Haswell Preview Benchmark - Page 35

nucleon · 2013-10-21, 09:22

Recall, we are pushing above the vendor's rated specs here. And as such, one shouldn't judge the platform for stability.

-- Craig

ldesnogu · 2013-10-21, 10:42

Quote:

Originally Posted by nucleon

Recall, we are pushing above the vendor's rated specs here. And as such, one shouldn't judge the platform for stability.

I wonder if people with stock cooler can run Linpack or latest Prime95 without the CPU throttling down. I bet they can't. Of course it could be argued it's not a stability issue, but given how Haswell behaves, I have decided not to OC it and regret to have picked a castrated 4770K instead of a 4770 or a 4771.

nucleon · 2013-10-21, 13:19

IVR + AVX + OC'd mem + highly optimized code, sends the temp thru the roof.

Running any other code is fine.

I pulled the CPU out again, looked at removing the locking mech, but need star screwdriver. Going to have to find one.

Gave me a chance to redo the paste. Not much difference, except core #2 is now the hottest, previously it was core #0. :)

-- Craig

Prime95 · 2013-10-21, 14:16

Quote:

Originally Posted by nucleon

IVR + AVX + OC'd mem + highly optimized code, sends the temp thru the roof.

My Haswells (4GHz, DDR3-2400) are running in the mid-70s and mid-80s doing normal LL testing. I doubt temps are the source of my crash-once-every-few-days instability.

fivemack · 2013-10-21, 17:50

Quote:

Originally Posted by Prime95

My Haswells (4GHz, DDR3-2400) are running in the mid-70s and mid-80s doing normal LL testing. I doubt temps are the source of my crash-once-every-few-days instability.

Ah, you have that too? I was assuming that it was a driver issue, to be expected from running a Linux distribution older than the CPU, and would magically fix itself when I went to Ubuntu-13.10.

Mark Rose · 2013-10-21, 18:23

I'm running Kubuntu 13.04 on a Haswell 3770 at stock clocks without issue (Hyper 212 Evo keeping it < 86°C).

Prime95 · 2013-10-21, 18:24

Quote:

Originally Posted by fivemack

Ah, you have that too? I was assuming that it was a driver issue, to be expected from running a Linux distribution.

Crap. I was going to try running Linux to see if my problem was a Windows-only issue.

ewmayer · 2013-10-21, 20:19

Quote:

Originally Posted by Prime95

My Haswells (4GHz, DDR3-2400) are running in the mid-70s and mid-80s doing normal LL testing. I doubt temps are the source of my crash-once-every-few-days instability.

Is this running v27 or 28? Tried reverting the CPU to stock speed? [I realize that would be annoying, after your time+money investment in high-end cooling.]

Prime95 · 2013-10-22, 00:09

Quote:

Originally Posted by ewmayer

Is this running v27 or 28? Tried reverting the CPU to stock speed? [I realize that would be annoying, after your time+money investment in high-end cooling.]

Running v28. I'm slowly increasing voltage to see if the problem goes away. It is a very slow process as it can take 3+ days before you realize that didn't solve the problem.

nucleon · 2013-10-22, 01:27

Quote:

Originally Posted by Prime95

My Haswells (4GHz, DDR3-2400) are running in the mid-70s and mid-80s doing normal LL testing. I doubt temps are the source of my crash-once-every-few-days instability.

Cool. My thinking is that if there was a invalid voltage - you'd be hitting that with invalid results rather than system reboot.

System reboots generally occur when all the auto-down clocking/voltages are turned off and it's the last course of action for the CPU to save itself.

If it isn't temperature, than it could be the power/amps failsafes.

There's an option in the bios to modify the allowed power/amps to the CPU. By default it's 84W/95amps.

Why it happens every few days - it's probably due to the probability of hitting high usage stuff all at once. All 4 cores sync with high power consumption at same time - then boom, reboot.

-- Craig

Prime95 · 2013-10-22, 02:09

Quote:

Originally Posted by nucleon

There's an option in the bios to modify the allowed power/amps to the CPU. By default it's 84W/95amps.

Why it happens every few days - it's probably due to the probability of hitting high usage stuff all at once. All 4 cores sync with high power consumption at same time - then boom, reboot.

I'm running at ~120W, so I suspect the power/amps failsafes are disabled.
If it were a high power spike problem, then the voltage increases I've been applying the last few weeks would be making the problem worse.

Can memory timings or overheating cause a kernel panic? Any other ideas as to the cause?

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Haswell-E Prelim. Benchmark	sdbardwick	Hardware	37	2015-02-10 18:49
Prime95 and Haswell	Pleco	Information & Answers	22	2014-07-13 16:03
Haswell Rig	Mini-Geek	Hardware	64	2014-05-27 13:22
Prime95 version 27.1 early preview, not-even-close-to-beta release	Prime95	Software	126	2012-02-09 16:17
Missing mouse-over preview text	retina	Forum Feedback	1	2011-09-12 15:32

2013-10-21, 09:22	#375
nucleon Mar 2003 Melbourne 5·103 Posts	Recall, we are pushing above the vendor's rated specs here. And as such, one shouldn't judge the platform for stability. -- Craig

2013-10-21, 13:19	#377
nucleon Mar 2003 Melbourne 5·103 Posts	IVR + AVX + OC'd mem + highly optimized code, sends the temp thru the roof. Running any other code is fine. I pulled the CPU out again, looked at removing the locking mech, but need star screwdriver. Going to have to find one. Gave me a chance to redo the paste. Not much difference, except core #2 is now the hottest, previously it was core #0. :) -- Craig

2013-10-21, 18:23	#380
Mark Rose "/X\(‘-‘)/X\" Jan 2013 2×5×293 Posts	I'm running Kubuntu 13.04 on a Haswell 3770 at stock clocks without issue (Hyper 212 Evo keeping it < 86°C).