Core i7/X58 Overclocking Thread

[Xello]

How many cores does crysis use?
Can we get more frame rate if we oc just 1 or 2 cores of an i7? (turn off unused cores?)

No more than 4, that goes for any game at the moment though.

I've found with many recent games that claim to be only optimized for 2 cores that i get very good usage on all 4, perhaps the other two are just dealing with windows tasks in the background or something.

Anyway about getting more frames, that's not going to happen from overclocking your cpu if it's already 3.6ghz+. Unless you're playing at 800x600!

[DavidNJ]

As far as I can see I agree, all the same. The key is what cause the issues. Like our old friend GTL, it looks like the QPI is sensitive around 4GHz. Nearly everyone can hit 200 Bclk, but a few drop off by 210, and nearly all by 220.

1.4v will get nearly everyone to 4GHz, that is, no one that high has reported being unable to reach that level. By 4.2GHz the voltages are higher. 21x210 is 4.4GHz and fewer have made it there. All air cooled have temp problems by now, and most waters are getting hot. There could be as much as 350w in the processor by this speed, making the core to water 42C or so with a good water block. Anything but the best radiator will add 5-10C to that above ambient. Those on air are getting core temps in the 80s.

Few have gotten beyond that on air or water. None reporting prime stability. Those on air report temps in the 90sC range. I would imagine nearly all should be capable of 21x200 or 21x205 with adequate cooling.

[Chicken Patty]

As far as I can see I agree, all the same. The key is what cause the issues. Like our old friend GTL, it looks like the QPI is sensitive around 4GHz. Nearly everyone can hit 200 Bclk, but a few drop off by 210, and nearly all by 220.

1.4v will get nearly everyone to 4GHz, that is, no one that high has reported being unable to reach that level. By 4.2GHz the voltages are higher. 21x210 is 4.4GHz and fewer have made it there. All air cooled have temp problems by now, and most waters are getting hot. There could be as much as 350w in the processor by this speed, making the core to water 42C or so with a good water block. Anything but the best radiator will add 5-10C to that above ambient. Those on air are getting core temps in the 80s.

Few have gotten beyond that on air or water. None reporting prime stability. Those on air report temps in the 90sC range. I would imagine nearly all should be capable of 21x200 or 21x205 with adequate cooling.

the most i have ran stable at is the highest i've tried so far 4.2 GHz, HT enabled, 1.411v 200x21

[SteveRo]

i ran at 4.2 GHz stable for about one week until I decided to re overclock the whole thing, i have it saved in my bios though, I can just go and load it, works like a charm. HT Enabled of course, if you want some benches let me know, I can post them

Thanks, but I don't need to see 4.2(HT on) benches - I know they are screaming. But if you would be so kind, perhaps better - post all your settings -
?? x ?? to get to ??
HT on
qpi, uncore and memory multi's
all voltages:
cpu
qpi
dram ...

Thanks!

[Chicken Patty]

Thanks, but I don't need to see 4.2(HT on) benches - I know they are screaming. But if you would be so kind, perhaps better - post all your settings -
?? x ?? to get to ??
HT on
qpi, uncore and memory multi's
all voltages:
cpu
qpi
dram ...

Thanks!

I was just asking

Anyhow, here are my settings. I believe i've posted this a few times already within the last week, here they are anyways, no biggie

200 x 21 all power saving features off, turbo enabled, ht enabled set at 4,800 G/T's

Uncore at x16
Memory at x6 had it low to take RAM stabilization of the equation, but I can run at x8

CPU vcore - 1.35v
CPU VTT - +300
CPU PLL - AUTO
Dimm V - 1.65v
DIMM DQ - +0
QPI PLL - 1.4v
IOH v - 1.4
IOH/ICH I/O - 1.65v
ICH - 1.250

[SteveRo]

I was just asking

Anyhow, here are my settings. I believe i've posted this a few times already within the last week, here they are anyways, no biggie

200 x 21 all power saving features off, turbo enabled, ht enabled set at 4,800 G/T's

CPU vcore - 1.35v
CPU VTT - +300
CPU PLL - AUTO
Dimm V - 1.65v
DIMM DQ - +0
QPI PLL - 1.4v
IOH v - 1.4
IOH/ICH I/O - 1.65v
ICH - 1.250

converting from evga to ggbt, would this equate to =
cpu vtt +300= 1.5+.3 for 1.8V? (I think ggbt is ~1.5)
DIMM DQ - don't know what that is is ggbt speak.
QPI PLL, IOH, IOH/ICH I/O and ICH - I need to check and see what ggbt defaults to for these.

Is your chip an A or B?

[Chicken Patty]

converting from evga to ggbt, would this equate to =
cpu vtt +300= 1.5+.3 for 1.8V? (I think ggbt is ~1.5)
DIMM DQ - don't know what that is is ggbt speak.
QPI PLL, IOH, IOH/ICH I/O and ICH - I need to check and see what ggbt defaults to for these.

Is your chip an A or B?

my chip is 3839A403

Well I Have my BIOS set at 1.35v +300 it equals to 1.411v in windows, if that helps.

[SteveRo]

I am currently into a prime run (3 hrs) at 21x206 (4.326), HT off.
Temps are through the roof as usual - real temp max so far - 93C.
CPU v - 1.5375, qpi v - 1.435, dram - 1.9v.
qpi, uncore and ram multi's at - x36, x14?, x6.

[Chicken Patty]

I am currently into a prime run (3 hrs) at 21x206 (4.326), HT off.
Temps are through the roof as usual - real temp max so far - 93C.
CPU v - 1.5375, qpi v - 1.435, dram - 1.9v.
qpi, uncore and ram multi's at - x36, x14?, x6.

UNCORE I BELIEVE SHOULD BE X16

[Splave]

how do you like the evga chicken

[jaredpace]

No more than 4, that goes for any game at the moment though.

I've found with many recent games that claim to be only optimized for 2 cores that i get very good usage on all 4, perhaps the other two are just dealing with windows tasks in the background or something.

Anyway about getting more frames, that's not going to happen from overclocking your cpu if it's already 3.6ghz+. Unless you're playing at 800x600!

IPC improvements with 45nm and Clockspeed make a much bigger impact on that game's performance than 2 vs. 4 cores

[SteveRo]

UNCORE I BELIEVE SHOULD BE X16

I was at x14, also just rebooted (failed) at just over 3 hours prime.
Why would x16 be better?
I am trying to keep uncore as low as possible to take it out of the equation.

What bios are you using, I am thinking of trying F4m... (oops my bad you are on evga).

[fornowagain]

Why would x16 be better?
I am trying to keep uncore as low as possible to take it out of the equation..

You have x6 mem right? x12 uncore is the min. Maybe he thought you were on x8 mem.

[SteveRo]

You have x6 mem right? x12 uncore is the min. Maybe he thought you were on x8 mem.

I cannot get past the bios at x12 with memory at x6, x13 and above works fine. maybe it is memoryx2<uncore ?

[Hot'n'Chilly]

I am currently into a prime run (3 hrs) at 21x206 (4.326), HT off.
Temps are through the roof as usual - real temp max so far - 93C.
CPU v - 1.5375, qpi v - 1.435, dram - 1.9v.
qpi, uncore and ram multi's at - x36, x14?, x6.

Careful with temps there. 100c is the absolute max temp according to RealTemp and I've heard rumors on other forums that you risk turning your i7 into a burnt little wafer if you ever break 100c.

None the less it is very impressive that you could even Prime95 2 whole hours at something that high (on air even). What are your temps and settings at 4.0ghz? If you can get a stable 4.26ghz I bet you have a heck of a stable 4.0ghz. I'm curious.

[fornowagain]

I cannot get past the bios at x12 with memory at x6, x13 and above works fine. maybe it is memoryx2<uncore ?

This quotes quite handy. eva2000 Intel Core i7 920 Overclocking Introduction Guide on X58

1. VDIMM has to be within 0.5v of Uncore voltage (refered to on some boards as QPI/PLL volts or CPU VTT) for CPU to be safe above >1.65v DRAM memory voltages
2. CPU VTT max 1.375v spec'd some have benched 1.55v so gives VDIMM a 1.875-2.05v range
3. UNCORE (CPUZ NB frequency) needs to optimally be twice that of MEM frequency
4. SOMETIMES for stability, having VCORE within 0.02-0.03v of CPU VTT might stabilise things
5. Yes, theoretical QPI frequency limit is around 4000Mhz but it's too early to say if we can surpass that, some boards and cpu combinations possible could in future.

[Kensek]

1. VDIMM has to be within 0.5v of Uncore voltage (refered to on some boards as QPI/PLL volts or CPU VTT) for CPU to be safe above >1.65v DRAM memory voltages True

2. CPU VTT max 1.375v spec'd some have benched 1.55v so gives VDIMM a 1.875-2.05v range FALSE
3. UNCORE (CPUZ NB frequency) needs to optimally be twice that of MEM frequency TRUE
4. SOMETIMES for stability, having VCORE within 0.02-0.03v of CPU VTT might stabilise things Way FALSE
5. Yes, theoretical QPI frequency limit is around 4000Mhz but it's too early to say if we can surpass that, some boards and cpu combinations possible could in future. Way TRUE

[T_Flight]

Kensek, nice points. I keep saying that, but people are expecting miracles right now. This technology is ***brand new***. It's a miracle it works at all really. That's not a slam, but a compliment to the people who have made it work as well as it does. It's gonna take time to get the BIOS's worked out, the brand new X58 Chipset Drivers worked out, and the brand new ICH10R worked out. There are gonna be new drivers and BIOS's versions for all of these things. ANy one of these things (or a combination of them) could be holding max clocks back.

[Xello]

IPC improvements with 45nm and Clockspeed make a much bigger impact on that game's performance than 2 vs. 4 cores

Like i said, 3.6ghz+ &/ 800x600

[Kensek]

I keep looking to squeeze a little more but I reached my Limit. I moved my Rig to a spare sparsely heated bedroom above the garage. It's winter here in Chicago so I purchased a thermometer to monitor the room temp. Being I'm water cooling I don't want the distilled water to freeze.

http://img156.imageshack.us/img156/7419/thermomlt0.jpg

[Chicken Patty]

how do you like the evga chicken

BRo, I cant be happier, its a great board IMO, clocks well, not too confusing of a BIOS, learning the i7 on any board takes time, but this board has worked flawlessly since I installed it A+

I was at x14, also just rebooted (failed) at just over 3 hours prime.
Why would x16 be better?
I am trying to keep uncore as low as possible to take it out of the equation.

What bios are you using, I am thinking of trying F4m... (oops my bad you are on evga).

Well at least on my board the minimum is 16x, my memory is at 8x anyways so its exactly where it should be. Don't know if your board is different though.

[Hot'n'Chilly]

1. VDIMM has to be within 0.5v of Uncore voltage (refered to on some boards as QPI/PLL volts or CPU VTT) for CPU to be safe above >1.65v DRAM memory voltages True

So on a P6T board you are saying that the QPI Voltage must be within .5v of the DRAM voltage if you want to try and pass the reccomended max DRAM Voltage of 1.65v? Or is the VDIMM voltage not the same thing as DRAM voltage?

[Chicken Patty]

So on a P6T board you are saying that the QPI Voltage must be within .5v of the DRAM voltage if you want to try and pass the reccomended max DRAM Voltage of 1.65v? Or is the VDIMM voltage not the same thing as DRAM voltage?

I find that weird, at least on my board the QPI voltage by default is 1.1v. Right now its running at 1.2V, I would think bringing it up to 1.60v would not be possible

[Hot'n'Chilly]

Well, assuming that your board's QIP Voltage is equivelant to my P6T's I'd say you'd be putting your system in danger by going that high. Mine is at 1.3375v right now on a 3.8ghz OC. I can't imagine needing more than the reccomended max 1.35v (some people argue that 1.45v is the max, but I'm not gonna try it... no way).

[Chicken Patty]

Well, assuming that your board's QIP Voltage is equivelant to my P6T's I'd say you'd be putting your system in danger by going that high. Mine is at 1.3375v right now on a 3.8ghz OC. I can't imagine needing more than the reccomended max 1.35v (some people argue that 1.45v is the max, but I'm not gonna try it... no way).

we are on the same boat here. The only other voltage that gets near that area on my board is the ICH/IOH I/O voltagel, I believe default is 1.5v just like my DRAM voltage.