- Jun 30, 2004
- 16,545
- 1,976
- 126
Posting here and there and starting some rather odd threads, it may already be common knowledge that I'm building an i7-6700K system with attention to every little detail, while taking my sweet time about it.
Cooling is currently a TR LGM, which might as well be the "RT" model because I hung a 140mm Viper "round" fan on it. So far, I found the minimums for stock speeds, and switched to adaptive voltage with a perfunctory offset of 0.001V+. And I'm now at the easy 4.4 Mhz OC, but needed to tweak the "Extra Turbo" voltage because I'd lowered LLC to level 3 when the motherboard "auto" gave it either 5 or 6. So running OCCT:Linpack, the peak package temperature is 47C (RA 77F). Agent God's IBT raises the peak package temperature to 56C. The drooped load voltage is 1.200 and the unloaded peak is 1.232V. If the temperatures seem too low for an air cooler, it's because I got the chip relidded with CLU from Silicon Lottery.
I come late to the Skylake club, and I'd skipped Haswell entirely. There's this myth going around that you shouldn't use the traditional stress tests with the AVX instruction sets, because you "might hurt" your processor. The truth in the myth is that the power-virus will push a stock factory-sealed processor to bothersome temperature levels. The inaccuracy of it is that if you can't stress with the full instruction-set, you haven't really proven absolute, reliable stability, now, have you?
And like I said -- I put the Agent GOD Burn Test "Maximum" on it, and my system eats it for lunch.
then I tried LinX 0.6.5. BSOD stop code 101 before completing the first iteration. Bumped up the boost voltage a few millivolts, but the BSOD occurs almost to the same run time.
And THEN I came across an article on the "SKYLAKE BUG" when I ran a search on "skylake and LinX." Intel had issued an alert that certain "combinations" of tasks would cause Skylake to BSOD, and they had passed information to board-makers and BIOS designers for a BIOS fix to the bug. This, in PC World's explanation, was supposed to occur by March, 2016. I think my BIOS revision was released in March or later, because I was trying to decide whether to flash it to the August revision. There had already been some two or three preceding revisions.
Now my question of the moment. Anybody know anything further about this which can clarify?
Also, one more thing. I see discussions around the forums that make it seem as though confusion over the difference between VID and vCORE had never been dispelled. Some of the monitoring programs that may have given VCORE for earlier processors only show the VID values. Some monitoring programs have a feature that turns off real-time VCORE monitoring, leaving you to see only real-time VID. Or at least that was my impression.
Right now, HWInfo is my new monitor program of choice.
Cooling is currently a TR LGM, which might as well be the "RT" model because I hung a 140mm Viper "round" fan on it. So far, I found the minimums for stock speeds, and switched to adaptive voltage with a perfunctory offset of 0.001V+. And I'm now at the easy 4.4 Mhz OC, but needed to tweak the "Extra Turbo" voltage because I'd lowered LLC to level 3 when the motherboard "auto" gave it either 5 or 6. So running OCCT:Linpack, the peak package temperature is 47C (RA 77F). Agent God's IBT raises the peak package temperature to 56C. The drooped load voltage is 1.200 and the unloaded peak is 1.232V. If the temperatures seem too low for an air cooler, it's because I got the chip relidded with CLU from Silicon Lottery.
I come late to the Skylake club, and I'd skipped Haswell entirely. There's this myth going around that you shouldn't use the traditional stress tests with the AVX instruction sets, because you "might hurt" your processor. The truth in the myth is that the power-virus will push a stock factory-sealed processor to bothersome temperature levels. The inaccuracy of it is that if you can't stress with the full instruction-set, you haven't really proven absolute, reliable stability, now, have you?
And like I said -- I put the Agent GOD Burn Test "Maximum" on it, and my system eats it for lunch.
then I tried LinX 0.6.5. BSOD stop code 101 before completing the first iteration. Bumped up the boost voltage a few millivolts, but the BSOD occurs almost to the same run time.
And THEN I came across an article on the "SKYLAKE BUG" when I ran a search on "skylake and LinX." Intel had issued an alert that certain "combinations" of tasks would cause Skylake to BSOD, and they had passed information to board-makers and BIOS designers for a BIOS fix to the bug. This, in PC World's explanation, was supposed to occur by March, 2016. I think my BIOS revision was released in March or later, because I was trying to decide whether to flash it to the August revision. There had already been some two or three preceding revisions.
Now my question of the moment. Anybody know anything further about this which can clarify?
Also, one more thing. I see discussions around the forums that make it seem as though confusion over the difference between VID and vCORE had never been dispelled. Some of the monitoring programs that may have given VCORE for earlier processors only show the VID values. Some monitoring programs have a feature that turns off real-time VCORE monitoring, leaving you to see only real-time VID. Or at least that was my impression.
Right now, HWInfo is my new monitor program of choice.
Facebook
Twitter