Hello everyone,
well as you can guess I'm not too happy as my new rig isn't quite the modest beast I intended it to be.
Remember here, I'm at stock clock. Also, Intel Turbo Boost on, RAM correctly displayed in the BIOS at 1600 Mhz, voltage on par with tech specs, no fail nor error whatsoever during computer fresh install and usual customizing procedures.
First ran Prime 95 small and large FFT yesterday for about 10 minutes each with no problems. I know it ain't nearly long enough, but I just wanted to see if heat was ok, and run the real test today. Heat was fine and no fail. Tonight, small FTT ran fine for the same lenght of time, so I switched to Large just to take a look before going with the real testing. I got one core failing after a couple of runs only. Re-tried for the same result, but this time on another core. Then ran Blend test, and 3 cores failed within the first minute.
Ran Intel Burn Test set to 10 pass and maximum burn... one pass was fine, second pass failed. Of course, both tools gave similar inconclusive info as to why, which you must already know.
Now the first thing that ticked me yesterday was that one of my cores was running a little fever compared to the others. Always 5-6 degrees (C) over the "coolest" core, and 2-3 degrees over the average of the three others - which presented in themselves no more significant difference than 1 degree average, and were mostly exchanging similar results during the load. Same thing today, same core always hotter with both test - though ITB did get some of the other cores to reach higher temps at one point. What is ticking me even more is that in all three or four fails I got running Prime95, never did the hottest core failed, even though all three others failed in turn. When they failed together running Blend torture, I let the test run for a number of minutes more and the hottest core was still up to the task.
I've read that memory could be a problem... but the last bit about the hottest core makes me wonder why it did not fail too at one point or another, would it be my problem. Still, after posting this I'm gonna remove one stick and see how it goes - then of course the other if the first fails. Then if both I got a spare pair to test too.
My concern is going more towards the CPU... By the way, I did check if there was a voltage drop when cores were failing, and there was not. CoreTemp gave me the correct, boosted 3.7 GHz speed and 1.3 V all of the time, without even a tick when cores were failing. And just to dismiss the obvious questions, my temps are real fine, between 21-29 at idle speed (hottest core still above giving that one time only 29 result), between 45-58 at 100% load (both range including the wildest results). Hottest it got was 61 during ITB - of course, again the fevered core, but it was for a fraction of second, nothing like the fairly consistent 58 I got with the hottest core during one of the Prime run. There's a question there, about that hotter core... Bad thermal pasting? Normal instance? Case/built particularity?
See rig in signature. All comments and help will be deeply appreciated. Will post back with RAM testing results soon...
well as you can guess I'm not too happy as my new rig isn't quite the modest beast I intended it to be.
Remember here, I'm at stock clock. Also, Intel Turbo Boost on, RAM correctly displayed in the BIOS at 1600 Mhz, voltage on par with tech specs, no fail nor error whatsoever during computer fresh install and usual customizing procedures.
First ran Prime 95 small and large FFT yesterday for about 10 minutes each with no problems. I know it ain't nearly long enough, but I just wanted to see if heat was ok, and run the real test today. Heat was fine and no fail. Tonight, small FTT ran fine for the same lenght of time, so I switched to Large just to take a look before going with the real testing. I got one core failing after a couple of runs only. Re-tried for the same result, but this time on another core. Then ran Blend test, and 3 cores failed within the first minute.
Ran Intel Burn Test set to 10 pass and maximum burn... one pass was fine, second pass failed. Of course, both tools gave similar inconclusive info as to why, which you must already know.
Now the first thing that ticked me yesterday was that one of my cores was running a little fever compared to the others. Always 5-6 degrees (C) over the "coolest" core, and 2-3 degrees over the average of the three others - which presented in themselves no more significant difference than 1 degree average, and were mostly exchanging similar results during the load. Same thing today, same core always hotter with both test - though ITB did get some of the other cores to reach higher temps at one point. What is ticking me even more is that in all three or four fails I got running Prime95, never did the hottest core failed, even though all three others failed in turn. When they failed together running Blend torture, I let the test run for a number of minutes more and the hottest core was still up to the task.
I've read that memory could be a problem... but the last bit about the hottest core makes me wonder why it did not fail too at one point or another, would it be my problem. Still, after posting this I'm gonna remove one stick and see how it goes - then of course the other if the first fails. Then if both I got a spare pair to test too.
My concern is going more towards the CPU... By the way, I did check if there was a voltage drop when cores were failing, and there was not. CoreTemp gave me the correct, boosted 3.7 GHz speed and 1.3 V all of the time, without even a tick when cores were failing. And just to dismiss the obvious questions, my temps are real fine, between 21-29 at idle speed (hottest core still above giving that one time only 29 result), between 45-58 at 100% load (both range including the wildest results). Hottest it got was 61 during ITB - of course, again the fevered core, but it was for a fraction of second, nothing like the fairly consistent 58 I got with the hottest core during one of the Prime run. There's a question there, about that hotter core... Bad thermal pasting? Normal instance? Case/built particularity?
See rig in signature. All comments and help will be deeply appreciated. Will post back with RAM testing results soon...
Last edited:
