• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

i7 920 first attempts

It's been discussed before, but I want to comment on the extra OCing headroom gained by disabling Hyperthreading.

With HT on, 4.0GHz required 1.45 volts and throttled back due to the heat within the first minute of a Prime95.

With HT off:
4.0GHz [200x20] - 1.35V - 1600 @ 8-8-8-24 - hottest core 72°C after 5 minutes of Prime95.
4.1GHz [205x20] - 1.40V - 1640 @ 9-8-8-24 - hottest core 80°C after 5 minutes of Prime95.
4.2GHz [210x20] - 1.45V - 1680 @ 9-8-8-24 - Prime95 instant BSOD
4.2GHz [200x21] - 1.45V - 1600 @ 8-8-8-24 - hottest core 90°C after 5 minutes of Prime95.

I've been running Prime95 on the latter settings for about 15 minutes now (screenshot) I'm gonna try some gaming on it and see if it makes any noticeable difference. For my CPU's sake, I hope not. ;)
 
Well, the reason it'll clock higher is because the execution units aren't being kept as busy. Then again, if what you use it for is faster anyway ... then who can argue with that? :) I for one will not be disabling HT, especially since one of the reasons I got this is for multithreaded development work. I also got an Atom 330 for lower-end performance testing :)

I finally got a working motherboard and I'm up to 3.46ghz. All I've done so far is increase the BCLK in the BIOS :D
 
I run a lot of multithreaded apps as well so HT stays on for my 24/7 oc. Still it's nice to know disabling HT can help me eke out a little bit of extra performance for older single-threaded apps, if it were ever worth the effort of rebooting and reconfiguring the BIOS temporarily for such an app. :)

As for gaming, I didn't really notice a subjective difference (4.2 w/o HT vs. 3.8 w/ HT). Loading a level, which is traditionally a single-threaded task, ran a bit faster. In Crysis Warhead SP loading "Adapt or Perish": 22.3s @ 4.2GHz vs. 25.9s @ 3.8GHz. Not from a fresh boot with either time though, so who knows... Either way, the progress bar absolutely flies across the screen compared to my old rig. :cool:
 
Vegasr said:
I bet you'll be good at the 8-7-7-20 setting. I'd be curious to see the differences in dropping those two to 7 does versus 8 on some memory bench tests like Everest and such.
Click to expand...

I played around with Everest but it said it wasn't optimized for my CPU. The memory read and copy scores varied as much as 10% from run to run.

After experimenting a bit, it seems that tRCD (the last number) should be the sum of the other 3. Deviating from that, either way, seemed to increase latencies. 8-7-7-20 had 0.2ns more latency than 8-7-7-22.

I'm trying a new Bclk now...went from 190x20 to 200x19. That way my memory runs at its full rated speed. Doing so increased my read/write speeds 1GB/s each. (16-17.5 read, 13.5-14.2 write, 19.5-20.5 copy)

The Everest latency readings were repeatable.
8-8-8-24 @ 1540: 32.5ns @ 1600: 31.1ns
8-7-7-22 @ 1540: 32.2ns @ 1600: 30.1ns
 
argh.. Intel Burn Test just burnt a huge hole through my OC, and in just minutes.

I'm up to 1.35V to stay stable with IBT *and* 4 threads of Prime95 running simultaneously. I should probably run Core Damage alongside the other 2 just to be sure. :p
 
Just run Core Damage by itself, that is max heat generation ... running other stuff will just slow it down.

I could have told you that at 90C running Prime95, that Core Damage would hit 100C+... I gave up with any voltage over 1.30v, my TRUE setup isn't enough.
 
MechGeek said:
I'm trying a new Bclk now...went from 190x20 to 200x19. That way my memory runs at its full rated speed. Doing so increased my read/write speeds 1GB/s each. (16-17.5 read, 13.5-14.2 write, 19.5-20.5 copy)

The Everest latency readings were repeatable.
8-8-8-24 @ 1540: 32.5ns @ 1600: 31.1ns
8-7-7-22 @ 1540: 32.2ns @ 1600: 30.1ns
Click to expand...

I'm gonna give that a try when I get home. I know it increases memory speeds, but does it effect the CPU at all?
 
You must log in or register to reply here.
Back
Top