4800mhz fine, 4900mhz no ?? 4790k

fat4l

New member
So guys, i've finally tweaked my cpu...

4800mhz_1.30V/cpu + 4600mhz_1.31v/cache (1.95V vccin under load)

Super stable, aida64 stability test 7hours.
occt:cpu_large data set = 1hour stable
occt:linpack = 1hour stable
Cinebench loop,xtu bench, aida64 benches, pc and 3d marks..etc..
All this with 70-85C max(delided cpu). No crashes/freezes whatsoever...:cool:

Now, the problem is 4900Mhz. Cant get it stable even with 1.4V. Thats 0.1V increase but only 100mhz gain.. :confused:
Tried vccin 2.02 under load and didnt help.... Can run aida64 stability test for hours, can do XTU for hours too, can do cinebench loops, wprime 1024M etc... But in occt:cpu it crashes to bsod in 5 mins.

Any ideas ? What might help, or should i "idgaf" about occt lol ? Is aida64 nuff?
 
Cache is only for memory speed - and you dont mention that but 1.3 is HIGH

4.8 maybe your wall but youve not mention anything about chipset volts or if the volts are stable in CPUZ when loaded etc. Dont mention BSOD code.... dear lord.

At least have a look around here first and try to help us help you.
 
4.8ghz is pretty good for Devils Canyon and as Tom says that may be your wall. I have seen a few go as high as 5.0 but most of the ones I see people bench top out at 4.8 or lower.
 
if you give us the bsod code "0x00..." then we can usually tell if its voltage and so on..
i would also bench test at 4.5 4.6 4.7 and 4.8 with the same voltages. see if your physics scores start to drop as you get higher.
3dmrk 11 physics test is fine for that. id do 3 runs for each and take an avarage.

as mentioned though its probably your limit. sillicone lottery and it dosent seem like you did too bad.
to me seems like a lot of V and temps for a 24/7 oc.. id probably tweek it down a bit personally. I dont like IBt to get my cpu that hot let alone anything els.
 
Last edited:
;)

The Overclock BSOD Codes are as follows.

0x101 = increase vcore
0x124 = increase/decrease QPI/VTT first, if not increase/decrease vcore
on i7 45nm, usually means too little VVT/QPI for the speed of Uncore
on i7 32nm SB, usually means too little vCore
0x0A = unstable RAM/IMC, increase QPI first, if that doesn't work increase vcore
0x1A = Memory management error. It usually means a bad stick of Ram. Test with Memtest or whatever you prefer. Try raising your Ram voltage
0x1E = increase vcore
0x3B = increase vcore
0x3D = increase vcore
0xD1 = QPI/VTT, increase/decrease as necessary, can also be unstable Ram, raise Ram voltage
0x9C = QPI/VTT most likely, but increasing vcore has helped in some instances
0x50 = RAM timings/Frequency or uncore multi unstable, increase RAM voltage or adjust QPI/VTT, or lower uncore if you're higher than 2x
0x109 = Not enough or too Much memory voltage
0x116 = Low IOH (NB voltage, GPU issue (most common when running multi-GPU/overclocking GPU)
0x7E = Corrupted OS file, possibly from overclocking. Run sfc /scannow and chkdsk /r

BSOD Codes for SandyBridge
0x124 = add/remove vcore or QPI/VTT voltage (usually Vcore, once it was QPI/VTT)
0x101 = add more vcore
0x50 = RAM timings/Frequency add DDR3 voltage or add QPI/VTT
0x1E = add more vcore
0x3B = add more vcore
0xD1 = add QPI/VTT voltage
“0x9C = QPI/VTT most likely, but increasing vcore has helped in some instances”
0X109 = add DDR3 voltage
0x0A = add QPI/VTT voltage

Plus more in the BSOD guide
 
oh cool guys. u rly are being helpful. The only "english" forum where ppl are helpful :) 5/5

im going to test bsod now. im really curios what the cause is :)


So i've tested it and its "0x101 = increase vcore" And it was with 1.375V :( (100MHz + 0.075V increase)
1.Can it be caused by some other Voltage? Like vccin or .....anything else ?


2.Also what do u think about what Tony said, cache voltage being so high ?

3.Also, should I rely on occt:cpu test or is any other stress program enough to test stability? For example, a lot of ppl use Aida64 Stability test and Im sure i can pass with no issues for hours...
 
Last edited:
You can try setting you VCCIN to 1.9v to make sure you have enough. Also if you really want to push the core as far as it can go then drop back on the memory frequency and cache frequency first. You can raise them afterwards.
 
You can try setting you VCCIN to 1.9v to make sure you have enough. Also if you really want to push the core as far as it can go then drop back on the memory frequency and cache frequency first. You can raise them afterwards.
I've tried this and it didnt help.
I set cache to 39, and ram to 1600... No change. I had vccin to 2.02V.... No change. Prolly thats the wall...
 
Well x39 on the cache is still technically an overclock. I would have said x30 and like 1033mhz on the RAM to make sure but to be honest that's only if you are trying to bench the core clock. It's not like you'd ever run it that way. So yea, probably not much you can do unless you want to turn off HT.
 
It is really dependent on the CPU. Run it with the minimum stable voltages. Mine runs at a 4.6GHz 24/7 OC and never crashes. IIRC I have the volts set quite high at 1.35v though because I am a lazy sod who hasn't bothered to fine-tune my stuff.

Get it stable, then start dropping the volts on each and see where instability creeps in. When it does, set the Volts back to the last stable setting and leave it at that. Maybe a small notch above for absolute safety and self-assurance (IMHO).

I lucked out with the silicon lottery though. 5GHz owner here -
7-214.png


Only ran stable for benchmarks, not so much for stress testing. I am pretty sure I could get mine to be stable 5.0 if I could be bothered to tweak and push to maybe around 1.45v because it only crashed when I put it under really heavy load.
 
It is really dependent on the CPU. Run it with the minimum stable voltages. Mine runs at a 4.6GHz 24/7 OC and never crashes. IIRC I have the volts set quite high at 1.35v though because I am a lazy sod who hasn't bothered to fine-tune my stuff.

Get it stable, then start dropping the volts on each and see where instability creeps in. When it does, set the Volts back to the last stable setting and leave it at that. Maybe a small notch above for absolute safety and self-assurance (IMHO).

I lucked out with the silicon lottery though. 5GHz owner here -
7-214.png


Only ran stable for benchmarks, not so much for stress testing. I am pretty sure I could get mine to be stable 5.0 if I could be bothered to tweak and push to maybe around 1.45v because it only crashed when I put it under really heavy load.
well, i can bench with 5 ghz too :) But like u said its not "stress testing" stable ...


For my cache, i tested it already rly nicely.

Cant get 4.7k stable on cache, even with 1.375v(real) cache.
For 4.6k its:
1.250v - Fail
1.275v - Fail after 25 Mins
1.280v - Pass 1H
(all votage above is bios voltage not real)

I set it 1.29v in bios which is 1.31v in real. Super stable...
:)

I just wasnt sure if 1.31v isnt too high as..Tom said.

PS: I thought everyone can bench at 5ghz :D
 
Last edited:
For an everyday overclock upto about 1.25v should be enough on the vRing. Remember that getting a 1:1 ratio between the cache and core clocks isn't important on Haswell because of the increased cache bandwidth. You can happily run the cache 3-500mhz under the core clock without having an impact on CPU performance. 1.3+v on the vRing is benching territory.
 
okay will lower it ;)


EDIT://
working on 4500mhz cache.
1.21v(real) LOCK
1.23v(real) STABLE
1.22v STABLE :)
1.215v LOCK after 38 mins


**Final= 4500mhz cache + 1.22v vring**

I see no point of going +100mhz on cache but with +0.09v...
 
Last edited:
Guys what do u think max vccin is ? on oc.uk forums they are telling me 2.2v is good .. hm

There's no need to have the VCCIN that high. The rule of thumb is vCore + 0.4v. Stock is around 1.8v so you should be able to leave it at stock unless you are seriously throwing volts at it. If anything you should be looking to lower the VCCIN as part of good practice but anything less that stock isn't really important.

I tend to start with mine around 1.9v-2.0 until I've dialled in the overclock and then start reducing it.
 
I've already said what I think. If you want to try adding more then do it. Chips react differently so if you want to add more then go and try it and see if it works for you.
 
Back
Top