Core i7 920 rev D0 stepping 5 :)

[ReverendMaynard]

Something is wrong! 24/7 for my C0 920 is 4.01Ghz, 1.184 Vcore and I'm seeing only 46C at idle, maybe 55C tops in games (RealTemp). Even the stock HS should do better than that. I've witnessed a 920 that looked normal, but it would get up to 60C pretty quickly, it was obviously defective. You either have an overheater, or your cooler is severely deficient.

Did I say 24/7? 1.184 is also good for prime.

Above is what I quoted and was referring to roller. I've been in this i7 game for a while and I know 1 hour of prime is not something I would tote at 24/7.

Still nice clocks/volts though.

[naokaji]

1 hour of prime is not something I would tote at 24/7.

QFT.

My I7 needs 0.05V more for actual 24/7 100% load stability compared to just a few hours 100% load stability.

Anyway, can't wait until D0 920's are out in a meaningful way (read: shops guaranteeing that you get one with actual stock), just hope I can get enough for my crappy clocking 940 on fleabay to pay for a new 920.

[roller11]

Above is what I quoted and was referring to roller. I've been in this i7 game for a while and I know 1 hour of prime is not something I would tote at 24/7.

Still nice clocks/volts though.

thanks for the compliment, rev. You're one of the top, most helpful posters in this forum, so coming from you it means a lot.

Anyhow...
My awkwardly worded post needs clarification, I didn't mean I primed for 24 hours straight.
Usually, guys will push their chips to voltage/heat extremes to get an impressive prime achievement, then back off their voltage for 24/7 operation.
If I did the same thing, (down volt after prime at 4.01Ghz/ 1.184 vcore) then no telling how low vcore I could go for 24/7 @ 4.01 Ghz. I never bothered to see how low vcore I can go since 1.184 is plenty low enough.

[ReverendMaynard]

thanks for the compliment, rev. You're one of the top, most helpful posters in this forum, so coming from you it means a lot.

Anyhow...
My awkwardly worded post needs clarification, I didn't mean I primed for 24 hours straight.
Usually, guys will push their chips to voltage/heat extremes to get an impressive prime achievement, then back off their voltage for 24/7 operation.
If I did the same thing, (down volt after prime at 4.01Ghz/ 1.184 vcore) then no telling how low vcore I could go for 24/7 @ 4.01 Ghz. I never bothered to see how low vcore I can go since 1.184 is plenty low enough.

1.184v at 4ghz is beautiful. I've actually taken the opposite approach and have been pushing my 24/7 max clocks and not really worrying about voltages anymore lol. I'll say this though, lapping this 920 has made a nice improvement. 4.57ghz HT off, 1.55v now might peak at 80C running wprime1024m, when before that would take it into the low 100's Linx? eff that action lol.

Thanks for the compliment though, coming from you that's pretty nice to hear because imo, you're 10x more valuable to these forums.

[roller11]

I'll say this though, lapping this 920 has made a nice improvement. 4.57ghz HT off, 1.55v now might peak at 80C running wprime1024m, when before that would take it into the low 100's Linx? eff that action lol.

.

The flatness of the heat sink certainly makes a difference, I know cause I've had good results lapping various HS. I've never lapped a CPU cause the warranty goes right out the window. Then again, perhaps I'll look into it since in a way, my CPU is irreplaceable anyway. If you want, you can PM me about particular technique for lapping a CPU.

BTW, you got me wondering about running in slow mode vs full speed QPI.
Turns out, it's a function of motherboard. I've tried eight different mobos, and only the two giga byte boards make a difference. That is,
I've validated 222x21 in SM with the giga UD5 but with full speed, I could do only
220x21. With the biostar x58, asus, DFI, and MSI boards, I can get 220-222x21 (depending on board) at full speed. In fact, I just set my personal record with the biostar x58 with full speed QPI.

[ReverendMaynard]

The flatness of the heat sink certainly makes a difference, I know cause I've had good results lapping various HS. I've never lapped a CPU cause the warranty goes right out the window. Then again, perhaps I'll look into it since in a way, my CPU is irreplaceable anyway. If you want, you can PM me about particular technique for lapping a CPU.

BTW, you got me wondering about running in slow mode vs full speed QPI.
Turns out, it's a function of motherboard. I've tried eight different mobos, and only the two giga byte boards make a difference. That is,
I've validated 222x21 in SM with the giga UD5 but with full speed, I could do only
220x21. With the biostar x58, asus, DFI, and MSI boards, I can get 220-222x21 (depending on board) at full speed. In fact, I just set my personal record with the biostar x58 with full speed QPI.

No need to lap the GTZ though, they are as flat as can be. During the various lapping stages that I did with this 920, the evidence of a severely warped IHS as I wore away the high spots was pretty clear. It took a good hour, going from 400g - 600g - 1000g - 1500g to get it down to where I deemed it as perfectly flat. Highly recommended if you don't give a rats ass about warranty