ERROR: SUM(INPUTS) != SUM(OUTPUTS)

AES · 2007-09-25, 14:30

This PC passes all torture tests and it just returned a matching double check. I was getting rounding errors on this one with version 25.4. I am getting the following with 25.5. I've restarted it now with 24.14.

Any ideas or suggestions?

Thanks.

Code:

[Tue Sep 25 03:34:03 2007]
Iteration: 625736/4010XXXX, ERROR: SUM(INPUTS) != SUM(OUTPUTS), 1887603325919654 != 1887603325922661
Possible hardware failure, consult the readme.txt file.
Continuing from last save file.
[Tue Sep 25 03:39:07 2007]
Error reading intermediate file: p1F00XXX
Renaming intermediate file q1F00XXX to p1F00XXX.
Iteration: 594731/4010XXXX, ERROR: SUM(INPUTS) != SUM(OUTPUTS), 4423346080195417 != 4423346080192381
Possible hardware failure, consult the readme.txt file.
Continuing from last save file.
[Tue Sep 25 03:44:20 2007]
Error reading intermediate file: p1F00XXX
Renaming intermediate file q1F00XXX to p1F00XXX.

Bundu · 2007-09-25, 20:59

Perhaps you should post this in the 25.5 sticky? So george sees it as a possible 25.5 bug?

AES · 2007-09-26, 03:13

Thanks. I've copied the entire directory to another system. I'll see if the errors are reproducible.

AES · 2007-09-26, 15:55

I'm not sure if it's a bug. The same setup runs fine on the other system (E4400 Conroe-2M).

It happens on the E2160 PC when two worker threads are running LL with FFTs 2560 and 2048.

If I use one worker thread using two cores, there are no errors. The Res64 matched on a double check as well.

I'm using the win32 version on XP. The CPU is an E2160 Conroe-1M, L1 32 KB per core, L2 Cache 1 MB (On-Die, ASC, Full-Speed).

AES · 2007-10-03, 13:49

This PC has problems with FFT size 2560 using v24.14.1 as well. It seems to be fine with FFT size <= 2048.

AES · 2007-10-04, 20:27

Bad RAM. It corrupted the LL when the FFT would not all fit in L2. It's strange that this RAM would pass the Windiag RAM tests.

This is the second new PNY DDR2 module that has been bad in the last month.

Bundu · 2007-10-05, 18:48

from what i know based on posts I've read. P95 is the ultimate stress tester and tends to point out hardware faults where other systems fail. that probably why the other diagnostic didn't tell you about the problem

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
LLR SUM(INPUTS) != SUM(OUTPUTS) error	jbristow	Software	4	2007-08-14 04:07
v. 2.13: SUM(INPUTS) != SUM(OUTPUTS)	Kaiw	Software	7	2005-10-26 14:49
SUM(INPUTS) != SUM(OUTPUTS)	rudi_m	Software	3	2005-08-22 15:42
ERROR: SUM(INPUTS) != SUM(OUTPUTS)	flava	Hardware	3	2004-01-19 17:52
ERROR: SUM(INPUTS) != SUM(OUTPUTS)	ebx	Software	5	2004-01-02 22:25

2007-09-25, 20:59	#2
Bundu Jul 2004 Mid Calder, Scotland 5·37 Posts	Perhaps you should post this in the 25.5 sticky? So george sees it as a possible 25.5 bug?

2007-09-26, 03:13	#3
AES Jul 2007 Tennessee 2⁵×19 Posts	Thanks. I've copied the entire directory to another system. I'll see if the errors are reproducible.

2007-09-26, 15:55	#4
AES Jul 2007 Tennessee 2⁵·19 Posts	I'm not sure if it's a bug. The same setup runs fine on the other system (E4400 Conroe-2M). It happens on the E2160 PC when two worker threads are running LL with FFTs 2560 and 2048. If I use one worker thread using two cores, there are no errors. The Res64 matched on a double check as well. I'm using the win32 version on XP. The CPU is an E2160 Conroe-1M, L1 32 KB per core, L2 Cache 1 MB (On-Die, ASC, Full-Speed).

2007-10-03, 13:49	#5
AES Jul 2007 Tennessee 2⁵·19 Posts	This PC has problems with FFT size 2560 using v24.14.1 as well. It seems to be fine with FFT size <= 2048.

2007-10-04, 20:27	#6
AES Jul 2007 Tennessee 1001100000₂ Posts	Bad RAM. It corrupted the LL when the FFT would not all fit in L2. It's strange that this RAM would pass the Windiag RAM tests. This is the second new PNY DDR2 module that has been bad in the last month.

2007-10-05, 18:48	#7
Bundu Jul 2004 Mid Calder, Scotland 5×37 Posts	from what i know based on posts I've read. P95 is the ultimate stress tester and tends to point out hardware faults where other systems fail. that probably why the other diagnostic didn't tell you about the problem