![]() |
![]() |
#1 |
Jul 2014
2·32·52 Posts |
![]()
Hi there,
I've two machines with AMD FX6300 cpus running LL tests on all 6 cores. On one of them, the system monitor says all cpus running at 99%. On the other, the monitor says all cpu's are at about 10%. Can anyone explain this? Is there something wrong? |
![]() |
![]() |
![]() |
#2 |
6809 > 6502
"""""""""""""""""""
Aug 2003
101×103 Posts
1088810 Posts |
![]()
Are they absolutely identical? Including memory and power supply?
Open up Prime95 and check the 10% machine. Check to see if it is running P-1's on the exponents that you are assigned (it might do this before the LL tests.) What is the range of the exponents you are testing on each machine (the nearest 1,000,000 is close enough.) How are the temperatures on the slow machine? |
![]() |
![]() |
![]() |
#3 |
Jul 2014
2×32×52 Posts |
![]()
>>>>>>>>>>>>>>>>Are they absolutely identical? Including memory and power supply?
the memory is the same. DDR31666 The boards are different and the PSU's. The range of exponents is 69-70 million. The worker threads say they are doing LL tests. The one question I'm not sure how to answer is the temperature question. How do I check that? |
![]() |
![]() |
![]() |
#4 |
Jun 2003
22·32·151 Posts |
![]()
Correct me if I'm wrong, but, if the CPU is throttling due to temperature, won't the task manager still show 100%, just that the actual runtimes would be slower?
|
![]() |
![]() |
![]() |
#5 |
1976 Toyota Corona years forever!
"Wayne"
Nov 2006
Saskatchewan, Canada
122338 Posts |
![]()
I found a BIG issues with memory contention on a new upper end PC I have doing PrimeNet. A 4-core Intel Haswell i7-4770.
It was build by a Server Admin; we have absolutely no say in the set up and have to live with it as is. The full-meal-deal thread is here http://www.mersenneforum.org/showthread.php?t=19727 Following is the short version. It was NOT built with dual channel memory....and only 4G If I run 1 LL test on 1 core in the 60M range I will see iteration times around 20ms ...whoo hooo!!!! With 2 LL tests on 2 cores (and the other 2 idle) the iteration times increase to about 33ms With 3 LL tests ... about 48ms With all 4 doing LL ... about 65ms. booo hooo!!!! I didn't look at CPU utilization but it is a safe bet with this much RAM contention it was well under 100%. I am sadly taking somewhat of a middle ground and running 2 LL tests and letting the other 2 cores do TF because it does NOT impact the LL test times of the other 2 cores. Even letting 1 or 2 of the other cores do small ECM slows down the LL workers about as much as more LL tests. You might want to see what times you get only running 1 or 2 or 3 ... LL tests by stopping the other workers temporarily. |
![]() |
![]() |
![]() |
#6 | ||
Dec 2014
3710 Posts |
![]() Quote:
Quote:
If you want to check CPU info, you can use CPU-Z. You can check if the FSB and multiplier are normal. Also I think you can set throttling in prime95 to 10% using Throttle=10 in prime.(something) Just in case you have set it for some odd reasons. |
||
![]() |
![]() |
![]() |
#7 |
Jul 2014
2×32×52 Posts |
![]()
Thanks guys. I've still to act on some of your advice. The last thing I checked is running 3 threads instead of 6 and it seems to make no difference.
|
![]() |
![]() |
![]() |
#8 |
1976 Toyota Corona years forever!
"Wayne"
Nov 2006
Saskatchewan, Canada
52×211 Posts |
![]()
At least version 28.5 of Prime95?
The 64 bit version on a 64 bit install of the OS? (or 32 and 32)? |
![]() |
![]() |
![]() |
#9 |
Jul 2014
2×32×52 Posts |
![]()
linux64 prime95 28.5 build 2
|
![]() |
![]() |
![]() |
#10 |
Undefined
"The unspeakable one"
Jun 2006
My evil lair
2·32·7·53 Posts |
![]()
I agree that the CPU core won't be at 100% but the task manager will still show 100% because the system scheduler has no idea of the finer details of what each task is doing. The task manager 100% value that you see is only indicative of the scheduler's software switching activity for each task; not the task's hardware activity of waiting for RAM, or the FPU, to complete each transaction.
|
![]() |
![]() |
![]() |
Thread Tools | |
![]() |
||||
Thread | Thread Starter | Forum | Replies | Last Post |
IHS TIM and de-lidding CPUs | kladner | Hardware | 2 | 2017-02-11 06:38 |
Number of CPUs to use | houding | Software | 3 | 2015-02-26 19:56 |
Running other programs while running Prime95. | Neimanator | PrimeNet | 14 | 2013-08-10 20:15 |
Whither Older CPUs? | Rodrigo | Operation Billion Digits | 4 | 2010-10-20 15:12 |
A tale of 3 CPUs | chris2be8 | Hardware | 7 | 2010-07-20 23:12 |