mersenneforum.org  

Go Back   mersenneforum.org > Great Internet Mersenne Prime Search > Hardware

Reply
 
Thread Tools
Old 2010-03-13, 02:34   #1
AZMango
 
Mar 2010

3·5 Posts
Question Iteration Times Vary on Core i7

I put Prime 95 on my new Core i7 920 PC. I used the default settings so Prime95 started four LL tests with Multithreading set to 2. My iteration times were around .060 to .065. I rebooted one day and after checking I saw that the iteration times had fallen to about .035. Those times stayed for awhile, but then after rebooting I noticed it was back to the .060 times. I rebooted again and it went back to .035 once again. I did a bios update and ever since the times stay at the .060 range and haven't dropped back down again. I did some experimenting and noticed that if I set the multithreading to just one CPU that the times drop to about .055.

So, my question is, is there some kind of built in overclocking going on? Why would the times vary by so much? Can I get the times back down to the .035 range once again? Should the Multithreading be set to just 1?

I did a few other tests and if I run one LL with eight cores helping then the time was .016. I'm guess that is still higher than it should be.

Any suggestions?
AZMango is offline   Reply With Quote
Old 2010-03-13, 17:20   #2
petrw1
1976 Toyota Corona years forever!
 
petrw1's Avatar
 
"Wayne"
Nov 2006
Saskatchewan, Canada

2·7·331 Posts
Default

Quote:
Originally Posted by AZMango View Post
I put Prime 95 on my new Core i7 920 PC. I used the default settings so Prime95 started four LL tests with Multithreading set to 2. My iteration times were around .060 to .065. I rebooted one day and after checking I saw that the iteration times had fallen to about .035. Those times stayed for awhile, but then after rebooting I noticed it was back to the .060 times. I rebooted again and it went back to .035 once again. I did a bios update and ever since the times stay at the .060 range and haven't dropped back down again. I did some experimenting and noticed that if I set the multithreading to just one CPU that the times drop to about .055.

So, my question is, is there some kind of built in overclocking going on? Why would the times vary by so much? Can I get the times back down to the .035 range once again? Should the Multithreading be set to just 1?

I did a few other tests and if I run one LL with eight cores helping then the time was .016. I'm guess that is still higher than it should be.

Any suggestions?
My first instinct tells me you should look at possible overheating: I've had a few PC's slow down noticeably when they get to hot.
- Better cooler
- More fans
- Open the case
- You can check quickly by installing software called coretemp.

I assume you have already checked in task Manager that Prime95 is in fact getting close to 100% of the CPU and nothing else is interfering?
petrw1 is offline   Reply With Quote
Old 2010-03-13, 21:10   #3
Prime95
P90 years forever!
 
Prime95's Avatar
 
Aug 2002
Yeehaw, FL

1D3D16 Posts
Default

Try running with 4 workers, but no multi-threading.
Prime95 is offline   Reply With Quote
Old 2010-03-14, 00:05   #4
AZMango
 
Mar 2010

3·5 Posts
Default

Quote:
Originally Posted by petrw1 View Post
My first instinct tells me you should look at possible overheating: I've had a few PC's slow down noticeably when they get to hot.
- Better cooler
- More fans
- Open the case
- You can check quickly by installing software called coretemp.

I assume you have already checked in task Manager that Prime95 is in fact getting close to 100% of the CPU and nothing else is interfering?
It's a new Dell PC so I didn't add anything extra to it. I installed Coretemp and with two helpers each the temps are from 82 to 86 C with 100% utilization. Average iteration times are about .061. Are these temps within a safe range? If I change to just one helper window then the times drop down to about .055, the temp is around 78 to 82, and the load is at about 52%.
AZMango is offline   Reply With Quote
Old 2010-03-14, 00:09   #5
AZMango
 
Mar 2010

11112 Posts
Post

Quote:
Originally Posted by Prime95 View Post
Try running with 4 workers, but no multi-threading.
I do get faster times this way, at about .055, but still nowhere near what had shown up before, about .035. The PC itself doesn't feel any slower and the Experience Index still comes in with the same ratings. Why is it that not having a helper would slow things down? I noticed if I run just one or two workers then adding extra helpers seems to speed things up.
AZMango is offline   Reply With Quote
Old 2010-03-14, 00:18   #6
petrw1
1976 Toyota Corona years forever!
 
petrw1's Avatar
 
"Wayne"
Nov 2006
Saskatchewan, Canada

110328 Posts
Default

Quote:
Originally Posted by AZMango View Post
It's a new Dell PC so I didn't add anything extra to it. I installed Coretemp and with two helpers each the temps are from 82 to 86 C with 100% utilization. Average iteration times are about .061. Are these temps within a safe range? If I change to just one helper window then the times drop down to about .055, the temp is around 78 to 82, and the load is at about 52%.
Temps seem high ... I don't know what kind of coolers Dell uses (you ordered?). If Dell did not know you intended to work the PC to 100% they would not know you need a high end cooler.
Before you consider a better cooler or more fans you could try a few simple test:
- see if opening the case helps the temp.
- be sure it is not in a hot room or enclosure
--- one daughter moved her PC into a cabinet and it could NOT run at 100% without overheating and crashing
--- the other daughter's PC got clogged with dust and the Prime95 iterations times eventually TRIPLED!!!!

A few of us have new i5-750 Quads and at 100% utilization and OverClocked by 20-50% we are still getting temps around 60.

In most cases reported here adding the hyperthreading helper thread either makes no difference or slows it down slightly. The helper is NOT any REAL hardware. Prime95 is coded very tight so it is unlikely there will be much idle CPU time that hyperthreading can help with.

Some of my Hyperthreaded PC's show 50% utilization in task manager without hyperthreading; even though the PC is truly 100% used.

Last fiddled with by petrw1 on 2010-03-14 at 00:26 Reason: daughters
petrw1 is offline   Reply With Quote
Old 2010-03-14, 00:23   #7
cheesehead
 
cheesehead's Avatar
 
"Richard B. Woods"
Aug 2002
Wisconsin USA

170148 Posts
Default

Quote:
Originally Posted by AZMango View Post
If I change to just one helper window < snip > the load is at about 52%.
My understanding is that the task manager computes load as though the virtual CPUs were as real as the real CPUs. They aren't.
cheesehead is offline   Reply With Quote
Old 2010-03-14, 00:51   #8
AZMango
 
Mar 2010

178 Posts
Default

Quote:
Originally Posted by petrw1 View Post
Temps seem high ... I don't know what kind of coolers Dell uses (you ordered?). If Dell did not know you intended to work the PC to 100% they would not know you need a high end cooler.
Before you consider a better cooler or more fans you could try a few simple test:
- see if opening the case helps the temp.
- be sure it is not in a hot room or enclosure
--- one daughter moved her PC into a cabinet and it could NOT run at 100% without overheating and crashing
--- the other daughter's PC got clogged with dust and the Prime95 iterations times eventually TRIPLED!!!!

A few of us have new i5-750 Quads and at 100% utilization and OverClocked by 20-50% we are still getting temps around 60.

In most cases reported here adding the hyperthreading helper thread either makes no difference or slows it down slightly. The helper is NOT any REAL hardware. Prime95 is coded very tight so it is unlikely there will be much idle CPU time that hyperthreading can help with.

Some of my Hyperthreaded PC's show 50% utilization in task manager without hyperthreading; even though the PC is truly 100% used.
I opened the case and the temps are at about 72 to 76 so about a six degree drop. It seems to have a big cooler from what I can tell, but not big enough, I assume. I do like how quiet this PC is so if those temps are safe enough, maybe that's good enough.

Back to the varying times. Do you think if I reboot with the temps cooler that the iteration times could go back to the .035 range? The lower temps didn't make any difference in the times at the moment.

Thanks for all of the replies.
AZMango is offline   Reply With Quote
Old 2010-03-14, 00:53   #9
AZMango
 
Mar 2010

F16 Posts
Default

Quote:
Originally Posted by cheesehead View Post
My understanding is that the task manager computes load as though the virtual CPUs were as real as the real CPUs. They aren't.
So, are you saying that even though it says 52%, it's really 100%? How about when the task manager thinks it's 100%?
AZMango is offline   Reply With Quote
Old 2010-03-14, 01:03   #10
petrw1
1976 Toyota Corona years forever!
 
petrw1's Avatar
 
"Wayne"
Nov 2006
Saskatchewan, Canada

10010000110102 Posts
Default

Quote:
Originally Posted by AZMango View Post
I opened the case and the temps are at about 72 to 76 so about a six degree drop. It seems to have a big cooler from what I can tell, but not big enough, I assume. I do like how quiet this PC is so if those temps are safe enough, maybe that's good enough.

Back to the varying times. Do you think if I reboot with the temps cooler that the iteration times could go back to the .035 range? The lower temps didn't make any difference in the times at the moment.

Thanks for all of the replies.
Did you say the 6 degree temp drop made no difference on the iteration times? Seems odd? 72-76 is not bad but a little higher than I would like.

If you routinely see Prime95 at about 100% (or 50% without Helpers) then that rules out other things stealing cycles.

Bigger coolers are not necessarily louder. The one recommended to me (Noctua) for my i5-750 is quieter than the standard cooler/fan on my other quad.

Not sure how you define "big cooler" but mine was big enough that I could not close the case...though bigger does NOT necessarily mean better cooling either.

But yes I would say that according to others benchmarks with the i7-920 you should be around 0.035 with 4 LL in the 30-35M range.

http://www.mersenneforum.org/showthr...d=1#post208318

Last fiddled with by petrw1 on 2010-03-14 at 01:10
petrw1 is offline   Reply With Quote
Old 2010-03-14, 01:14   #11
axn
 
axn's Avatar
 
Jun 2003

115408 Posts
Default

Quote:
Originally Posted by AZMango View Post
I put Prime 95 on my new Core i7 920 PC. I used the default settings so Prime95 started four LL tests with Multithreading set to 2. My iteration times were around .060 to .065. I rebooted one day and after checking I saw that the iteration times had fallen to about .035.
Did all 4 LL tests show .035 simultaneously?! (Doubtful). Your timing of 0.055 is more like what is expected of your CPU. Hyperthreading is not at all useful for P95. Running 4 single-threaded LL is the most efficient use of your CPU (as far as P95 is concerned).
axn is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Iteration times in i5 and i7 Jud McCranie Information & Answers 53 2013-08-17 19:09
What are your per-iteration times? LiquidNitrogen Hardware 22 2011-07-12 23:15
CPU frequency and iteration times. rx7350 Hardware 12 2006-05-08 21:54
LLR.exe FFT crossovers and iteration times SlashDude 15k Search 0 2004-01-28 05:47
slow iteration times PLeopard Hardware 9 2003-10-29 05:48

All times are UTC. The time now is 11:43.

Sun May 9 11:43:30 UTC 2021 up 31 days, 6:24, 0 users, load averages: 2.69, 2.89, 2.78

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, Jelsoft Enterprises Ltd.

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.