-
Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!
How to Overclock Intel xeon x5680 on ASUS P6T Deluxe v1
- Thread starter MuHcOw
- Start date
- Joined
- Oct 28, 2017
Great then!
How high is voltage range for the x5680?
On Asus p6t it changes with colors, I like to keep mine no higher than middle range of purple.. about 1.325volts. by the way, HT will create more heat and requires higher cpu voltage....vcore.
I usually check spot where vootage begins to climb sharply and decide that's where my limit is.
For example this is an example::, if 3.8 needed 1.25 and 4ghz needed 1.4vcore, that's too high and I would settle at like no higher than 3.9ghz.
How high is voltage range for the x5680?On Asus p6t it changes with colors, I like to keep mine no higher than middle range of purple.. about 1.325volts. by the way, HT will create more heat and requires higher cpu voltage....vcore.
I usually check spot where vootage begins to climb sharply and decide that's where my limit is.
For example this is an example::, if 3.8 needed 1.25 and 4ghz needed 1.4vcore, that's too high and I would settle at like no higher than 3.9ghz.
- Joined
- Oct 28, 2017
Found her
View attachment 196224
View attachment 196225
Okay, so she seems like mine in terms of voltage, just higher could multiplier? I'd just use the highest cpu multi there is, set 1.3vcore, 1.3qpi/dram and begin cranking.
Anther thing you want to do is drop qpi link frequency to lowest because overxloxking will cause it to rise to high value and cause you to hit wall because of that. In Asus p6t there are usually 3 settings, choose lowest and start increasing baseclock.
Oh man, this is such a wrong way to overclock.
- - - Updated - - -
BTW, bsod 124 will also be caused by low vcore. I tried lowering vcore and I get same bsod.
View attachment 196224
View attachment 196225
Okay, so she seems like mine in terms of voltage, just higher could multiplier? I'd just use the highest cpu multi there is, set 1.3vcore, 1.3qpi/dram and begin cranking
.Anther thing you want to do is drop qpi link frequency to lowest because overxloxking will cause it to rise to high value and cause you to hit wall because of that. In Asus p6t there are usually 3 settings, choose lowest and start increasing baseclock.
Oh man, this is such a wrong way to overclock.
- - - Updated - - -
BTW, bsod 124 will also be caused by low vcore. I tried lowering vcore and I get same bsod.
OP
- Joined
- Jul 19, 2013
- Thread Starter
- #27
I've read that maximum multiplier is x25 - But I'm able to write x26 in the BIOS???
If I'm to run max multiplier, then I have to drop BCLK a long way down - Is that good or bad? or....?
(edit)
Wrong way to overclock???
Well it's been running super stabile for 12 minutes and 39 seconds - Only 23 hours and 48 minutes to go!!
If I'm to run max multiplier, then I have to drop BCLK a long way down - Is that good or bad? or....?
(edit)
Wrong way to overclock???
Well it's been running super stabile for 12 minutes and 39 seconds - Only 23 hours and 48 minutes to go!!
- Joined
- Oct 28, 2017
Well, it's probably up to 26 because of turbo. Earlier bios only way to have turbo multi is have turboboost enabled, but latest p6t bios, you can now simply choose turbo multi.
Regarding highest multi, I guess you can choose lower. On mine it only goes up to 21, so that's what I chose to overclock, and my board only goes to 191bclk stably, but if you can go higher, then yeah, you can easily drop multi lower and crank that baseclock.
Regarding wrong way ocercloxking, most people recommend taking one step at a time. Find max bclk, then memory multi, then uncore and last one is cpu speed. I can never find enough patience to do it that way because you have to run strss tests on each step and often multiple times, so I just set my frequencies that I'd ideally like to use and see if "my safe volts" can work..then I juggle things bit by bit tweaking to make it work somewhere close to what I wanted to have.
Note: I get punished for doing it this way sometimes because it just won't werk
and I frantically change things only to give up and start the correct way. I usually succeed though..(please no one link to struggles in my posts)
Regarding highest multi, I guess you can choose lower. On mine it only goes up to 21, so that's what I chose to overclock, and my board only goes to 191bclk stably, but if you can go higher, then yeah, you can easily drop multi lower and crank that baseclock.
Regarding wrong way ocercloxking, most people recommend taking one step at a time. Find max bclk, then memory multi, then uncore and last one is cpu speed. I can never find enough patience to do it that way because you have to run strss tests on each step and often multiple times, so I just set my frequencies that I'd ideally like to use and see if "my safe volts" can work..then I juggle things bit by bit tweaking to make it work somewhere close to what I wanted to have.
Note: I get punished for doing it this way sometimes because it just won't werk
and I frantically change things only to give up and start the correct way. I usually succeed though..(please no one link to struggles in my posts)
- Joined
- Oct 28, 2017
Donkey, I've had errors in 16th and 18th hours, but once it rolls over 20 hours I never had anymore eerros, so I just stop at like 21 hour.
BTW, when I'm doing "unknown" configuration overclock, I don't run it all 20 hours, I would run a couple hours and then drop vcore a little to see if she can do another couple hours, or I sometimes increase cpu frequency a bit more and then run few hours again, if she crashes after like 5 minutes, I know I went too far and revert back to previous setting, but if she crashes like after 2 hours, u know I'm in the ballpark and I would just add 2-3 notches of vcore and then let her prime until she either crashes again or I go all the way to 20+ hours mark. Doing it this way I get general idea fairly quick where I am in stability zone.
I was doing 184x21 HT off, with 1.23vcore and I had a feeling it was still more than enough based on tests from long time ago, so I drop it down to 1.18- quick crash...then 1.2 little longer, but crash, 1.208, much longer, but crash, so up 2 notches to 1.218vcore and she went 20 hours without hickup.
You don't have to follow this methology, but just sharing what I was doing. once you begin also iverclocking ram and uncore at the same time, it gets more difficult to figure out what made bsod,
so I up a voltage and see if it changed....if not, I got for other voltage and adjust that instead...if it gives me more runtime, I go little higher and if she gets me even longer, I keep it.
BTW, when I'm doing "unknown" configuration overclock, I don't run it all 20 hours, I would run a couple hours and then drop vcore a little to see if she can do another couple hours, or I sometimes increase cpu frequency a bit more and then run few hours again, if she crashes after like 5 minutes, I know I went too far and revert back to previous setting, but if she crashes like after 2 hours, u know I'm in the ballpark and I would just add 2-3 notches of vcore and then let her prime until she either crashes again or I go all the way to 20+ hours mark. Doing it this way I get general idea fairly quick where I am in stability zone.
I was doing 184x21 HT off, with 1.23vcore and I had a feeling it was still more than enough based on tests from long time ago, so I drop it down to 1.18- quick crash...then 1.2 little longer, but crash, 1.208, much longer, but crash, so up 2 notches to 1.218vcore and she went 20 hours without hickup.
You don't have to follow this methology, but just sharing what I was doing. once you begin also iverclocking ram and uncore at the same time, it gets more difficult to figure out what made bsod,
so I up a voltage and see if it changed....if not, I got for other voltage and adjust that instead...if it gives me more runtime, I go little higher and if she gets me even longer, I keep it.
- Joined
- Oct 28, 2017
Donkey, are you still awake? I did a test. I purposely dropped my known stable vcore by 3 notches(lowered it) and fired up stress test
prime95 small fft
Vs
Occt large data set
My finding shows occt bsod within 2-3 minutes, while prime kept chugging along. (Separate tests after restart)Thought I'd let you know prime small fft may give you false sense of stability, or it may take longer to crash if it's unstable.
For final settings, I like to run prime95 blend setting, but clicking "custom" and inserting maximum free ram I have by looking into task manager. And keep her grinding 20+ hours.
...just not sure choosing small fft will provide quick or at all definitive stability test.
prime95 small fft
Vs
Occt large data set
My finding shows occt bsod within 2-3 minutes, while prime kept chugging along. (Separate tests after restart)Thought I'd let you know prime small fft may give you false sense of stability, or it may take longer to crash if it's unstable.
For final settings, I like to run prime95 blend setting, but clicking "custom" and inserting maximum free ram I have by looking into task manager. And keep her grinding 20+ hours.
...just not sure choosing small fft will provide quick or at all definitive stability test.
OP
- Joined
- Jul 19, 2013
- Thread Starter
- #31
I'm back - 3 hours of sleep surely must be enough 
I'm running Small FFTs to find out if the CPU is getting enough power (not much ram tested) - If small FFTs fail then there's a good possibility that I just have to add more CPU voltage and such.
I MAY BE COMPLETELY WRONG ABOUT THIS !!!
This is where I read about it (http://www.playtool.com/pages/prime95/prime95.html)
I do like Blend test - But it's difficult to tell WHY the test failed, right??
One more thing - Let's say I haven't got the CPU running without errors when reaching 1.35000 Vcore - What do I do, lower BCLK and raise Multiplier? Increase Vcore and hope for the best?? Or should I just accept a lower speed???
(edit)
Vcore is now at 1.35000
QPI is at 1.28125
Multiplier x21
BCLK is still at 210
Prime95 gives me an error after about 20 minutes with "small FFTs"
Peak temperature has been 69°C
What to do?????
I'm running Small FFTs to find out if the CPU is getting enough power (not much ram tested) - If small FFTs fail then there's a good possibility that I just have to add more CPU voltage and such.
I MAY BE COMPLETELY WRONG ABOUT THIS !!!
This is where I read about it (http://www.playtool.com/pages/prime95/prime95.html)
I do like Blend test - But it's difficult to tell WHY the test failed, right??
One more thing - Let's say I haven't got the CPU running without errors when reaching 1.35000 Vcore - What do I do, lower BCLK and raise Multiplier? Increase Vcore and hope for the best?? Or should I just accept a lower speed???
(edit)
Vcore is now at 1.35000
QPI is at 1.28125
Multiplier x21
BCLK is still at 210
Prime95 gives me an error after about 20 minutes with "small FFTs"
Peak temperature has been 69°C
What to do?????
Last edited:
- Joined
- Oct 28, 2017
Usually after 4ghz, there's a heavy vcore required unless chip is very good. Some people say 1.35 vcore is fine, but if she's not stable, probably drop bclk lower or drop multi.
Here's the thing about bclk vs multi. Because ram is tied to bclk and ram multiplier doesn't go low enough, you may not be able to go too much bclk, because in turn it'll drive ram too much frequency....they're tied together. Lowest multi in p6t is 6x, which at 210multi will make ram 1260. I don't see that much of a problem though because is ram even being 1066 can take it with looser timings, but if you have 1333 ram sticks and don't want to undercloxk them, here's what happens:
Next ram multi is 8x, which makes ram go 210x8=1680!! Some 1333 ram can take it, but others will need very loose timings to run it. And another thing, if uncore(integrated memory controller) isn't very good, it will need bigger QPI/DRAM(vtt) voltage bumps. Most people I've read can run 1600 ram though.
Don't forget, uncoee has to be minimum 2x multi of ram, so if ram is 1333, uncore will be 2666...it can go higher than 2x, but not lower.
Another thing with higher bclk is, your QPI link speed, even being at lowest multi will make it still very high and in some cases may not work reliably. My board for example can't stay stable even 196 bclk and 200 won't even boot. Yours looks better because it in boots, but I guess this is where guide is correct about checking max bclk you can run. In your case if you're pusshing it beyond 200, probably good idea to see if it's even stable. In this case, like I mentioned, guide is excellent advice. They say drop everything multis to lowest and see if she's stable.
Now if you don't feel like doing that extra work and want to "jump over that", you have high miltis to play around with. You can replicate similar could frequency by cranking CPU multi and dropping bclk and see if she's good now.
210bclk x21 =4400 MHz cpu
169bclk x26 =4396 fairly similar, which you could try and if it doesn't work then it's definitely too high court frequency because 169 bclk *anyone* can run
Here's the thing about bclk vs multi. Because ram is tied to bclk and ram multiplier doesn't go low enough, you may not be able to go too much bclk, because in turn it'll drive ram too much frequency....they're tied together. Lowest multi in p6t is 6x, which at 210multi will make ram 1260. I don't see that much of a problem though because is ram even being 1066 can take it with looser timings, but if you have 1333 ram sticks and don't want to undercloxk them, here's what happens:
Next ram multi is 8x, which makes ram go 210x8=1680!! Some 1333 ram can take it, but others will need very loose timings to run it. And another thing, if uncore(integrated memory controller) isn't very good, it will need bigger QPI/DRAM(vtt) voltage bumps. Most people I've read can run 1600 ram though.
Don't forget, uncoee has to be minimum 2x multi of ram, so if ram is 1333, uncore will be 2666...it can go higher than 2x, but not lower.
Another thing with higher bclk is, your QPI link speed, even being at lowest multi will make it still very high and in some cases may not work reliably. My board for example can't stay stable even 196 bclk and 200 won't even boot. Yours looks better because it in boots, but I guess this is where guide is correct about checking max bclk you can run. In your case if you're pusshing it beyond 200, probably good idea to see if it's even stable. In this case, like I mentioned, guide is excellent advice. They say drop everything multis to lowest and see if she's stable.
Now if you don't feel like doing that extra work and want to "jump over that", you have high miltis to play around with. You can replicate similar could frequency by cranking CPU multi and dropping bclk and see if she's good now.
210bclk x21 =4400 MHz cpu
169bclk x26 =4396 fairly similar, which you could try and if it doesn't work then it's definitely too high court frequency because 169 bclk *anyone* can run
OP
- Joined
- Jul 19, 2013
- Thread Starter
- #33
I made a spreadsheet
I'm trying to find a sweet spot where DRAM Frequency is pretty close to 1333MHz.
The frequencies in this table do not mach up 100% with what's listed in BIOS - But it's somewhat close.
View attachment Donkeys_OC.txt
And - I have to shut down for tonight
I'll be back Sunday evening.
I'm trying to find a sweet spot where DRAM Frequency is pretty close to 1333MHz.
The frequencies in this table do not mach up 100% with what's listed in BIOS - But it's somewhat close.
View attachment Donkeys_OC.txt
And - I have to shut down for tonight
I'll be back Sunday evening.
Last edited:
- Joined
- Oct 28, 2017
Excellent table! Looks great! BTW, ram will happily work a little faster over advertised speed, you just want to check it t make sure my mushkin silver line 1333 9-9-9-24 works fine at those timings at 1528, so you're not locked down to strictly advertised speed..
Have good weekend!
my mushkin silver line 1333 9-9-9-24 works fine at those timings at 1528, so you're not locked down to strictly advertised speed..Have good weekend!
OP
- Joined
- Jul 19, 2013
- Thread Starter
- #39
Not dead yet
I know I'm doing this the WRONG wrong way
(two wrongs doesn't make a right in this case). I'm putting my CPU at risk here, something I actually said that I wouldn't do!
But the guys at the other forum just made me a LITTLE confident about the CPU and that it would be just fine as long as it has proper cooling.
Coming up on 8 hours now.
But I really can't justify why I'm doing this.
Maybe I should just rethink my strategy and start from scratch.
Wanted to try these:
x21 210 4410 < here now
x22 200 4400
x23 192 4416
x24 184 4416
x25 176 4400
I know I'm doing this the WRONG wrong way
(two wrongs doesn't make a right in this case). I'm putting my CPU at risk here, something I actually said that I wouldn't do!But the guys at the other forum just made me a LITTLE confident about the CPU and that it would be just fine as long as it has proper cooling.
Coming up on 8 hours now.
But I really can't justify why I'm doing this.
Maybe I should just rethink my strategy and start from scratch.
Wanted to try these:
x21 210 4410 < here now
x22 200 4400
x23 192 4416
x24 184 4416
x25 176 4400
Last edited:
- Joined
- Oct 28, 2017
Donkey, ask yourself, if it passes, you're stable at xx frequency with 1.4volts. are you sure you want to keep that voltage? But TBH...t 4.4ghz you may not be able to go much lower anyways
Keep on burning, let us know how it goes (if you want to be at 4.4) perhaps if it passes, you may try dropping it lower, maybe it'll work at 1.35?
What temp??
Keep on burning, let us know how it goes
(if you want to be at 4.4) perhaps if it passes, you may try dropping it lower, maybe it'll work at 1.35?What temp??
Similar threads
