Sign in with
Sign up | Sign in
Your question

Turbo Boost Overclock Problem

Last response: in Overclocking
Share
February 17, 2013 8:38:08 PM

ok here is my problem, i wanted to make a turbo boost overclock, i upped the multiplier cap to 45 on all cores, but when i run prime95 it runs at 4500mhz for 1-2 seconds and it automatically downclocks to 3800mhz im not sure why is this.

i have intel speedstep disabled and my temp goes to 80max on hottest core, i upped my voltage to 1.37v thinking it needed more juice but with no luck

any recommendations are greatly appreciated.

thanks in advance.

edit1: forgot to put my specs
MB: MSI Z77 GD-65
CPU: i7 3770K w H100i Cooler
750w psu
2x 126gb vertex 4 raid 0
Dual GTX 560 Ti in SLI
February 18, 2013 4:03:24 AM

Quote:
Throttling. It can happen for a few reasons.

1. Your motherboard VRM is limiting. With the motherboard that you have, something might be wrong (honestly I wouldn't suspect this).
2. Your overclock is unstable (I would say with 99% certainty this is it)
3. Your temps are too high

Your voltage is pretty high, it's entirely possible that you are hitting 105*C+ immediately when starting prime95 and causing damage to your chip, or at least hitting thermal limits causing it to downclock.

That is WAY too much voltage for a 4.5ghz overclock. You should be able to do 4.5 on like 1.2vcore.

If you want to push the higher voltages on a Ivy Bridge (and the i7-3770k runs hotter than the i5, by the way), you are going to have to delid the chip or get custom water cooling. Delidding is a rather extreme solution, I've delidded 3 ivy bridges but it's not for everyone. It's where you take a razor blade underneath the IHS, pop it off, and rub off the glue.

The reason Ivy Bridge gets so hot is because there is a large gap between the actual CPU die on the chips PCB, and the IHS. On Sandy Bridge, this was okay because they used fluxless solder to attach the IHS to the die, which transmits heat near-perfectly, but on Ivy Bridge they instead used decent thermal paste (okay, but not great, and not nearly as good as fluxless solder).

Wow 1.37v on a non-delidded chip.... you are hurting your chip every time you run prime95. Try being more cautious, that's why you raise voltages incrementally, you don't just max it out right away. If you incrementally went up from 1.15, 1.2, 1.225, 1.25, you'd realize hey, I'm hitting 90*C+ around 1.3, 1.335 vcore, I should be a bit careful when trying to push higher voltages.

The first ivy bridge I played with, even after delidding, I was having issues just like you. I didn't realize it only takes a nanosecond for the CPU to go right up to 105*C, I didn't realize it was getting so hot until I backed down the chip voltage and was seeing temps of 90*C+, because before my system was just crashing on prime95 immediately so I never got to see what the temp value was in HWInfo.

More voltage can often cause more instability, rather than help with stability. You should really try 4.5 at like 1.2v, and then see if you need more voltage. And I doubt your cooling is adequate without delidding the chip or custom water.

There's no need to disable power saving options when overclocking. Also, manually setting your core ratio, is pretty much the same thing as setting X value to all your turbo ratios.

What PSU are you using? A psu can also limit overclocks if the quality is bad (wattage literally means nothing when buying a PSU, I had a 500w PSU cause instability during prime95 and instant crashes, driver crashes most of the time but sometimes system crashes and black screens and resets). The PSu has to have a good voltage regulator to be able to handle the instant switch from low to high load.

And that was a cx500, not some unknown psu, too.

thank you for your well elaboorated response. I think it is something else. my main reason to think that is because I turned on enhanced turbo boost which kind of locks the freq to max turbo boost and when I run prime 95 it stays on 4.5ghz and the temps get around 82 degrees. the reason I have the vcore that hogh is because it was not stable at a lower vcore

the psu is a ocz modxtream pro 700w

any suggestions on how to get a better overclock are fully welcome

thank you very much for your response
m
0
l
February 18, 2013 5:52:59 AM

Quote:
basically do this:

1. Set all settings to stock, and your RAM to their XMP. Run prime95 for 30 hours. Make sure your system actually is good. Set prime95 to priority 10, and go to blend, then hit 'custom' and set your max ram to use to the maximum available ram you have (the stock blend test sets it at 1600, which is only if you have only 2gb of ram, not 16).

If you must use the computer, just pull up task manager, set right click on prime95, set priority to normal and use your computer for light browsing or whatever, then set it back to above normal when you are done (30 hours straight is much better than 6 hours uninterrupted)

2. If your system is straight, set a conservative 4.3ghz@1.3vcore, and run OCCT PSU test. If your system fails, your PSU may be suspect.

3. If all that is good, then it's probably temps. Try the max overclocking frequency you can hit at 1.25vcore, as long as temps are okay. Once you get your stable overclock on ~1.25, you can try 100mhz more and a little more voltage, if it's worth it, and keeping volts in line.

You really can't go above 1.3vcore, ever, on Ivy Bridge, unless you delid the CPU. Otherwise Ivy gets too hot. If you want to push past 1.3, you gotta delid or get custom water cooling, which i wouldn't recommend (better off delidding and accidentally killing a chip and buying a new one then buying water :X)

http://i.imgur.com/DVYzXif.jpg

i think im going to go with the OC Genie overclock for a while until I get a better power supply since I am running dual gtx 560 tis on sli on a 700w psu.

do you still think 83 degrees on hottest core at 4.5ghz 1.37 vcore after 10 mins prime 95 is bad? also I AM running a H100i water cooler
m
0
l
Related resources
a b K Overclocking
February 18, 2013 5:58:15 AM

not really, i run 1.35V 4.83ghz 3770k no problem, i don't get past the 85 Celsius mark with a Noctua NHU12P se2...

Ivy heat a lot but they're good to 105C so it's not a problem until you get about 90C, but you need a good cpu cooler to overclock them for sure...

Run 4.2 stock Vcore, 4.6 @ 1.3V so you should get about that kind of results too...
m
0
l
February 18, 2013 6:03:17 AM

spawnkiller said:
not really, i run 1.35V 4.83ghz 3770k no problem, i don't get past the 85 Celsius mark with a Noctua NHU12P se2...

Ivy heat a lot but they're good to 105C so it's not a problem until you get about 90C, but you need a good cpu cooler to overclock them for sure...

Run 4.2 stock Vcore, 4.6 @ 1.3V so you should get about that kind of results too...

I think I might have something wrong because when I tried I wasnt stable ultil around 1.36vcore. I will stick with OC Genie overclock for a while until I have enough time to test it all out since it is time consuming even having an ssd fast boot time
m
0
l
February 18, 2013 4:27:48 PM

Quote:
Quote:
not really, i run 1.35V 4.83ghz 3770k no problem, i don't get past the 85 Celsius mark with a Noctua NHU12P se2...

Ivy heat a lot but they're good to 105C so it's not a problem until you get about 90C, but you need a good cpu cooler to overclock them for sure...

Run 4.2 stock Vcore, 4.6 @ 1.3V so you should get about that kind of results too...
------------------------------


I find it hard to believe you don't push past 85*C on small fft with a noctua nhu12p.

And the chips have thermal protections at 105. That doesn't mean you want to push the chip to it's thermal case limit. You should consider 85-95*C a maximum temperature, depending on how okay with damaging your chip you are. I would say 85*C is really the max you want to be, but if you spike up to 95*C during 12 hours of testing I would be okay with that as long as it wasn't there for long. It depends on how desperate I was to reach an overclock. If it was running 90*C+ for more than a few seconds than I'd have a problem for sure.

Quote:

I think I might have something wrong because when I tried I wasnt stable ultil around 1.36vcore. I will stick with OC Genie overclock for a while until I have enough time to test it all out since it is time consuming even having an ssd fast boot time


Sounds like either you have a hardware problem (which is quite common, I'd say on average 1-2 components of every computer I make is faulty in some way) or you just have a bad chip (more often than not I've gotten terrible Ivy bridges).

Also, if you are using OC genie to overclock, then no wonder you are unstable. There's a reason people avoid software overclocking, they are always terribly unstable, unreliable, and push extreme voltages for low overclocks. They don't give you a realistic portrayal of your CPU at all.

Like I can boot at 4.5ghz@1.3vcore, whatever, and then use software to tweak my CPU so it's 5ghz@1.3vcore. It'll run just fine, and maybe just crash immediately in prime95. I don't sit there and wonder how do I make 5ghz work, oh i need more voltage, I realize that idling in the computer means nothing and hitting 5ghz like that means nothing.

You are doing things completely backwards it sounds. I have no idea why you think it's a good idea to overclock using the software but there are tons of guides out on the internet on how to overclock and I can't imagine any of them recommended to do what you are doing - pushing extreme voltages on a non-delidded ivy bridge, using software to overclock...


1. I havent uses oc genie yet. but it is not software in windows. it is a part of the msi uefi bios that does a slight overclock to 4.2 ghz

2. like I said earlier. I used such a hogh voltage because I couldnt seem to get it stable at a lower vcore. I has crashes after 10seg on prime 95

3. even at 4.5ghz and 1.37vcore I am not totally sure if it is totallly stable I jist ran prime 95 for 10 mins and it seemed to work fine

4. I learned from a couple of guides that you should chooose the freq you want and then adjust the voltage from up to dawn until its not longer stable, but I never saw needong so mich voltage for a 4.5 oc

so iI might be the ram or psu or a not so good chip. ill test thhe ram later on default freq to see if it is the ram
m
0
l
February 18, 2013 5:51:51 PM

Quote:
1. Oh right, MSI OC Genie, their special little thing. Don't use it. It's terrible.

2. You are crashing most likely because your temps are too high for a non-delidded ivy bridge. You need to back it wayyyy down and see what your max temp is in 5 minutes of small fft, and make sure you stay under 90-95*C, and then raise it up slowly, using small fft to make sure you never go above 95*C as an absolute highest core temp

If you are unstable at 4.5ghz@1.37vcore, then you either have faulty hardware or already damaged your chip. I would say with 99% certainty you have either bad hardware, like something faulty in your hardware, or you've damaged your chip already. I wouldn't be surprised if it's both.

Your chip should also be stable with 4.5@1.35 at the very, very, very most, so something is wrong.

What kind of instability are you getting, exactly? What kind of crashes? Certain kinds of crashes can be indicative of faulty hardware.

3. 10 minutes of prime95 is okay for a quick test but you are going to have to test for 24 hours to really be sure. I hope you are setting it to max ram too.

This is what i do though: For 4gb of RAM, I set custom blend to 2400. That means it uses about 80% of the ram, so if I need to use the computer, I still can open up a tab or two, and then just stop using it and it keeps testing. It's better to test for 24 hours straight and maybe used the pc for a few of those hours, then to not test for 24 hours straight at all.

4. That should probably tell you something then. There's something wrong if you cannot get 4.5ghz stable.

I am pretty sure you got a bad power supply, bad board, bad ram, or already degraded your chip because you are overheating it (i guarantee 1.37vcore on an h100 is overheating on small fft on a non-delidded chip).

You should really run a 24 hour prime95 custom blend test priority 10 on stock settings to make sure you don't have any faulty hardware.

A bad chip means you need 1.3vcore for 4.5ghz. 1.37vcore for 4.5ghz means something is wrong.


I will do that 24hour prime blendpriority 10 with stock settings and then I will post here again. hopefully you can guide me trough

- this is my first build btw, also my first oc.

should I leave ram default or x.m.p.

thanks for all advice
m
0
l
February 19, 2013 9:50:12 PM

Quote:
XMP, as your RAM is rated to run at that speed (I do hope when overclocking you turn off XMP though).

I would recommend you do a PSU test first though, since that's pretty quick and easy to do.


ok so i did cpu stress test and gpu stress test at the same time to test the psu as you said but i have no idea what the numbers mean here is a screenshot

also i dont know why the gpus are throttling, the temperatures i guess



m
0
l
February 20, 2013 2:09:28 AM

Quote:
There's no numbers to make sense of. If your system didn't blue screen, black screen, reset, turn off, have a driver crash, the screen didn't flash, your case speaker didn't beep at you, or the many other things that can be a sign of instability... then you are good.

Prime95 just tells you it's running. The exact numbers tell you what test it's on (8k = cpu intensive, something like 8-50 or some other low number equaling the size of your cpu cache means it's stressing your IMC, and then like large numbers mean it's stressing your rAM). But you really don't need to know that.

You just want to make sure that Prime95 is running constantly when on, instead of saying "HARDWARE FAILURE: WORKER STOPPED" or "Rounding was .6, less than max tolerance of .5".

Which, by the way, make sure you go to 'advanced' and check Sum checking and rounding (you'll know when you look at it, it's very obvious, it's one of those top menu dropdown bars).

As for your GPU throttling, that isn't an issue. That's actually kind of funny that's what you were talking about. Yea sometimes a GPU, instead of being 500mhz, it'll be 510mhz, throttled by 10mhz, to meet the setting of 500mhz. It's just the GPU's way of hitting 500mhz.

So you set your GPU to 597mhz, what the GPU is doing is that it's setting 900mhz, and then throttling itself 303mhz so it runs at 597mhz instead of 900mhz or whatever. You did set it to 597mhz right?


no, both are at the settings of my factory superclocked one, but only one is factory superclocked the other one is normal 2gb.

so the next step is run prime 95 for 24 hours to test if my cpu/board/ram is faulty right?

can you tell me at which settings to run prime95 ?

edit1: also, the gpus are throttling because of the temp right? im not really sure

edit2: i also think the temp on the gpus is so high because i only have 2 intake fans while there is the h100i fans + 2 gpus + back fan on exhaust
m
0
l
February 20, 2013 3:02:43 AM

Quote:
First off, you should always run both of your GPUs at the same frequency.

Secondly, 'factory overclock' is just a very, very minor overclock done to a GPU at the 'factory'. It's basically a way to rip people off and sell overpriced GPUs to dumb people. The 560, as you know, is just the exact same card as the 460, and just like the 460, overclocks generally to 900-1000mhz at a little less than 1.1v.

So what I mean by set your GPU to stock, I mean the reference clock speeds. You can google the 560's actual, stock speed. Running one of your cards at 'factory overclock' and the other at reference speed, is not the right way to do it. ALways set them to same frequencies.

I don't know what you mean by 2GB. I think you are confusing VRAM and frequency or something here. In SLI, you will use the VRAM of the smaller card. So if you have a 1gb and 2gb card, you will only use 1gb of VRAM (yes, technically, VRAM is doubled, but your effective vram doesn't double, because the amount of textures you have to store doubles, as both cards have to save all of the textures to their vram, because they don't share the work, each just renders half the screen).

Yes, run prime95 for 24 hours to make sure everything is fine. I already told you how to run prime95, go to advanced and make sure error checking and sum input are checked off, then select blend, then select custom, so all the blend settings are written in for the custom values, then change the max memory to use to whatever the max ram you have available.

So generally for an 8gb system I use 6600 and for 4gb system I use around 2400. This gives me a little extra space to open task manager, drop the priority of prime95 to normal, and then check my email real quick, and then open task manager back up, set priority back to above normal, and let it keep testing. I mean to be realistic gotta use the computer, right? Ideally you wouldn't touch it at all during testing though, obviously.

Go to worker windows (it's under the far left drop down menu under 'test') and set priority to 10. Seriously just google how to change prime95 priority. What this does is the same thing as when you go to task manager, right click on a task, and set the priority of something to Above Normal.

You know... you can change priorities and affinities of tasks. Like you can set the priority of your call of duty to above normal to make sure your cpu focuses on the game and less on background applications and such, you can set affinities like set your game to cores 0-1 and then your music player to core 3 so that way it never interferes...

Your GPU temps aren't high. As long as you are under 90*C you are fine. And that's being conservative, I think somewhere around 100*C is the actual max...

Once you pass 24 hours then try like 1.25vcore and see the highest frequency you can get with that, and you know, watch temps, raise frequency/voltage as necessary.

Hopefully you have a hardware problem, instead of just a really shitty chip.

I run both gpus at the same frequency and voltage. they are both the 2gb model

do you know why are they throttling when I run furmark?

I will run prime 95 for around 12 hours... do you think that is enough? its just that I need to use the computer when im at home
m
0
l
February 20, 2013 4:28:39 AM

Quote:
Quote:
es, run prime95 for 24 hours to make sure everything is fine. I already told you how to run prime95, go to advanced and make sure error checking and sum input are checked off, then select blend, then select custom, so all the blend settings are written in for the custom values, then change the max memory to use to whatever the max ram you have available.

So generally for an 8gb system I use 6600 and for 4gb system I use around 2400. This gives me a little extra space to open task manager, drop the priority of prime95 to normal, and then check my email real quick, and then open task manager back up, set priority back to above normal, and let it keep testing. I mean to be realistic gotta use the computer, right? Ideally you wouldn't touch it at all during testing though, obviously.


So go to test-> worker windows -> priority 10. When you need to use the computer, open up task manager -> processes, right click on prime95 -> set priority -> Normal. Do your light browsing or whatever, check reddit and forums, then when you are done, close it up, and go back to task manager and set priority to above normal.

You want p95 to use max ram available, but doing that will prevent you from using your computer in any capacity. So instead of setting ~3100 as max ram to use on a 4gb system, I set it as ~2400, so it's testing about 80% of my ram and I can use the computer for a few hours of the day and keep the test running continuously. You can check how much ram you have available, and how much you have available with a few tabs open, with hwinfo, task manager, etc....


first of all thanks for all your support and to be patient with me. im still in high school so there are a lot of things I am not experienced with, and I might ask too many questions. but asking questions helps to learn.

I will do the test and if everything goes well then I will try upping the voltage to 1.25 to see what I can get. I will update here later.
m
0
l
a c 150 K Overclocking
February 20, 2013 5:46:12 AM

Fixed voltage, turn off turbo boost, and set multi to 45x.
m
0
l
February 21, 2013 3:19:54 AM

been running the test for 15 hours until now, will update again in 9 hours

m
0
l
February 21, 2013 4:22:26 AM

Quote:
Wow, see how hot your system is only at stock? I'm pretty sure you were overheating your CPU and you might have degraded it. Needing 4.5@1.37 is definite sign of a damaged chip. Not a bad chip, a damaged one.

Tommorow, assuming you pass, try to see the lowest voltage you can do 24 hours of prime95 again on 4.4ghz. You should be able to hit 4.4 generally around stock voltage ~ 1.25ish or so, not sure what the worst chip on that is, but really shouldnt take much more to hit.


how would I know for sure I damaged it. I might try to exchange it for a new one. maybe I get lucky and get a better one
m
0
l
February 21, 2013 4:50:57 AM

Quote:
You require a lot more voltage than you used to for an overclock. Since your CPU is new and you've been really pushing it with the temps (i'm guaranteeing those instant crashes on prime95 were because you overheated your cpu because you just ramped your voltage so high right off the bat without slowly adding voltage so you could keep track of your temps) than it'll probably just be that you require a ton more voltage than even a really bad chip would need.

So one of the worst chips I had, needed 4.5@1.336vcore, and it failed at the 17th hour. It needed about 1.35 vcore to be stable at 4.5. I suppose 1.37 isn't too far off but you didn't test for 24 hours on that, you said you just were able to at least not instantly crash. I mean I hadn't heard of any chip worse than the one I had, neither of them.

The worst reported chip on OCN's 24/7 stable club is 4.5@1.36. He passed 12 hours (doesnt say if he did 24 hours). I mean it's possible you just have the worst chip in history, I don't know, 4.5@1.37 seems a bit unlikely.

Try to see if you can pass 24 hours on 4.4ghz on the lowest voltage you possibly can, make sure to test temps with small fft after 1.3vcore to make sure you won't overheat, tommorow, after you pass 24 hours.

Frankly, I'd let this initial test go for about 30+ hours just to be 100% certain. I know it's temping to go like "ah 20 hours that's pretty good" but I've had many, many prime95 tests fail after the 21st hour and once or twice it turned out to be faulty hardware.

I was pulling my hair out for about a month once when I tested a chip for 24 hours on stock to make sure it was all straight, I stopped testing at around 22 hours because I figured that was good enough. For a month I was just going crazy, it was really hard to get 3.8, 3.7, stable, and what really made it confusing, was there was no correlation with anything. Like 3.8ghz@1.3vcore was more stable than 3.6ghz@1.5vcore (and temps were low too), in the sense that 3.8@1.3 failed at like 8 hours and 3.6 failed in 15 minutes.

It just went back and forth, like increasing voltage, decreasing voltage, there was no sense to make of any of it. Eventually I was like "Okay, I'll just be happy if I can do 3.5ghz" and even with 1.55vcore I couldn't. Then I started to realize something was screwy, but I mean if a part of the chip was damaged, it wouldn't pass more than 10 minutes of prime95, it'd just crash instantly on everything.

Then I had the idea to underclock my ram, not just run stock 1333mhz, but underclock it, and it worked anjd passed 24 hours on 3.8ghz. Then I wasn't sure if it was the mobo or ram, because that ram ran a high overclock when I had a different CPu, everything else the same, just fine, and walla, it was the board.

Oh, and that CPU had 6 broken pins and I had used copper wire stuck into the motherboard socket, to make it work. So you can imagine what was going through my head every time an overclock failed. It still doesn't really make much sense, I mean I ran the exact same parts for 2 years with a different CPU in it, overclocked, no problem.

Moral of the story - test for 24 hours, at the minimum. 30+ even if you really want to be sure. And you'll waste a lot of time and really be frustrated if everything all comes back to you not doing a surefire, 24 hour test to begin with.

I will let it running until I get home tomorrow which will be around 30 hoirs total test time. since there is no point in stoppping it sonce I am not going to. be home anyways.

I will also change the h100 fans to pull air from the outside of the case. since I installed them wrong because they had no marking to show the way the air flows.

frabkly I want to find out if something is faulty so I can exchange it as soon as possible
m
0
l
a c 150 K Overclocking
February 21, 2013 6:34:04 AM

wax147 said:
I will let it running until I get home tomorrow which will be around 30 hoirs total test time. since there is no point in stoppping it sonce I am not going to. be home anyways.

I will also change the h100 fans to pull air from the outside of the case. since I installed them wrong because they had no marking to show the way the air flows.

frabkly I want to find out if something is faulty so I can exchange it as soon as possible

You need the fans to PUSH air through the radiator.

Fan ^
rad
Fan^


m
0
l
a c 150 K Overclocking
February 21, 2013 6:34:52 AM

Quote:
^ Typical TH post. If it takes 1.37vcore+ just for 10 minutes p95 stable for 4.5, I would not tell the guy "fix voltage, turn off turbo, set multi to 45). I'd tell him something is seriously wrong and he should not do that.

OP doesn't even need to OC. [:rayner428]
m
0
l
February 21, 2013 2:25:40 PM

you just make sure your overclock is stable and vcore are enough ad also disable c1E if you don't it will be hard to overclock or you can use the turbo multiplier method it help save life time too.
m
0
l
February 21, 2013 4:37:27 PM

Quote:


It's fine as long as the fans run air over the h100. I don't see how you can possibly install the fans incorrectly, maybe you put them opposite of what you desire, but either orientation of fans on the h100 would work, either pulling air through the radiator, or pushing air through the radiator.

I'm pretty sure they have markings on them, too.


I was confident it had markings on. I examined them for like 5 minutes and didnt find anything. maybe I didnt notice them. im human after all.

the real temps are around 62 peak on hottest core at default clocks. the reason it peaks to 71~ on the screenshot is because it was on silent mode.

I will get home a in about 3 hours. interested if prime95 crashed or not

my plan now is to do a slightly lower oc would be happy with 4.4ghz at a reasonable temps.

we'll see how it goes

thanks again for the support
m
0
l
February 21, 2013 5:06:39 PM

Quote:


It's fine as long as the fans run air over the h100. I don't see how you can possibly install the fans incorrectly, maybe you put them opposite of what you desire, but either orientation of fans on the h100 would work, either pulling air through the radiator, or pushing air through the radiator.

I'm pretty sure they have markings on them, too.


I was confident it had markings on. I examined them for like 5 minutes and didnt find anything. maybe I didnt notice them. im human after all.

the real temps are around 62 peak on hottest core at default clocks. the reason it peaks to 71~ on the screenshot is because it was on silent mode.

I will get home a in about 3 hours. interested if prime95 crashed or not

my plan now is to do a slightly lower oc would be happy with 4.4ghz at a reasonable temps.

we'll see how it goes

thanks again for the support
m
0
l
February 21, 2013 8:00:35 PM

Quote:


You don't really need to disable c1e, this is just a myth like cpu-gpu bottlenecking and disabling page file, that keeps getting perpetuated because it was true years ago on older hardware but isn't true anymore.

Turbo multiplier overclocking is the same thing as setting a core ratio value if your bios has one.

That's also just a terrible post.


so it didnt fail after 31~ hours. now ill try too see what i can get with 1.25vcore with all default settings

m
0
l
February 21, 2013 8:26:45 PM

Quote:
Can you scroll down to the bottom of each worker on prime95? I can't see the end of worker 1 and 4. When you end the test, each worker should say something like Test Stopped, passed certain number of runs successfully or something. If it didn't, it'll say something like Worker Stopped, hardware failure, rounding error, etc.

Prime95 can run forever and you have had a crash on a worker, meaning instability exists, but the other workers keep going.

If you closed prime95, just check your results.txt (in w7 start-> search "results" and it'll come up, or its in p95 folder) and make sure you had 8 workers all completing the last FFT done (they will be slightly out of sync due to certain workers being faster or quicker, usually because of background applications causing certain workers to do things slower). If you had a hardware failure/worker stop, it will say so. If you have just 7 lines of tests being done for each fft than that clearly means one worker stopped.

Maybe you just have the worst chip in history. So try to see the lowest voltage required to do 4.4ghz, for 24 hours prime95. Set ram to the same settings you had them when you did the 24 hour run (1333 profile disabled or xmp, whatever you had it set to).


i checked the results log and it didnt say anything about hardware failure or worker stopped, the last hardware failure on the results.txt was feb 17th so i dont think it was from last run

maybe you could double check since im not really sure of what im looking for > http://pastebin.com/xNs81sjn

also tried 4.2ghz at 1.25v and it didnt crash also the temps were not so bad, i just tested for 10 mins on small fft but im going to up the freq to see how high can i get with that voltage



edit1: it was large fft to test heat, i testes 4.3ghz at 1.25v and it didnt crash either. clearly i must have been doing something wrong since i dont think it would take .12v to gain just 200mhz

m
0
l
!