LLR beta Version 3.8.13 (deprecated) - Page 3

Prime95 · 2014-03-07, 00:57

Quote:

Originally Posted by pinhodecarlos

I made the test. My result, same as yours....humm...better call Prime95.

I had no problems in prime95 on 32-bit Windows and 64-bit Linux (Haswell with FMA). I'm not all that familiar with LLR, I need to know what your CPU is, 32-bit or 64-bit, multi-threaded or not, etc.

pinhodecarlos · 2014-03-07, 01:26

Quote:

Originally Posted by Prime95

I had no problems in prime95 on 32-bit Windows and 64-bit Linux (Haswell with FMA). I'm not all that familiar with LLR, I need to know what your CPU is, 32-bit or 64-bit, multi-threaded or not, etc.

My laptop is an i7-3630QM (Ivy Bridge) running Win 7 64 bit.
I re-run the test shutting down my current instances of LLR and didn't get any error. The first run was made with two LLR instances running in parallel. I think it is related to heat problems because I can't run on this machine more than 2 instances of LLR otherwise the CPU starts throttling with high temps.

pepi37 · 2014-03-07, 07:43

If 48°C is high temperature for Intel.....I have I7-2700K under 64 bit Win 7.
On one or on all four, error is same...since 3.8.11 has no error it is something in new LLR code.

rebirther · 2014-03-07, 08:15

Quote:

Originally Posted by Prime95

I had no problems in prime95 on 32-bit Windows and 64-bit Linux (Haswell with FMA). I'm not all that familiar with LLR, I need to know what your CPU is, 32-bit or 64-bit, multi-threaded or not, etc.

I have made the same test with the same result on my i7-4930k, 64bit, HT off.

pepi37 · 2014-03-07, 11:25

http://www.dropbox.com/s/zvsvhm7k9kc6jmq/Clipboard02.jpg

When printing errors show - FERMAT PRP TEST IN PROGRESS

pinhodecarlos · 2014-03-07, 11:39

pepi37,

Would you try only run one instance of LLR? We need to filter what is going on.
In my case it runs well if I only have one instance running. When I add a third it breaks.

pepi37 · 2014-03-07, 12:07

When I something dont understand ( English is not my language) and when someone of you write it before then you "yell on me"
I already answer to this in my previous post

Quote:

On one or on all four, error is same...since 3.8.11 has no error it is something in new LLR code.

I do retest on GUI and cllr ( both 64 bit) of 3.8.11 and both works perfect.
I found two versions of GUI 64 bit LLR (3.8.12) and both have same error.

Prime 28.4 have also same error, results are attached
Prime 28.3 have also same error.
Prime 27.7 and 27.9 are error free.....

pinhodecarlos · 2014-03-07, 13:21

Sorry, I red your message at 4 am, I was sleepy. Sorry.

Prime95 · 2014-03-07, 14:40

I'm able to reproduce the problem. It affects the AVX FFTs, but not the FMA FFTs.

pepi37 · 2014-03-07, 14:50

Great news!
So it is not my hardware fault or temperature issue...
I hope fixed version will be available soon...

Prime95 · 2014-03-07, 15:39

The bug only affects zero-padded non-base-2 AVX FFTs larger than 6K. The FFT is fine, but the calculated roundoff error is garbage. I'll have a fix soon.

2014-03-07, 14:50	#32
pepi37 Dec 2011 After milion nines:) 5×17² Posts	Great news! So it is not my hardware fault or temperature issue... I hope fixed version will be available soon... Last fiddled with by pepi37 on 2014-03-07 at 14:52

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
LLR Version 3.8.18 released [deprecated]	Jean Penné	Software	43	2017-02-20 12:05
LLR Version 3.8.17 released [deprecated]	Jean Penné	Software	18	2017-02-01 12:49
LLR Version 3.8.14 released (deprecated)	Jean Penné	Software	67	2015-05-02 07:24
Prime95 version 28.5 (deprecated, use 28.7)	Prime95	Software	162	2015-04-05 16:19
Beta version 24.12 available	Prime95	Software	33	2005-06-14 13:19

2014-03-07, 07:43	#25
pepi37 Dec 2011 After milion nines:) 5×17² Posts	If 48°C is high temperature for Intel.....I have I7-2700K under 64 bit Win 7. On one or on all four, error is same...since 3.8.11 has no error it is something in new LLR code.

2014-03-07, 11:25	#27
pepi37 Dec 2011 After milion nines:) 5×17² Posts	http://www.dropbox.com/s/zvsvhm7k9kc6jmq/Clipboard02.jpg When printing errors show - FERMAT PRP TEST IN PROGRESS

2014-03-07, 11:39	#28
pinhodecarlos "Carlos Pinho" Oct 2011 Milton Keynes, UK 3×17×97 Posts	pepi37, Would you try only run one instance of LLR? We need to filter what is going on. In my case it runs well if I only have one instance running. When I add a third it breaks.

2014-03-07, 13:21	#30
pinhodecarlos "Carlos Pinho" Oct 2011 Milton Keynes, UK 3×17×97 Posts	Sorry, I red your message at 4 am, I was sleepy. Sorry.

2014-03-07, 14:40	#31
Prime95 P90 years forever! Aug 2002 Yeehaw, FL 7526₁₀ Posts	I'm able to reproduce the problem. It affects the AVX FFTs, but not the FMA FFTs.

2014-03-07, 15:39	#33
Prime95 P90 years forever! Aug 2002 Yeehaw, FL 2·53·71 Posts	The bug only affects zero-padded non-base-2 AVX FFTs larger than 6K. The FFT is fine, but the calculated roundoff error is garbage. I'll have a fix soon.