Cpu Only Blend test fail

Blend test failure is usually memory. Are you trying to overclock your memory?

If so, put the memory settings back at the stock 1333 MHz, then leave it alone.\

Does small fft's pass or fail?

Small fft always passed with either oc or base memory setting. blend used to pass to but it doesn't for some reason cause I core keeps failing. I disabled all memory oc and set it to 1333mhz but for some reason I STILL can't be stable with blend. small fft is stable though. Could it be because I manually set vccio and vccsa? I put it on base voltage though.

Raising vccio causes it to fail faster with bsod 124...

I don't know. Strange that small fft's pass and blend test fails with stock memory settings.

Does memtest86+ pass?

I'll have to do that. I think I did it when I first bought it. But I don't remember. I'll do it tonight.

No fail. It passed.

I'm assuming we're talking some type of Sandy Bridge K CPU, and if that is the case when you discover the CPUs voltage that will allow the 5.0ghz clock to pass P95 small ffts, with the memory set exactly as specified, regarding DDR3 voltage, speed, and timings, < Very Important!, then run the Blend Test.

At each failure drop the memory speed to the next lower speed, (Leaving the other settings alone drop memory speed only!), rerun the test, continuing to drop the memory speed for each failure, until you discover the memory speed the 5.0ghz CPU clock will stably support.

Once you discover the memory speed that Blend Test does not fail, you can tighten the memory timings and experiment to fine tune the overall performance.

This basically happens because the memory modules were never tested to run the advertised ratings at a 5.0ghz CPU clock, they were only tested and rated at the maximum stock CPU capability range.

It used to be 100% stable though. I think this started to happen after I set vccio and vccsa to manual but I'm not sure. The memory was at default settings but because it failed I dropped it all the way down to 1333mhz. But there is no difference.

Re Read what I just posted you'll have to go lower than 1333mhz.

This is a "been there done that" situation for my response, you don't mention what CPU we're even discussing here, but my discoveries were with a 2500K and the same holds true for a 2600K.

You almost have to be talking a 2500K or 2600K, because I don't know of any other CPU you'll even get to 5.0ghz without refrigerated or extreme cooling, and that certainly wouldn't be a desktop setup you'd be selling to someone.

2500k.
I don't have any lower settings  .

Are you serious?

What motherboard are we talking about?

I'll help if I can, but right now I have a birthday dinner I have to leave for. Ryan

Well there is but I can't boot with them. It always reverts to 1333mhz. I have a z68x-ud3-b3. The lowest setting is 800 and 1066.

Should I try reinstalling windows? Maybe I installed something or something changed. This was working before  

Or could it be because of a bios update? I think I was using F9 originally than I updated to F11.

Try completely hard clearing your CMOS, that should force you to reset your date and time because not having, or missing BIOS memory setting has nothing to do with the operating system.

If it's automatically reverting it sounds like something is still on auto that should be on manual.

You may very well end up dropping your memory speed all the way down to 800mhz to get blend test stability, it all depends on the quality of the memory modules themselves.

By reinstalling windows I meant to resolve the overall issue not the memory. I've read some guides where there was a suggestion to reinstall windows to make sure drivers are not causing the issue. So I will shoot for that.

Hard clearing cmos can be done by...? This is done by taking the battery out right?

Um... someone would have to be pretty thick to buy a PC with such a hugely unsafe OC applied. Those voltages are definitely going to kill the CPU in no time. I've sold a few PCs and most people want long term reliability. I sold a 2600k at 4.6ghz and a 2500k at 4ghz, and other parties have asked for little to no OCing...

Also please post full specs.

Those voltages are fine for a 2500k...

What are the rest of your settings? PLL overvoltage? CPU Spread Spectrum? Energy saving features?

Gigabyte z68x-ud3
i5 2500k
g. skil 2x4gb 1866mhz 1.5v

Settings:
llc: disabled
cpu multi 50x
vcore: normal
offset: +0.165
vccio: 1.05v
vccsa: 0.920v(intel default is 0.925 but gigabyte default is 0.920)
ram: 1.5v

all c-states enabled.

Intel's max is 1.52v. I know that its abnormally high but because Intel says it you can sue them if your chip dies  . Besides I've haven't found ANY person who has had their chip died because of high cpu voltages. Most of the time it was heat or some other voltage. I would like to know one if you do. Either way people buy apple and alienware   , why would they not buy a 5ghz cpu? Many know nothing of voltages  

Reinstalling windows 7-64bit only gave my pc only an extra 30 min to live on the blend test and Prime 95 doesn't fail anymore, my pc just bsods. And now I've seen only bsod 121s and no more bsod 101s. So improvement??

So I kept experimenting with voltages. Raising vccio makes it die faster. Lowering it down to 0.975 had no effect on stability(but raised it back up to default to isolate the variable). Raising pll had no effect. Lowering to 1.50v caused it to bsod on start so I left it at 1.55v which is the lowest v without affecting stability. Vccsa doesn't want to be touched pc has to be forced shut down and enters boot loop until I unplug it. Lowering vcore caused faster death. Raising vcore had absolutely no effect, died 2hrs 30min no matter what, even up to 1.465v. Since all the voltages are acting up on me I lowered multi to 4.9 and I will try to find all the lowest stable voltages than raise the multi to 5.0 again and raise vcore slowly. Hope it works  .

Intel does not say you can sue them if the chip dies. In fact as far as I can find they don't technically say that 1.52V is the max useable, only that it is the highest VID. You may want to read up on what you're doing. For example Vccsa has a very limited range of values - 0.879 min to 0.971 max.

http://www.intel.com/content/www/us/en/processors/core/...

Your Blend test solution is in my first post!

You're playing with voltages you do not even have to mess with!

I 2nd wolfram23's suggestion; "You may want to read up on what you're doing."

This topic has been closed by 4ryan6