Solution

Cool, I've had mine up there without a problem. ;)

RJR

Distinguished
Jan 7, 2009
1,065
0
19,360
However, I think that it's possible some PCH voltage will help smooth things out. Or not?? Stock I think is 1.05V, I might try 1.1V and see what happens... I guess this will be a long drawn out process for me as I first want to get my CPU OC voltages fine tuned (I think I'm at 1.28 core/1.23 VTT and it passed Intel Burn Test but haven't done Prime95). Then again, could be the crashes are from the GPU mem or core clocks. Tho, fairly certain those are ok (I've done my FurMark tests and plenty of gaming).

That's the fun part of OCing :) lots of trial and error, didn't check the data sheet but I'm sure it's over 1.10v, I think 1.14v or something (you can check) so go for it. :D
 
Isn't that the truth! I'll probably spend at least a few hours trying to fine tune my "new" vcore and vtt... and then it's on to figure out if I can fix the freeze/crashes with PCH.

I did just read the Intel specs on the chipset, got this:
PCH Absolute Maximum Ratings
Parameter Maximum Limits
Voltage on any 5 V Tolerant Pin with respect to Ground (V5REF = 5 V) -0.5 to V5REF + 0.5 V
Voltage on any 3.3 V Pin with respect to Ground -0.5 to Vcc3_3 + 0.4 V
Voltage on any 1.8 V Tolerant Pin with respect to Ground -0.5 to VccVRM + 0.5 V
Voltage on any 1.5 V Pin with respect to Ground -0.5 to VccVRM + 0.5 V
Voltage on any 1.05 V Tolerant Pin with respect to Ground -0.5 to VccIO + 0.5 V
1.05 V Supply Voltage with respect to VSS -0.5 to 1.3 V
1.8 V Supply Voltage with respect to VSS -0.5 to 3.7 V
3.3 V Supply Voltage with respect to VSS -0.5 to 3.7 V
5.0 V Supply Voltage with respect to VSS -0.5 to 5.5 V
V_CPU_IO Supply Voltage with respect to VSS -0.5 to 1.3 V
1.8 V Supply Voltage for the analog PLL with respect to VSS -0.5 to 1.98 V

PCH Power Supply Range
Power Supply Minimum Nominal Maximum
1.05 V 1.00 V 1.05 V 1.10 V
1.5 V 1.43 V 1.50 V 1.58 V
1.8 V 1.71 V 1.80 V 1.89 V
3.3 V 3.14 V 3.30 V 3.47 V
5 V 4.75 V 5.00 V 5.25 V

Sorry, I know it's a bit confusing. Here's the link http://www.intel.com/Assets/PDF/datasheet/322169.pdf (Section 8.2)

Looks like PCH can only have up to 1.1V... although I guess maybe there's different ones I have to mess with. Definitely a bit confusing but I'm fairly sure it's PCH 1.05V I need to change...
 

RJR

Distinguished
Jan 7, 2009
1,065
0
19,360
Well, that's a fun table you got there :pt1cable:

Yeah, I would stick to 1.10v for now. It appears to be the max, but it looks like the absolute max is 1.3v and that just sounds a little too high to me.
 

RJR

Distinguished
Jan 7, 2009
1,065
0
19,360

Cool, I've had mine up there without a problem. ;)
 
Solution
Well I spent almost the whole time trying to get stable. I wanted to raise my bclk in the 175-180 range but had a hell of a time trying to get stable. Finally at the end of the night I set it back to 175 and 4 times it was unstable at iirc 1.31 vcore and 1.225 vtt... apparently it was unstable from overvolt cuz I set it back down to where my previous settings were at 1.306 vcore 1.21 vtt and it passed a linpack.

Anyway, while that was all going on, for a short while I did have the PCH at 1.15, but there was 2 big crashes after that. First time the screen went all... how to explain it... ok, if the screen was a fresh paining, it was kind of like someone dragged a rake across it horizontally lol. Then I got a BSOD. After that, I changed the voltages but left PCH at 1.15, booted up. Started a burn test but it failed. I was in the kitchen at this time and after a couple minutes I look up and it's BSOD. I went over and it was something in regards to USBHub maybe. Definitely something USB. Which kindof tipped me off that maybe the 1.15V is too much and the PCH effed up since it controls the USB and all that. I rebooted and again had the screen issue but this time it simply froze so I did a hard reboot and then put the PCH back to 1.05. I think that either 1.15 was just too much, or it was ok but amplified my CPU instability (which at that time I was also trying to get stable).

I have some more time tonight to try again... I'll first try with my old settings that I know are stable (well... Intel Burn Test and Prime95 stable).
 

RJR

Distinguished
Jan 7, 2009
1,065
0
19,360
OK, if memory serves and your signature is a hint, you leave turbo enabled?

I was sitting here thinking, how could he have so much trouble going from 175 to 180, it should be so easy, now I get it.

What clock and voltage does CPU-Z show when you run IBT/Prime95?
 
Yeah I like turbo boost :D

CPUZ shows it accurately. I have load line calibration on so it's exactly what I set, rounds the vcore up from 1.30625 to 1.31 and the clock is generally 175x21... and yeah, goes up to x24.

EDIT: If I disable LLC, the voltage drops a lot. I think it said 1.28V max and drops to about 1.21V under 4 cores. I think it was 1.26V with 1 core.

I'm debating disabling turbo and go for maybe a straight 4ghz, but it seems pretty good to have it set up this way since gaming is generally 2 maybe 3 cores which will then be getting that boost.
 

RJR

Distinguished
Jan 7, 2009
1,065
0
19,360
Trying to figure out what exact voltage you need for 1 core at this speed and 2 cores at this speed, etc. is way to much for my short attention span :)

If you can figure it out, more power to you :D , I just go for the single speed OC's.

 
Well as of right now I got it stable at a flat 4ghz - 200x20 with 1.31 vcore and 1.225 vtt so nearly the same voltage as 175bclk with turbo. Right now I'm just trying to squeeze a little more speed out of the ram so 202 bclk for 1616mhz cl8. Still havn't done the pch voltag stuff lol.
Oh ya one thing with turbo off the CPU power, according to HWMonitor dropped from 108W to 95W although temps are slightly higher.
 
Well, wierd result after getting the 4ghz stable, I set my GPUs to the speeds I had last time I ran Vantage (960core, 1200 mem). Reran the test. Scored about 2k better on CPU, but my GPU score actually decreased! I still have lots of testing to do, and going to try the PCH voltage increase at some point but anyway, thanks for the insights RJR.