-
Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!
SOLVED My 2600k overclock so far, need some help
- Thread starter DaMaGeLIVE
- Start date
- Joined
- Apr 29, 2001
Thanks much as well!
I've started at settings for 4.0GHz. I've noted temps are 55c or less (they were 58c).
I know that using 4.1GHz upward will require much more voltage... sad that I've already ran into a wall, but in Windows, stability seems to drop quickly. (I know this issue is inevitable, but it still seems depressing in a way.)
At 4.3GHz, which I tried very briefly, the BIOS wants 1.360v (yours read 1.355).
DVID, for me, is set to -0.20v.
CPU PLL is 1.790v (default 1.8v)
QPI/vtt is 1.050 as well.
On POST it reads 1.284v, idle in Windows it's 0.96v, and under load it's between 1.19 and 1.20v. (Using 1.20v core forced and LLC5, it would go up to 1.22v on load).
I'll dig into the RAM later (I've 4x4GB of Corsair Vengeance Blue (9-9-9-24. IO Latency 2. I did change the IO timing to 1 and, so far, it's been very stable - which is great. I've kept latency at 37, whereas yours is 33 by default...)
After that, I'll try for 4.2GHz stable, using 1.30v or less (BIOS value; when powering up the PC I don't want to apply any more voltage than necessary).
Apart from the inevitably higher BIOS settings, voltage levels for power received is fantastic.
I've started at settings for 4.0GHz. I've noted temps are 55c or less (they were 58c).
I know that using 4.1GHz upward will require much more voltage... sad that I've already ran into a wall, but in Windows, stability seems to drop quickly. (I know this issue is inevitable, but it still seems depressing in a way.)
At 4.3GHz, which I tried very briefly, the BIOS wants 1.360v (yours read 1.355).
DVID, for me, is set to -0.20v.
CPU PLL is 1.790v (default 1.8v)
QPI/vtt is 1.050 as well.
On POST it reads 1.284v, idle in Windows it's 0.96v, and under load it's between 1.19 and 1.20v. (Using 1.20v core forced and LLC5, it would go up to 1.22v on load).
I'll dig into the RAM later (I've 4x4GB of Corsair Vengeance Blue (9-9-9-24. IO Latency 2. I did change the IO timing to 1 and, so far, it's been very stable - which is great. I've kept latency at 37, whereas yours is 33 by default...)
After that, I'll try for 4.2GHz stable, using 1.30v or less (BIOS value; when powering up the PC I don't want to apply any more voltage than necessary).
Apart from the inevitably higher BIOS settings, voltage levels for power received is fantastic.
I just tried 44x, my F4 BIOS suggested 1.355V IIRC when VCore was on "Normal".
I used a DVID of -0.020 which made it read Vcore 1.320V in BIOS once rebooted. I jumped into Windows and CPU-Z read 1.284V like it had done before at 1.32V but 43x. However, this time while prime95'ing, it jumped back and forth between 1.272 and 1.284 quite often, but 1.284V was seen most often.
I ran prime95 for 30 minutes without problems, then I left the room and played guitar for 30 minutes and when I came back I had a BSOD. Don't know when it happened..
Here are some stills:
Right before I left the room
When I came back
It would be awesome if I could get 44x to work but I don't know what to touch except the VCore. If anyone know how to read BSOD screens than maybe it could help to isolate possible reasons to the crash. If not, 43x will do just fine
I also had a look at that SPD tab in CPU-Z and it said 1.5V for all speeds except the highest supported speed of my memory which is 1600MHz, for that frequency it said 1.65V. I've always had them at 1066MHz but on 1.55 voltage so I just went back to 43x, DVID of -0.015 and set the memory voltage to 1.50V. If it works a night in prime95 I'm going to see if [email protected] works as well.
I also have 4x4GB RAM like you Thermodynamic, btw, are you sure you had DVID @ -0.20 and not -0.020?
Edit reason: Change image host to own server
I used a DVID of -0.020 which made it read Vcore 1.320V in BIOS once rebooted. I jumped into Windows and CPU-Z read 1.284V like it had done before at 1.32V but 43x. However, this time while prime95'ing, it jumped back and forth between 1.272 and 1.284 quite often, but 1.284V was seen most often.
I ran prime95 for 30 minutes without problems, then I left the room and played guitar for 30 minutes and when I came back I had a BSOD. Don't know when it happened..
Here are some stills:
Right before I left the room
When I came back
It would be awesome if I could get 44x to work but I don't know what to touch except the VCore. If anyone know how to read BSOD screens than maybe it could help to isolate possible reasons to the crash. If not, 43x will do just fine
I also had a look at that SPD tab in CPU-Z and it said 1.5V for all speeds except the highest supported speed of my memory which is 1600MHz, for that frequency it said 1.65V. I've always had them at 1066MHz but on 1.55 voltage so I just went back to 43x, DVID of -0.015 and set the memory voltage to 1.50V. If it works a night in prime95 I'm going to see if [email protected] works as well.
I also have 4x4GB RAM like you Thermodynamic, btw, are you sure you had DVID @ -0.20 and not -0.020?
Edit reason: Change image host to own server
Last edited:
- Joined
- Apr 29, 2001
Oops, 0.02v - sorry! 
As for the BSOD - x00000124 = it needs more voltage (or alter DVID to -0.015v from -0.020v). That might resolve the problem. Every x00000124 I've gotten is voltage-related.
As for the BSOD - x00000124 = it needs more voltage (or alter DVID to -0.015v from -0.020v). That might resolve the problem. Every x00000124 I've gotten is voltage-related.
Oops, 0.02v - sorry!
As for the BSOD - x00000124 = it needs more voltage (or alter DVID to -0.015v from -0.020v). That might resolve the problem. Every x00000124 I've gotten is voltage-related.
Thanks!
Focusing on the slightly higher voltage that is shown when in BIOS. Isn't 1.32V still far from damageable for the CPU, I mean, isn't Intel's own max recommended CPU voltage higher, I've read many stuff in forums but most seems to agree that it is around 1.37-1.38 (and that must be BIOS voltage because cpu-z seems more sketchy) for the i7 2600k size CPU's...
Might give the 44x a shot if I only can feel safe going from BIOS 1.320V to 1.325, 1.330 or whatever would be require for it to pass a prime95 and not BSOD after 30-60 minutes... I guess it is not something I can do with the QPI/PLL if the BSOD code complains about cpu core
- Joined
- Apr 29, 2001
I know that maybe 50% of the i2600ks can make it stable at 4.4Ghz...
( http://hardforum.com/showthread.php?t=1578110 )
1.38v is the max safe value, but YMMV. Lower is always better.
I'm sticking with 4.0GHz. 4.1 or higher requires a LOT more voltage than what I've read other people getting away with, and even the values for my 2600k are higher than what was said earlier:
(I see 1.360 for 43, 42 and 42 see 1.345, etc...)
Granted, QTT I've kept at 1.075 and another poster had a screengrab showing 1.085, so that MIGHT be a factor, but it's possible I've just got a poorer-grade 2600 and should be happy 4.0GHz is obtainable as it is...
I've never looked at VRM frequency settings yet, assuming it's in the BIOS. Increasing that is said to help overclocking as well... people say it should be at "350"...
When I ran 4.8GHz just for a quick cinebench test I was at fixed 1.35V. I have tried a lot of stuff and what I found was that whenever I had the RAM at 1600MHz I got BSOD right after I clicked blend test, downclocked to 1300 I could have it running at least a few minutes.
Right now I have it at 1066 but I will see if 1300 works as well at 43x. I don't even know the difference between RAM frequency but my stomach tells me that this is my personal priority list when it comes to RAM:
1. Amount of gigabytes!!!!!
2. Timings
3. Frequency
I've been using a Mac Pro at work since 2008 and that is 800MHz RAM, never complained about anything except that 8 GB is waaay too little for what I do.
If you have the time you could always downclock the RAM frequency to 1066 instead of 1600 and see if it runs stable with the setting krose posted photos of. I think I'd rather run 44x and RAM@1066 than 43x and RAM@1600
Also, I don't use profile RAM in BIOS because it sets the voltage high at 1.65V!! (which is what the corsair xms3 wants at 1600MHz. I also use "Standard" instead of "Turbo" in the RAM BIOS settings...
Right now I have it at 1066 but I will see if 1300 works as well at 43x. I don't even know the difference between RAM frequency but my stomach tells me that this is my personal priority list when it comes to RAM:
1. Amount of gigabytes!!!!!
2. Timings
3. Frequency
I've been using a Mac Pro at work since 2008 and that is 800MHz RAM, never complained about anything except that 8 GB is waaay too little for what I do.
If you have the time you could always downclock the RAM frequency to 1066 instead of 1600 and see if it runs stable with the setting krose posted photos of. I think I'd rather run 44x and RAM@1066 than 43x and RAM@1600
Also, I don't use profile RAM in BIOS because it sets the voltage high at 1.65V!! (which is what the corsair xms3 wants at 1600MHz. I also use "Standard" instead of "Turbo" in the RAM BIOS settings...
- Joined
- Apr 29, 2001
Then I might try underclocking the RAM... it's weird as my RAM is native for 1600MHz, but given Sandy Bridge has the integrated memory controller, no more FSB, et al, CPU speed is far more important than RAM speed...
Thanks much for the info...
What it really comes down to is SB you can't up the BCLK = (FSB)speed so upping the ram speed will not help because you run into SB BCLK speed bottle neck.
But what about lowering the ram frequency, not the timings?
I get BSOD instantly when I run at 1600MHz (4x4GB corsair xms3 1600mhz 1.65V memory). But when I lower them to run at 1066MHz, prime95 blend can go on for hours and hours.
That might as well be RAM voltage related and not depending on the frequency the RAM runs at... I don't feel like giving DRAM voltage of 1.65V on sandy bridge. I guess what I tried to say was that unless you give your RAM sticks the voltage it wants (1.65V for corsair xms3 1600MHz) then try to set Voltage to 1.50 - 1.55 and run them at 1066MHz or 1300MHz, that will increase stability.
Gigabyte GA-Z68XP-UD4 Normal Vcore under different multipliers
Like to say thanks to everyone who has helped out.
In return I'd like to post my boards Normal Vcore so that anyone in the future can see what they are:
Gigabyte Z68XP-UD4
BIOS F4
Normal Vcore - (The Vcore that is shown in BIOS status after rebooting at that multiplier)
DVID was at 0.000 for all multipliers below:
40x = 1.340V - (1.320V)
41x = 1.340V - (1.332V)
42x = 1.340V - (1.332V)
43x = 1.345V - (1.332V)
44x = 1.355V - (1.332V)
45x = 1.360V - (1.344V)
46x = Don't want to try...
I ran prime95 again for a few hours (43x) and it was stable.
I believe I'm not going to spend more time geting 44x to work... isn't worth it.
BIOS photos are on the way, though it's pretty much exactly like Krose posted, except that I have easier/slower settings for RAM.
UPDATE:
Not shown: Bi-Directional PROCHOT: Auto
Cinebench 11.5 scores at 4.3GHz
8.32 - 8.40
Unclocked score: 6.79
CPU-Z CPU Voltage:
Idle: 0.972V
Load: 1.284V
Idle temp: ~28C
Load temp: ~62C (maximum seen is 70C on core 3 during a particular FFT test in prime95 "blend")
Room temp: 21C
Happy overclocking and Merry Christmas!
Like to say thanks to everyone who has helped out.
In return I'd like to post my boards Normal Vcore so that anyone in the future can see what they are:
Gigabyte Z68XP-UD4
BIOS F4
Normal Vcore - (The Vcore that is shown in BIOS status after rebooting at that multiplier)
DVID was at 0.000 for all multipliers below:
40x = 1.340V - (1.320V)
41x = 1.340V - (1.332V)
42x = 1.340V - (1.332V)
43x = 1.345V - (1.332V)
44x = 1.355V - (1.332V)
45x = 1.360V - (1.344V)
46x = Don't want to try...
I ran prime95 again for a few hours (43x) and it was stable.
I believe I'm not going to spend more time geting 44x to work... isn't worth it.
BIOS photos are on the way, though it's pretty much exactly like Krose posted, except that I have easier/slower settings for RAM.
UPDATE:
Not shown: Bi-Directional PROCHOT: Auto
Cinebench 11.5 scores at 4.3GHz
8.32 - 8.40
Unclocked score: 6.79
CPU-Z CPU Voltage:
Idle: 0.972V
Load: 1.284V
Idle temp: ~28C
Load temp: ~62C (maximum seen is 70C on core 3 during a particular FFT test in prime95 "blend")
Room temp: 21C
Happy overclocking and Merry Christmas!
Last edited:
I'm curious to hear if anyone has tried to clock the turbo instead of the multiplier?
I'm not sure if it is possible to do that, but what I was thinking is:
Leave the CPU Clock Ratio at 34X
Enable Intel(R) Turbo Boost Tech.
Then, since we got it stable at 43x on 4 cores... fill in the cores as such:
X -Turbo Ratio(1-core) - 46
X -Turbo Ratio(2-core) - 45
X -Turbo Ratio(3-core) - 44
X -Turbo Ratio(4-core) - 43
Having the CPU Clock Ratio at 43x, is that the same as:
Leaving CPU Clock Ratio default at 34x and setting Turbo Boost to:
X -Turbo Ratio(1-core) - 43
X -Turbo Ratio(2-core) - 43
X -Turbo Ratio(3-core) - 43
X -Turbo Ratio(4-core) - 43
????
I believe that for single core applications, it may be likely that a single core can run steady at higher multipliers, for example 46x.
I'll fiddle with this a bit and write about the progress. Post your thoughts/experience
I'm not sure if it is possible to do that, but what I was thinking is:
Leave the CPU Clock Ratio at 34X
Enable Intel(R) Turbo Boost Tech.
Then, since we got it stable at 43x on 4 cores... fill in the cores as such:
X -Turbo Ratio(1-core) - 46
X -Turbo Ratio(2-core) - 45
X -Turbo Ratio(3-core) - 44
X -Turbo Ratio(4-core) - 43
Having the CPU Clock Ratio at 43x, is that the same as:
Leaving CPU Clock Ratio default at 34x and setting Turbo Boost to:
X -Turbo Ratio(1-core) - 43
X -Turbo Ratio(2-core) - 43
X -Turbo Ratio(3-core) - 43
X -Turbo Ratio(4-core) - 43
????
I believe that for single core applications, it may be likely that a single core can run steady at higher multipliers, for example 46x.
I'll fiddle with this a bit and write about the progress. Post your thoughts/experience
I just tried it as you proposed and it locked up on the Windows logo splash screen. It seemed like a great idea too... I guess it needs more vcore to do 46x during boot, but I'm just going to run it like I was. I will be interested to see how you make out.
I tried:
1core - 46
2core - 45
3core - 44
4core - 43
It booted up into Windows. I don't know if Windows 7 uses 4 cores or only 1?
I ran cinebench a few times and noticed that everytime I clicked the "Run CPU test" button in Cinebench, CPU-Z showed me 4.6GHz for 1 second (while the first render boxes showed up) and then it went to 4.3GHz and stayed there.. I guess it maybe ramp up the cores from 1-4 being used right when you start the test.
It worked fine a few times but the 4th or 5th time I reran the test it frooze, no BSOD, just regular freeze with no mouse movement etc.
I now have the Multi stock at 34x and instead using Turbo Boost Enabled:
1core - 43
2core - 43
3core - 43
4core - 43
Which seems to do EXACTLY the same thing as disabling turbo and having 43x as the CPU Clock Multiplier. I will do more tests later. I guess something like this should be possible without altering the DVID setting.
1core - 45
2core - 44
3core - 43
4core - 43
Or whatever
1core - 46
2core - 45
3core - 44
4core - 43
It booted up into Windows. I don't know if Windows 7 uses 4 cores or only 1?
I ran cinebench a few times and noticed that everytime I clicked the "Run CPU test" button in Cinebench, CPU-Z showed me 4.6GHz for 1 second (while the first render boxes showed up) and then it went to 4.3GHz and stayed there.. I guess it maybe ramp up the cores from 1-4 being used right when you start the test.
It worked fine a few times but the 4th or 5th time I reran the test it frooze, no BSOD, just regular freeze with no mouse movement etc.
I now have the Multi stock at 34x and instead using Turbo Boost Enabled:
1core - 43
2core - 43
3core - 43
4core - 43
Which seems to do EXACTLY the same thing as disabling turbo and having 43x as the CPU Clock Multiplier. I will do more tests later. I guess something like this should be possible without altering the DVID setting.
1core - 45
2core - 44
3core - 43
4core - 43
Or whatever
- Joined
- Nov 26, 2005
- Location
- Concord, NC
What it really comes down to is SB you can't up the BCLK = (FSB)speed so upping the ram speed will not help because you run into SB BCLK speed bottle neck.
What bottleneck?
http://www.xbitlabs.com/articles/memory/display/sandy-bridge-ddr3_4.html
Having applications that can use the ridiculous bandwidth is another story.
Increaisng memory speed wont help because of BCLK bottleneck on SB.What bottleneck?
http://www.xbitlabs.com/articles/memory/display/sandy-bridge-ddr3_4.html
Having applications that can use the ridiculous bandwidth is another story.
If you could increase BCLK speed on SB there would not be a problem with more memory speed.
Similar threads
