everything, lots of RAM tested).
11,12,13 = Allows you to fine tune the above three selections.
Blend is the default.  NOTE: if you fail the blend test, but can pass the 
small FFT test then your problem is likely bad memory or a bad memory 
controller.
Type of torture test to run (3): 1

Accept the answers above? (Y): y
[Main thread Apr 27 11:51:51] Starting workers.
[Worker #2 Apr 27 11:51:51] Worker starting
[Worker #1 Apr 27 11:51:51] Worker starting
[Worker #1 Apr 27 11:51:51] Setting affinity to run worker on any logical CPU.
[Worker #3 Apr 27 11:51:51] Worker starting
[Worker #3 Apr 27 11:51:51] Setting affinity to run worker on any logical CPU.
[Worker #1 Apr 27 11:51:51] Beginning a continuous self-test to check your computer.
[Worker #1 Apr 27 11:51:51] Please read stress.txt.  Hit ^C to end this test.
[Worker #3 Apr 27 11:51:51] Beginning a continuous self-test to check your computer.
[Worker #3 Apr 27 11:51:51] Please read stress.txt.  Hit ^C to end this test.
[Worker #2 Apr 27 11:51:51] Setting affinity to run worker on any logical CPU.
[Worker #2 Apr 27 11:51:51] Beginning a continuous self-test to check your computer.
[Worker #2 Apr 27 11:51:51] Please read stress.txt.  Hit ^C to end this test.
[Worker #3 Apr 27 11:51:51] Test 1, 180000 Lucas-Lehmer iterations of M580673 using AVX FFT length 28K, Pass1=448, Pass2=64.
[Worker #2 Apr 27 11:51:51] Test 1, 180000 Lucas-Lehmer iterations of M580673 using AVX FFT length 28K, Pass1=448, Pass2=64.
[Worker #1 Apr 27 11:51:51] Test 1, 180000 Lucas-Lehmer iterations of M580673 using AVX FFT length 28K, Pass1=448, Pass2=64.
[Worker #2 Apr 27 11:52:14] Test 2, 180000 Lucas-Lehmer iterations of M573441 using AVX FFT length 28K, Pass1=448, Pass2=64.
[Worker #3 Apr 27 11:52:14] Test 2, 180000 Lucas-Lehmer iterations of M573441 using AVX FFT length 28K, Pass1=448, Pass2=64.
[Worker #1 Apr 27 11:52:14] Test 2, 180000 Lucas-Lehmer iterations of M573441 using AVX FFT length 28K, Pass1=448, Pass2=64.
[Worker #1 Apr 27 11:52:14] FATAL ERROR: Rounding was 38103.86312, expected less than 0.4
[Worker #1 Apr 27 11:52:14] Hardware failure detected, consult stress.txt file.
[Worker #1 Apr 27 11:52:14] Torture Test completed 1 tests in 0 minutes - 1 errors, 0 warnings.
[Worker #1 Apr 27 11:52:14] Worker stopped.
[Worker #2 Apr 27 11:52:36] Test 3, 180000 Lucas-Lehmer iterations of M565247 using AVX FFT length 28K, Pass1=448, Pass2=64.
[Worker #3 Apr 27 11:52:36] Test 3, 180000 Lucas-Lehmer iterations of M565247 using AVX FFT length 28K, Pass1=448, Pass2=64.
[Worker #2 Apr 27 11:53:00] Test 4, 180000 Lucas-Lehmer iterations of M557057 using AVX FFT length 28K, Pass1=448, Pass2=64.
[Worker #3 Apr 27 11:53:01] Test 4, 180000 Lucas-Lehmer iterations of M557057 using AVX FFT length 28K, Pass1=448, Pass2=64.
[Worker #2 Apr 27 11:53:22] Test 5, 180000 Lucas-Lehmer iterations of M544767 using AVX FFT length 28K, Pass1=448, Pass2=64.
[Worker #3 Apr 27 11:53:23] Test 5, 180000 Lucas-Lehmer iterations of M544767 using AVX FFT length 28K, Pass1=448, Pass2=64.

(Using Linux 27.6, HT OFF, 3 workers (4th is occupied by YAFU/NFS). I'll try and test 27.4. (Blend did not produce any immediate errors, though I let it run for 1 minute instead of 5. I tried small because his screenshot indicated a 12K length; I got 28K.)

Edit2: Notice that in both my case and Brian's, the roundoff was not even less than 1, rather it was E3, E4, E75. How could that possibly be a valid roundoff?

BrianOC · 2012-04-27, 17:19

Quote:

Originally Posted by Dubslow

What's "Linx"? And my question was, which torture test are you using: Blend, In-place, or the other one? (Or does it fail with all three?)

To humor us, what are the temperatures you see when running the torture test? I'd recommend HWMonitor, it's by the same people as CPUZ.

Edit: Definitely a bug somewhere. Again, I'm absolutely sure of stability: I've turned in nothing but good DCs for the last 4 months, and temps run less than 65C, typically <60C.

Code:

Your choice: 15

Number of torture test threads to run (4): 3
Choose a type of torture test to run.
1 = Small FFTs (maximum FPU stress, data fits in L2 cache, RAM not tested 
much).
2 = In-place large FFTs (maximum heat and power consumption, some RAM 
tested).
3 = Blend (tests some of everything, lots of RAM tested).
11,12,13 = Allows you to fine tune the above three selections.
Blend is the default.  NOTE: if you fail the blend test, but can pass the 
small FFT test then your problem is likely bad memory or a bad memory 
controller.
Type of torture test to run (3): 1

Accept the answers above? (Y): y
[Main thread Apr 27 11:51:51] Starting workers.
[Worker #2 Apr 27 11:51:51] Worker starting
[Worker #1 Apr 27 11:51:51] Worker starting
[Worker #1 Apr 27 11:51:51] Setting affinity to run worker on any logical CPU.
[Worker #3 Apr 27 11:51:51] Worker starting
[Worker #3 Apr 27 11:51:51] Setting affinity to run worker on any logical CPU.
[Worker #1 Apr 27 11:51:51] Beginning a continuous self-test to check your computer.
[Worker #1 Apr 27 11:51:51] Please read stress.txt.  Hit ^C to end this test.
[Worker #3 Apr 27 11:51:51] Beginning a continuous self-test to check your computer.
[Worker #3 Apr 27 11:51:51] Please read stress.txt.  Hit ^C to end this test.
[Worker #2 Apr 27 11:51:51] Setting affinity to run worker on any logical CPU.
[Worker #2 Apr 27 11:51:51] Beginning a continuous self-test to check your computer.
[Worker #2 Apr 27 11:51:51] Please read stress.txt.  Hit ^C to end this test.
[Worker #3 Apr 27 11:51:51] Test 1, 180000 Lucas-Lehmer iterations of M580673 using AVX FFT length 28K, Pass1=448, Pass2=64.
[Worker #2 Apr 27 11:51:51] Test 1, 180000 Lucas-Lehmer iterations of M580673 using AVX FFT length 28K, Pass1=448, Pass2=64.
[Worker #1 Apr 27 11:51:51] Test 1, 180000 Lucas-Lehmer iterations of M580673 using AVX FFT length 28K, Pass1=448, Pass2=64.
[Worker #2 Apr 27 11:52:14] Test 2, 180000 Lucas-Lehmer iterations of M573441 using AVX FFT length 28K, Pass1=448, Pass2=64.
[Worker #3 Apr 27 11:52:14] Test 2, 180000 Lucas-Lehmer iterations of M573441 using AVX FFT length 28K, Pass1=448, Pass2=64.
[Worker #1 Apr 27 11:52:14] Test 2, 180000 Lucas-Lehmer iterations of M573441 using AVX FFT length 28K, Pass1=448, Pass2=64.
[Worker #1 Apr 27 11:52:14] FATAL ERROR: Rounding was 38103.86312, expected less than 0.4
[Worker #1 Apr 27 11:52:14] Hardware failure detected, consult stress.txt file.
[Worker #1 Apr 27 11:52:14] Torture Test completed 1 tests in 0 minutes - 1 errors, 0 warnings.
[Worker #1 Apr 27 11:52:14] Worker stopped.
[Worker #2 Apr 27 11:52:36] Test 3, 180000 Lucas-Lehmer iterations of M565247 using AVX FFT length 28K, Pass1=448, Pass2=64.
[Worker #3 Apr 27 11:52:36] Test 3, 180000 Lucas-Lehmer iterations of M565247 using AVX FFT length 28K, Pass1=448, Pass2=64.
[Worker #2 Apr 27 11:53:00] Test 4, 180000 Lucas-Lehmer iterations of M557057 using AVX FFT length 28K, Pass1=448, Pass2=64.
[Worker #3 Apr 27 11:53:01] Test 4, 180000 Lucas-Lehmer iterations of M557057 using AVX FFT length 28K, Pass1=448, Pass2=64.
[Worker #2 Apr 27 11:53:22] Test 5, 180000 Lucas-Lehmer iterations of M544767 using AVX FFT length 28K, Pass1=448, Pass2=64.
[Worker #3 Apr 27 11:53:23] Test 5, 180000 Lucas-Lehmer iterations of M544767 using AVX FFT length 28K, Pass1=448, Pass2=64.

(Using Linux 27.6, HT OFF, 3 workers (4th is occupied by YAFU/NFS). I'll try and test 27.4. (Blend did not produce any immediate errors, though I let it run for 1 minute instead of 5. I tried small because his screenshot indicated a 12K length; I got 28K.)

Edit2: Notice that in both my case and Brian's, the roundoff was not even less than 1, rather it was E3, E4, E75. How could that possibly be a valid roundoff?

linx stress test program.
27.6 Smal FTT error
blend and large fft no error http://i47.tinypic.com/34gt5di.png

Dubslow · 2012-04-27, 17:29

Okay, after about half an hour of Small FFT testing, that's the only error I got; the other two workers have passed 28K, 8K, and are halfway through 35K.

Here's the LinX: A GUI for Linpack. http://www.xtremesystems.org/forums/...pack-interface

I'll start running a blend (same settings). Brian, which FFT size failed on the Blend?

Edit: The Blend 448K and 8K passed, no errors about halfway through 512K. I'll revert to small FFTs and see what else fails.
Edit2: Heh, this time Worker 2 failed right off the bat, fast enough that it didn't appear on my CPU monitor:

Code:

Your choice: 15

Number of torture test threads to run (4): 3
Choose a type of torture test to run.
1 = Small FFTs (maximum FPU stress, data fits in L2 cache, RAM not tested 
much).
2 = In-place large FFTs (maximum heat and power consumption, some RAM 
tested).
3 = Blend (tests some of everything, lots of RAM tested).
11,12,13 = Allows you to fine tune the above three selections.
Blend is the default.  NOTE: if you fail the blend test, but can pass the 
small FFT test then your problem is likely bad memory or a bad memory 
controller.
Type of torture test to run (3): 1

Accept the answers above? (Y): y
[Main thread Apr 27 13:08:46] Starting workers.
[Comm thread Apr 27 13:08:46] Updating computer information on the server
[Worker #2 Apr 27 13:08:46] Worker starting
[Worker #2 Apr 27 13:08:46] Setting affinity to run worker on any logical CPU.
[Worker #3 Apr 27 13:08:46] Worker starting
[Worker #3 Apr 27 13:08:46] Setting affinity to run worker on any logical CPU.
[Worker #2 Apr 27 13:08:46] Beginning a continuous self-test to check your computer.
[Worker #2 Apr 27 13:08:46] Please read stress.txt.  Hit ^C to end this test.
[Worker #3 Apr 27 13:08:46] Beginning a continuous self-test to check your computer.
[Worker #3 Apr 27 13:08:46] Please read stress.txt.  Hit ^C to end this test.
[Worker #1 Apr 27 13:08:46] Worker starting
[Worker #1 Apr 27 13:08:46] Setting affinity to run worker on any logical CPU.
[Worker #1 Apr 27 13:08:46] Beginning a continuous self-test to check your computer.
[Worker #1 Apr 27 13:08:46] Please read stress.txt.  Hit ^C to end this test.
[Worker #3 Apr 27 13:08:46] Test 1, 180000 Lucas-Lehmer iterations of M580673 using AVX FFT length 28K, Pass1=448, Pass2=64.
[Worker #1 Apr 27 13:08:46] Test 1, 180000 Lucas-Lehmer iterations of M580673 using AVX FFT length 28K, Pass1=448, Pass2=64.
[Worker #2 Apr 27 13:08:46] Test 1, 180000 Lucas-Lehmer iterations of M580673 using AVX FFT length 28K, Pass1=448, Pass2=64.
[Worker #2 Apr 27 13:08:46] FATAL ERROR: Rounding was 2.213541735e+20, expected less than 0.4
[Worker #2 Apr 27 13:08:46] Hardware failure detected, consult stress.txt file.
[Worker #2 Apr 27 13:08:46] Torture Test completed 0 tests in 0 minutes - 1 errors, 0 warnings.
[Worker #2 Apr 27 13:08:46] Worker stopped.

Whoops, Worker 1 also failed within 15 seconds

Code:

[Worker #1 Apr 27 13:09:09] Test 2, 180000 Lucas-Lehmer iterations of M573441 using AVX FFT length 28K, Pass1=448, Pass2=64.
[Worker #1 Apr 27 13:09:09] FATAL ERROR: Rounding was 2.213541735e+20, expected less than 0.4
[Worker #1 Apr 27 13:09:09] Hardware failure detected, consult stress.txt file.
[Worker #1 Apr 27 13:09:09] Torture Test completed 1 tests in 0 minutes - 1 errors, 0 warnings.
[Worker #1 Apr 27 13:09:09] Worker stopped.
[Worker #3 Apr 27 13:09:09] Test 2, 180000 Lucas-Lehmer iterations of M573441 using AVX FFT length 28K, Pass1=448, Pass2=64.
[Worker #3 Apr 27 13:09:30] Test 3, 180000 Lucas-Lehmer iterations of M565247 using AVX FFT length 28K, Pass1=448, Pass2=64.

Note the rounding errors of E20 OoM.

BrianOC · 2012-04-27, 17:35

Quote:

Originally Posted by Dubslow

Okay, after about half an hour of Small FFT testing, that's the only error I got; the other two workers have passed 28K, 8K, and are halfway through 35K.

Here's the LinX: A GUI for Linpack. http://www.xtremesystems.org/forums/...pack-interface

I'll start running a blend (same settings). Brian, which FFT size failed on the Blend?

Sorry! Blend worked now tested 10 minutes, Small FTT error 0-1 minutes.

Dubslow · 2012-04-27, 17:36

Quote:

Originally Posted by BrianOC

Sorry! Blend worked now tested 10 minutes, Small FTT error 0-1 minutes.

Ah, likely the Blend was running through the "Small" portion of the blend by chance and then failed.

Dubslow · 2012-04-27, 19:17

Okay, with 27.4, all three workers passed the 28K test, which is the only size that failed with 27.6. They're mostly through 8K as well, though I don't think that'll fail either; I'll go back to production now until George/others request otherwise.

2012-04-27, 16:11	#26
Dubslow Basketry That Evening! "Bunslow the Bold" Jun 2011 40<A<43 -89<O<-88 1110000110101₂ Posts	Guys, what's he's saying is that he gets errors with 27.6, but NOT with 27.4. In theory, if he gets errors with one version, he should get errors with all versions. Edit: More specifically, I think he said he ran 27.4 for 6 hours, all three torture test styles, no errors, but 27.6 fails in 5 minutes. There's a bug with one or the other. BrianOC, which test did you run that fails on 27.6? I also have a 2600K which I'm absolutely sure is stable and try and reproduce your problem. Last fiddled with by Dubslow on 2012-04-27 at 16:16

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Prime95 version 27.3	Prime95	Software	148	2012-03-18 19:24
Prime95 version 26.3	Prime95	Software	76	2010-12-11 00:11
Prime95 version 25.5	Prime95	PrimeNet	369	2008-02-26 05:21
Prime95 version 25.4	Prime95	PrimeNet	143	2007-09-24 21:01
When the next prime95 version ?	pacionet	Software	74	2006-12-07 20:30

2012-04-27, 19:17	#33
Dubslow Basketry That Evening! "Bunslow the Bold" Jun 2011 40<A<43 -89<O<-88 3·29·83 Posts	Okay, with 27.4, all three workers passed the 28K test, which is the only size that failed with 27.6. They're mostly through 8K as well, though I don't think that'll fail either; I'll go back to production now until George/others request otherwise.