Core i7/X58 Overclocking Thread

[beast200]

To lap or not to lap that is the question, whether it is nobler ...
OK - enough - so it has been suggested that I should lap my cpu to try to get better heat xfer.
I am having 2nd thoughts about doing this because doing so voids the intel warranty (on a $1K processor!)
The reason this was suggested was the load temps that I have been posting.
So those of you that have an i7 under water - please take a look at the three screen captures below.
All three are on water - the first is with everything at stock, the second is at 4.0 HT on and the third is at 4.2 HT on and at vcore of 1.55v.
Can anyone confirm that these are inline with your experience or are they high for water?

i am at 21 x 2000 @1.425volts with HT off and my max temps are 69 -70
i think our temps are ok as you have higher volts and ht on

[SteveRo]

Unless Asus has released a new BIOS that changed the requirements for Turbo to be enabled then no the P6T does not. Asus still has either a TDP or temp requirement in place in order for Turbo to be enabled.[/QUOTE]

well, I think that is one big vote for the ggbt boards vs the asus?

[SteveRo]

i am at 21 x 2000 @1.425volts with HT off and my max temps are 69 -70
i think our temps are ok as you have higher volts and ht on

Thanks beast - yes I have 4.4 with ht off with max temp of 69 - so that matches up - good deal - maybe I am ok after all!

[icecold316]

My Memory timings are weird. Checkout Case Latency of 4. My memory goes into latency 4 after I resume from sleep. Is it Really 4 or just a error reading it by two software's.

[Vega]

How bad is it to use a QPI/VTT voltage above 1.35V? I have read that Intel dont recommend it!
To use high blck a need 1.4V for stability.

[SteveRo]

How bad is it to use a QPI/VTT voltage above 1.35V? I have read that Intel dont recommend it!
To use high blck a need 1.4V for stability.

time will tell how much wiggle room (if any) there is on these intel max voltages.
if you are prepared to be one of the test cases - go for it - someone has to do it.
if you want to stay under the 1.35 something else to try is to increase (just one, no more than two clicks) qpi pll and (no greater than .5v over qpi v) your memory v.

[SteveRo]

My Memory timings are weird. Checkout Case Latency of 4. My memory goes into latency 4 after I resume from sleep. Is it Really 4 or just a error reading it by two software's.

Most likely something weird going on - in how s/w is coming back from sleep.
any other weird symptoms?

[vikingboy]

appreciate your opinions - am I being uneccessarily worried about stability?

I can run a 200pass of Linx (768MB, 10000) simultaneously with a prime95 blend or lrage calculation for 12 hours. No stability problems.

Moving Linx to a 4096MB test (problem size 23000) I can not even complete two passes - it always fails on the second calculation regardless of how much power I put in to the CP, QPI, mem, slacken timings etc. This is at 19x205.

Am I being too harsh - and can anybody else with an overclocked rig circa 4ghz run a 200 pass on a 4096MB (23k) or greater problem size?

thanks in adv
Ian

PS Congrats on the 100 pages :-)

[SteveRo]

appreciate your opinions - am I being uneccessarily worried about stability?

I can run a 200pass of Linx (768MB, 10000) simultaneously with a prime95 blend or lrage calculation for 12 hours. No stability problems.

Moving Linx to a 4096MB test (problem size 23000) I can not even complete two passes - it always fails on the second calculation regardless of how much power I put in to the CP, QPI, mem, slacken timings etc. This is at 19x205.

Am I being too harsh - and can anybody else with an overclocked rig circa 4ghz run a 200 pass on a 4096MB (23k) or greater problem size?

thanks in adv
Ian

PS Congrats on the 100 pages :-)

My opinion - if you are prime stable 12 hours you are stable enough - I have not used linx yet but unless you see something else unexpected showing up in regular apps I wouldn't worry about it.

[JWMc]

I continue using it in the present state till I am broken

Kazukun is doing it the Xtreme way for us!

[icecold316]

Most likely something weird going on - in how s/w is coming back from sleep.
any other weird symptoms?

Everything runs smooth at Latency 4. I played for around 3-4 hrs of Gaming and no problems. Ran Linx and prime95, no problems at all.

Is there anyone else here able to run the Dominator at Latency4 ?

[SteveRo]

I also posted this question in the ggbt thread -
I have some new memory arriving tomorrow - upgrading from 3x1gb 1066 to 2000.
I am trying to figure out how to go about ocing it based on current best cpu oc, bclock and uncore.
I have a lot of options (965 cpu) but it seems like the uncore speed is going to be my likely limiter.
At 200+ bclock I will want to run the mem mutli at 10 so that means the uncore multi needs to be at 21 right?
But 200x21 is uncore running at 4200 which it seems it will not do.
So questions -
1. I assume cpu cycles are still king and that I do not want to give up on any of my cpu oc (almost 4.3 ht on, and 4.4 ht off).
2. How do I get the uncore speed high enough to run the memory at 2k?

[vikingboy]

Everything runs smooth at Latency 4. I played for around 3-4 hrs of Gaming and no problems. Ran Linx and prime95, no problems at all.

Is there anyone else here able to run the Dominator at Latency4 ?

you could etst by setting latency to 4 in bios but my guess is it would not boot (or post?). I guess its an error although good if it isnt :-)

[286SX]

Another Air Cooled 4.0Ghz OC
System:

• Asus P6T Deluxe
• i7 920 [Batch 3839A606 - that's right, a 3839A worked]
• Corsair XMS 1333 3GB
• Antec 900
• Corsair 650TX
• Cooler Master V8 - knock off Arctic Silver
• Diamond 4870X2 [one core identified as disabled for some reason...]
• 1.5Tb Seagate
• 1.0Tb Seagate
• OEM Optical DVD-RW
• WinXP sp3

1) OC STABLE - 3.31 Ghz @ 1.14v
Prime95 blend stable for 9 hours with stock cooler. MaxTemps were off due to RealTemp 2.7, but read 75/76/75/75. No bios edits except BCLK set to 166.

2) OC STABLE - 3.5 Ghz @ 1.14v
Prime95 blend stable for 30 mins with V8. MaxTemps were off due to RealTemp 2.7, but read 58/58/56/56.

3) OC STABLE - 3.6 Ghz @ 1.14
Prime95 blend stable for 30 mins with V8. Temps with RealTemp 2.9 = 65/64/64/62. 3.6 is easy and stable OC with low voltage.

4) OC UNSTABLE - 3.8 @ 1.16v
Prime95 blend caused reboot under 20 minutes. Voltage too low.

5) OC FAIL - 4.0 Ghz @ 1.20v
Prime95 caused reboot in less than 2 minutes. Only BIOS changes were BLCK@200 and CPUv @ 1.2000v.

6) OC STABLE - 4.0 Ghz @ 1.325v
QPIv = 1.225v. Prime95 blend torture test @ 4.0 Ghz = 7 hours stable. DRAMf = 1223. Everything else was on Auto. I jumped from 1.2 to 1.325, so the temps were a bit high: 97/95/91/90. I had success with low voltage OC in the 3.6-3.8 range, so I am resetting CPUv to 1.3125. See below.

[attachment showing 1.325v removed]

7) HIGHEST STABLE OC - 4.0 Ghz @ 1.3125v
Reduced voltage from #6. Prime95 blend stable for 11+ hours. QPIv = 1.225v. DRAMv = 1.6v. Everything else set on Auto. Max temps are 93/92/89/87.



8) OC UNSTABLE - 4.1 Ghz @ various voltages
1.300-1.325; QPIv 1.2-1.275; PLLv 1.8-1.96. Also randomly set CPU Diff Amplitude to 800 and CPU clock skew to 100ps. Nothing would keep prime stable longer than 20 minutes. Temps were not high, but need to up CPUv to have this work in prime95. Windows boot stable, internet surf stable, CPUMark stable. No other apps tested.

9) OC UNSTABLE - 4.2 Ghz @ noted voltages
CPUv 1.323-1.350; QPIv 1.325-1.35; PLLv 1.90-1.96. As with 4.1Ghz try, also randomly set CPU Diff Amplitude to 800 and CPU clock skew to 100ps. Prime died within 5 minutes. CPUMark only worked at CPUv 1.350; QPIv 1.35; PLLv 1.96. Prime95 reboot in 1 minute. Temps were not high, but need to up CPUv to have this prime95 stable. Windows boot stable at CPUMark OK voltages and internet surf-able. No other apps tested.

Need to figure out how to get 4.1+ stable. I figure it must be something besides voltage since 4.0G works with 1.3125 - and maybe a little less, but I haven't tested 1.3000 or 1.30625 yet.

10) OC STABLE - 4020hz (201 X 20) @ 1.31875
CPUv 1.31875; QPIv 1.225; PLLv A. Prime Small for 8 hours. Yeah, I ran small, and I'm not going back to run blend

11) OC ~STABLE - 4047hz (213 X 19) @ 1.31875
CPUv 1.31875; QPIv 1.30v; PLLv 1.80. LinX 20 stable. Temps 96/96/92/90. Will update with Prime Blend later.

12) OC ~STABLE - 4066hz (214 X 19) @ 1.3125
Reduced CPUv to 3.125; QPIv 1.30v; PLLv 1.80. LinX 20 stable. Temps 97/96/91/90. Will update with Prime Blend later.

13) OC ~STABLE - 4085 (215 x 19) @ 1.3125
CPUv 3.125; QPIv 1.30v; PLLv 1.80. LinX 20 stable. Temps 98/97/93/91. Will update with Prime Blend later.

Looks like I am getting to the top of my cooler's ability. I was able to run 4.104Ghz (216 X 19), but temps got up to 99 in LinX, so I shut it off.

**NOTE** interestingly, in prime blend, temps are 10 degrees cooler than LinX. I may be able to get a higher prime blend stable OC than LinX simply due to temps...

Does that still count??

-td

[ReverendMaynard]

1 second of prime doesn't mean much though Wait until it hits 8k's and watch the temps hit mid 80's.

[icecold316]

Another 4.0Ghz OC


4.0 Ghz @ 1.325v
Jacked up the voltage and set prime95 to torture test @ 4.0 Ghz before I came to work. Will edit with BIOS settings later, but I know I set DRAM Freq at 1200 Mhz to give me the best chance that the system would "live" while I was at work. I think everything else was on Auto.

-td

What is ur QPI Voltage. I guess for Stability of I7, both CPU Voltage and QPI Voltage plays a big factor.

Temp 90 seems too high for 1.325V though.

[SteveRo]

What is ur QPI Voltage. I guess for Stability of I7, both CPU Voltage and QPI Voltage plays a big factor.

Temp 90 seems too high for 1.325V though.

for benching and stress testing i7 will do fine with an occasional max temp of 90. the proc does not throttle until 100c -

[286SX]

1 second of prime doesn't mean much though Wait until it hits 8k's and watch the temps hit mid 80's.

Mid-80s. Bah. Try mid-90s. lol!

In any event, it was prime95 stable (but hot) for 7 hours. 3839A engaged.

-td

[rge]

Since I changed from windows xp 32bit to windows 7 64bit, decided to change my ram from 3x1gb 1600, 8x200 (corsair dom 8,8,8 24) to OCZ 6gb (3x2gb) plat running stock at 7,7,7 21 1600 (8x200). Only change I needed for 4.2ghz prime/linpack stable was increasing qpi/vtt 2 notches from 1.31v to 1.35v.

[SteveRo]

Since I changed from windows xp 32bit to windows 7 64bit, decided to change my ram from 3x1gb 1600, 8x200 (corsair dom 8,8,8 24) to OCZ plat running stock at 7,7,7 21 1600 (8x200). Only change I needed for 4.2ghz prime/linpack stable was increasing qpi/vtt 2 notches from 1.31v to 1.35v.

rge - ok - excellent news for you and maybe me too!
What are all your multi's now.
I will be doing a mem upgrade tomorrow myself.
if you don't mind - I would like very much to try your multi's!

[rge]

I am using qpi/UC/mem of 18/17/8, and of course 21x200. I forgot to add above I am mainly switching from 3gb to 6gb, running at same speed, except OCZ rated 7,7,7 also.

But if you are planning on running 2000mem, I have benched that high with my OCZ 3x1 gb, but of course mine I have to use crazy volts as mine is not rated for 2000 at 1.65v. I had to use something like 1.96DRAM and 1.6qpi/vtt...but I was also running uncore at 22x or 4400. Using 10/20x, which you will probably use if trying to run at stock, dont know what you will need for qpi/vtt. I think my qpi/vtt was not as bad for running 10/20..but I was benching not running prime either.

Actually since you got i965, you can run at 1800, 1900, etc...you should be able to find a sweet spot with good mem speed and not too high qpi/vtt....be interesting to see what you get.

[icecold316]

What is Load Line Calibration ?

I just Disabled that - Now I see my load temperatures is lower by 4c on linx. Is it good to have it disabled ?

My Load Temperature before in CPUZ was 1.248 after disabling it - 1.216 for 3.9 Ghz OC.

[JWMc]

What is Load Line Calibration ?

I just Disabled that - Now I see my load temperatures is lower by 4c on linx. Is it good to have it disabled ?

My Load Temperature before in CPUZ was 1.248 after disabling it - 1.216 for 3.9 Ghz OC.

All it does is increase the vdroop you are seeing when you disable it. If you were on the edge of stability with it on you will definitely crash with it off unless you raise vCore.

[hlonipha]

I posted this over the GIG board thread but here it goes here.

Any ideas here please*****
I can boot and prime @189x20 with turbo enable HT disable for 12hrs. at:
1.31v cpuz idle & 1.296v load, with temps 62c /62c/60c/60c
QPI/VTT 1.18V
x36
x16
x8
rest setting auto.
Now if I change the Bclock to 190+ no boot at all, do not matter if I change multi to 19 turbo off. tried bios f5g, f4,f4r.
Any ideas?????

[namegt]

i7 920 @4.6G 32M PI

BCLK 220
Multi: x21
QPI/VTT : 1.4v
Vdim : 1.8v
Vcore : 1.55v
cpu vtt : 350%
cpu pll 1.8v
qpi pll 1.4 v
ioh core 1.25v
ich 1/0 1.45v
ich core 1.2v





i7 920 [3838B079]