(note: I mistakenly posted this in the wrong section previously - have asked for the other posting to be removed)
Hi folks,
I've just put together my next machine, and decided to give overclocking a go for the first time, just to see what'd happen!
I'd be interested in any comments on my approach, or on my settings - especially whether there is anything i could change that would relatively safely let me take it further, or if I’ve made any mistakes. I've also a few questions as you'll see...
OK, so after a lot of reading around my approach was two fold - first of all see what I can get whilst keeping the temperature to around 70 degrees under a burn-in test (I picked 70 degrees based on comments from various places on the net) and I decided to keep the turbo mode enabled, so that I’d get a max on the speed when necessary, but then when the CPU is not so heavily loaded it could run at a slightly lower speed, and therefore be a little cooler.
First of all the machine specification:
CPU - Intel i7 950
Cooling - Corsair H70 Hydro Series CPU Cooler (mounted as exhaust not as recommended as an intake)
Motherboard - Gigabyte GA-X58A-UD3R Rev 2
Memory - Corsair TwinX 4GB DDR3 1333 (2x2Gb in dual configuration)
PSU - 1000W Coolermaster Silent Pro M, Modular Power Supply
GPU - SLI - 2x Point of View GTX 470 TGT Ultra Charged (3500MHz GDDR5, GPU 680MHz, Shader 1360MHz) (Std 3348MHz GDDR5, GPU 607MHz, Shader 1215MHz)
The settings I've come up with are:
BCLK = 165 at standard 23x/24x multiplier (giving approx 3.8GHz to 4GHz turbo) (reported max speed=3961.09)
VCPU=1.25v (monitoring actual 1.232v)
Memory=9-9-9-24 (I have configured the memory as per the settings on the actual sticks)
VDRAM=1.58v (monitoring actual 1.568v)
QPI Speed=5.94GHz (multipler is set to auto and defaulting to 36x which is the smallest available)
Memory speed=1320 (8x)
Uncore speed=3300 (20x)
HT is enabled
EIST is enabled
Intel turbo boost tech is enabled
These setting run various benchmarking and burn in tools stably with the following temps and results:
Prime95 - 65-70 degree max core temp
Intelliburn test v2.4 - 69 degree max core temp
3d Mark Vantage - score=34,383 CPU=58,834 GPU=30,200 max temp=64 degrees
3d Mark06 - score=26,316 SM2=9,663 SM3=13,588 CPU=6,854 max temp=61 degrees
Other things I've not done yet because I had questions:
I’ve read about is disabling C1E but haven't done this yet as I'm not sure why it would be good to disable a power save function?
I've upped BCLK and not decreased the multiplier - I’ve read this as an alternate approach i.e. lower multiplier and increase BCLK - why is this better?
One final question which has me confused is I’ve been reading about the different temperature sensors (tJuntion/tCase etc). And something I’ve noticed is that (as expected) the core temperatures vary based upon processor loading, idling in the 30's, when loaded as above. However I’ve noticed the CPU sensor stays in the 60s irrespective of loading and i'm unsure why that would be? I read somewhere saying there should be a 5 degree gradient difference between the two, but think I’ve clearly misunderstood something. if anyone can point me to a better description of these I’d be grateful.
I’m running dual channel DDR3 as I had it from my prior machine – will I really see any performance difference in replacing it with triple?
I tried taking the BCLK up another 2 notches (167 rather than 165) to just get it over the 4GHz barrier and found that whilst it would boot, within seconds of starting Intelliburn or Prime it would BSOD. I pushed vcore up 2 notches (from 1.25v to 1.2625v) and this stabilized it for Intelliburn which completed successfully, however Prime still BSOD’s, albeit after about 6 minutes. I was considering putting vcore up a little more as its still pretty low at this point compared to the max specification (I believe this is 1.375?) however the temperatures had jumped by 5-10 degrees and idle and load already so held off. Is there something else I should have looked at?
Last question! Will how much difference does the case make in lowering overall temperatures? I've an Akasa Eclipse 62 case now, which ii'd think is reasonable, but only has one fan in and the H70 out.
Thanks for any comments!
Nick
Hi folks,
I've just put together my next machine, and decided to give overclocking a go for the first time, just to see what'd happen!
I'd be interested in any comments on my approach, or on my settings - especially whether there is anything i could change that would relatively safely let me take it further, or if I’ve made any mistakes. I've also a few questions as you'll see...
OK, so after a lot of reading around my approach was two fold - first of all see what I can get whilst keeping the temperature to around 70 degrees under a burn-in test (I picked 70 degrees based on comments from various places on the net) and I decided to keep the turbo mode enabled, so that I’d get a max on the speed when necessary, but then when the CPU is not so heavily loaded it could run at a slightly lower speed, and therefore be a little cooler.
First of all the machine specification:
CPU - Intel i7 950
Cooling - Corsair H70 Hydro Series CPU Cooler (mounted as exhaust not as recommended as an intake)
Motherboard - Gigabyte GA-X58A-UD3R Rev 2
Memory - Corsair TwinX 4GB DDR3 1333 (2x2Gb in dual configuration)
PSU - 1000W Coolermaster Silent Pro M, Modular Power Supply
GPU - SLI - 2x Point of View GTX 470 TGT Ultra Charged (3500MHz GDDR5, GPU 680MHz, Shader 1360MHz) (Std 3348MHz GDDR5, GPU 607MHz, Shader 1215MHz)
The settings I've come up with are:
BCLK = 165 at standard 23x/24x multiplier (giving approx 3.8GHz to 4GHz turbo) (reported max speed=3961.09)
VCPU=1.25v (monitoring actual 1.232v)
Memory=9-9-9-24 (I have configured the memory as per the settings on the actual sticks)
VDRAM=1.58v (monitoring actual 1.568v)
QPI Speed=5.94GHz (multipler is set to auto and defaulting to 36x which is the smallest available)
Memory speed=1320 (8x)
Uncore speed=3300 (20x)
HT is enabled
EIST is enabled
Intel turbo boost tech is enabled
These setting run various benchmarking and burn in tools stably with the following temps and results:
Prime95 - 65-70 degree max core temp
Intelliburn test v2.4 - 69 degree max core temp
3d Mark Vantage - score=34,383 CPU=58,834 GPU=30,200 max temp=64 degrees
3d Mark06 - score=26,316 SM2=9,663 SM3=13,588 CPU=6,854 max temp=61 degrees
Other things I've not done yet because I had questions:
I’ve read about is disabling C1E but haven't done this yet as I'm not sure why it would be good to disable a power save function?
I've upped BCLK and not decreased the multiplier - I’ve read this as an alternate approach i.e. lower multiplier and increase BCLK - why is this better?
One final question which has me confused is I’ve been reading about the different temperature sensors (tJuntion/tCase etc). And something I’ve noticed is that (as expected) the core temperatures vary based upon processor loading, idling in the 30's, when loaded as above. However I’ve noticed the CPU sensor stays in the 60s irrespective of loading and i'm unsure why that would be? I read somewhere saying there should be a 5 degree gradient difference between the two, but think I’ve clearly misunderstood something. if anyone can point me to a better description of these I’d be grateful.
I’m running dual channel DDR3 as I had it from my prior machine – will I really see any performance difference in replacing it with triple?
I tried taking the BCLK up another 2 notches (167 rather than 165) to just get it over the 4GHz barrier and found that whilst it would boot, within seconds of starting Intelliburn or Prime it would BSOD. I pushed vcore up 2 notches (from 1.25v to 1.2625v) and this stabilized it for Intelliburn which completed successfully, however Prime still BSOD’s, albeit after about 6 minutes. I was considering putting vcore up a little more as its still pretty low at this point compared to the max specification (I believe this is 1.375?) however the temperatures had jumped by 5-10 degrees and idle and load already so held off. Is there something else I should have looked at?
Last question! Will how much difference does the case make in lowering overall temperatures? I've an Akasa Eclipse 62 case now, which ii'd think is reasonable, but only has one fan in and the H70 out.
Thanks for any comments!
Nick
