**LETS SEE YOUR PILEDRIVER OVERCLOCKS - LET ME START WITH 5GHz+!!**

[Martini1991]

It fails AOD before it fails Prime95, not the other way round, that's what i said and that's the point i'm making, Prime95 is useless for stability testing.

How many hours of Prime95 did your CPU pass?

No way do I believe you crash AOD before Prime.
Only possible explanation would be due to AOD's lack of intensity not making LLC kick up the voltage (Which isn't all that unreasonable, plenty of people have had CPU's stable at load but crap themselves under light load due to a lack of voltage)

AOD was/is nothing compared to Prime in intensity.

Could literally do hours of AOD to a half a second of prime.

 

[humbug]

No way do I believe you crash AOD before Prime.
Only possible explanation would be due to AOD's lack of intensity not making LLC kick up the voltage (Which isn't all that unreasonable, plenty of people have had CPU's stable at load but crap themselves under light load due to a lack of voltage)

AOD was/is nothing compared to Prime in intensity.

Could literally do hours of AOD to a half a second of prime.

Oh so i'm lying?

What an idiotic thing to say. 100% CPU usage not enough to pick up LLC,.... lol

Its a 90's throw back app that only does one thing to the CPU, AOD runs a cycles of all possible aspects of the CPU so of course it will crash before Prime95
its a better stress testing tool than Prime95, application snobbery now as well... get over it....

 

[Chrisoldinho]

Well, i'm irritated!

A few pages ago I showed I was able to hit 4.8 Stable @ 1.4625v, core temp circa 54.
This was tested in AIDA64 and many games such as BF3 never went above 47.
Handbrake is clearly a lot more intensive and was pushing temps as high as 77 (panic stations!).

I have 2 issues here, the first voltage required, and the second temps. I can run 4.8 on this CPU, and pass AIDA no issues, and experience no crashes in any day-day application, but OCCT and Prime95 both fail miserably when I try and run 4.8 @ 1.4625, they both require much closer to 1.5v to be "stable", IBT seems OK at 1.475.

The second issue is temps, again most applications are fine, but Hanbrake notably exposed a weakness here.

I lowered the overclock to 4.6, and found a stable OCCT setting at 1.4375 for nearly an hour before I cancelled the test, I also reduced current capability to 110%, left LLC on Ultra High to counteract vdroop. Core temps are now circa 65 which is still too high for my liking, AIDA now never goes above 51 core temp however (is this even a stress test application?). I imagine it would probably still fail Prime95, as this seems to require higher voltages than all others, not tested this yet.

I can run decent voltages for the overclock I am running at, but the temps are still uncomfortably high for me in stressed scenarios. I am thinking I might need a stronger cooler (H100i or similar) to drive the temps down further, but would prefer not to spend £100 for the benefit of a couple of hundred Mhz. So it seems my CPU is not really voltage limited, rather limited by the capability of my Antec 620.

 

[mmj_uk]

It fails AOD before it fails Prime95, not the other way round, that's what i said and that's the point i'm making, Prime95 is useless for stability testing.

How many hours of Prime95 did your CPU pass?

It will run for as long as I want same with AOD and any other test that's the essence of a stable CPU, although like Chrisoldinho I can only get about 4.2-4.4 without heat being a problem in Prime95 (it still goes over AMD's recommendations which is worrying but it doesn't throttle at least).

Prime95 has always been a thorough test and is optimized to stress AMD unlike a lot of tests, imo if AOD was crashing instead of Prime95 I'd suspect that the extra stress of Prime95 is triggering throttling of some sort which is making it run more stably than AOD.

Load = heat, so when you say AOD runs 5C cooler and crashes you're basically saying is that your CPU is less stable under less stress which is a bit of an anomaly.

Its a 90's throw back app that only does one thing to the CPU, AOD runs a cycles of all possible aspects of the CPU so of course it will crash before Prime95
its a better stress testing tool than Prime95, application snobbery now as well... get over it....

I disagree and the fact that you get 5C lower temps running AOD does too, whatever your issue is you'll have to figure it out but I doubt it's because Prime95 is a bad test compared to AOD.

 

['Wes]

I dislike Prime, i either test it via 3D Mark or intense gaming.

 

[Chrisoldinho]

Prime certainly seems to stress my CPU more than anything else, whether this is necessary will always be open for debate. Until I ran handbrake nothing seemed wrong with my setup @ 4.8, it's only when I began running tests more intensive than AIDA such as handbrake, IBT, OCCT and Prime that my true temps under stress unravelled. It's still comforting to know that temps aside, the voltage I was hitting stable in AIDA at didn't crash handbrake or any other application I use, so if I could get the temps under control I could run a 1/2 way house between Prime stable and OCCT stable and achieve 4.8. I might back down to 4.5 tonight, and accept that my cooler simply isn't up to the job of a "true" stress test at higher clocks.

 

[~>Dg<~]

If it never fails for what a user in their daily use of the computer need it for, then its stable for them.

I don't agree with the premiss that a computer is not stable unless it can Prime95 for x number of hours.
I can run Prime95 for 8 hours solid @ 4.2Ghz with temps just under the threshold, and then run AOD stress testing where it will BSOD in less than 30 seconds (no really)

I turn up the vCore and memory volts up just one notch and AOD will stress indefinably but with core temps about 5c lower than Prime95.
While using all 6 cores at 100% in Handbreak will get me temps 5c lower still.

Prime95 will get the temps higher than anything, that does not mean its stable, nor will it ever get to those temperatures where 90% of users are concerned.

Stability is relative.

thing is you may just do a task that requires some horsepower then bang off goes your pc. who oc there pc just to surf the web?

if you want your pc to be unstable and not really knowing whether its going to crash then so be it but it is daft running it this way to be honest.

still find this thread funny tbh . show us your 5 ghz + thread yet basically nonone has hahahaha great marketing gibbo !

 

[Hagar]

still find this thread funny tbh . show us your 5 ghz + thread yet basically nonone has hahahaha great marketing gibbo !

Agreed ++1.

There are many who are getting 5GHz + but not prime stable. It was a mistake to just do a 5.4 and a very quick bench and say that was a proper overclock. The volts were silly to start with.

Like all the people who bought 2500K in the first few months, banged up a 50x multi and ran superpi. WHOOEE.

I can do 5GHz at 1.5V, run cinebench, x264 FHD and a few other programs, run arma 2 bench etc. It is not a prime stable overclock on 8 workers.

The computer however has not crashed yet running at 5GHz doing all sorts of benches and will run 4 workers prime stable at 4.9GHz.

However day to day I run a 4.6GHz fully prime stable 8 core almost silent solution which I think is very good for the money, has absolutely no problems, 100% reliable and OC's the memory bus brilliantly. It can be tweaked in many more ways than an SB or IB and I think is superior in a lot of other ways.

Also we did not have SATA degredation on the chipset.

my 2 cents worth.

 

[wolvers]

There's loads of people on OC.net with fully stable 5Ghz. No one here seems to be running them with proper water cooling loops.

I could do 5Ghz 24/7 easy with Andy's chip, 1.55v is no problem for me. As it is, I have to be happy with 4.9Ghz 24/7.

Still, I agree the title is a little misleading!

 

[Hagar]

I could do 5Ghz 24/7 easy with Andy's chip, 1.55v is no problem for me. As it is, I have to be happy with 4.9Ghz 24/7.

Still, I agree the title is a little misleading!

I did do 5GHz 24/7 and for quite a while, at less than 1.55V

The issue is not voltage, nor is it power usage. The main issue with eight cores is heat as you imply. With only passive cooling on the motherboard and air cooling for the processor, priming on eight cores at 5GHz does need a proper loop and decent air flow to maintain the heat specification.

It is still a very good processor for the enthusiast who does not mind spending a bit and keeping it cool at 5GHz+. It is also a very good processor for all types of applications at 4.6-4.8GHz with normal air cooling aspirations

 

[Chrisoldinho]

I did do 5GHz 24/7 and for quite a while, at less than 1.55V

The issue is not voltage, nor is it power usage. The main issue with eight cores is heat as you imply. With only passive cooling on the motherboard and air cooling for the processor, priming on eight cores at 5GHz does need a proper loop and decent air flow to maintain the heat specification.

It is still a very good processor for the enthusiast who does not mind spending a bit and keeping it cool at 5GHz+. It is also a very good processor for all types of applications at 4.6-4.8GHz with normal air cooling aspirations

Completely agree with this on the temperature front. Mine is rock solid probably even at 5.0, but the temps are high, so I don't risk it. Even after 10 minutes of OCCT @ 4.6 core temp is circa 65, which frankly isn't good enough. Clearly an Antec 620 isn't a great water cooler.

 

[Chrisoldinho]

Backed down to 4.5 now, Prime finally stable at 1.44375, using a high LLC, so once compensated it comes in at 1.43v in HWMonitor. Happy with the voltages to achieve this tbh and finally, have core temps under control, not gone above 57 now.

If I run 4.6 core temps hitting circa 65, which I am not happy with, clearly my Antec 620 struggles after this.

Next stop will definately have to be a new cooler, either a top end air cooler or something like a H80i/H100i or the new Kraken X40/X60 (although the X60 pricing puts me off).

 

[PCZ]

There's loads of people on OC.net with fully stable 5Ghz. No one here seems to be running them with proper water cooling loops.

If you add the cost of custom water to a BD build it doesn't make sense.
There budget chips.

 

[Hagar]

Completely agree with this on the temperature front. Mine is rock solid probably even at 5.0, but the temps are high, so I don't risk it. Even after 10 minutes of OCCT @ 4.6 core temp is circa 65, which frankly isn't good enough. Clearly an Antec 620 isn't a great water cooler.

My comments on temps at 5GHz were on priming (and OCCT) on eight cores. 24/7 use for me has shown no large temperatures for games or applications.

If you add the cost of custom water to a PD build it doesn't make sense.
There budget chips.

You could say the same for 2500K, 3570K even 2700K but many do use custom water. Personally I would not.

 

[wolvers]

Yeah exactly, I'm not advocating a custom loop specifically for a PD build. Either you have gone to the trouble of adding a custom loop, or not. The platform is irrelevant, and if you change platform then you just change the CPU block (or adapter) just as you would with an air cooler.

 

[Martini1991]

Watercooling in my honest opinion is an investment.
It helps that AMD have kept the same cooling bracket.

Intel have helped somewhat with keeping 1156/1155 the same, 1150 is meant to be the same too.

But with AMD, my custom loop I bought in 09 I could be using now for example.

Obviously when the bracket change is needed, as Wolvers says, you can just get a new adaptor for the bracket, or change CPU block (Which can be expensive if you want top end)

 

[wolvers]

How does a top end air cooler compare in price to a water block? Some blocks don't have interchangeable mounting brackets, I would steer clear of them. TBH, I'm thinking of going back to a core only universal GPU block on my next GPU.

Intel have helped somewhat with keeping 1156/1155 the same, 1150 is meant to be the same too.

So will either of my 775 or 1366 block brackets fit 1150?

 

[Martini1991]

How does a top end air cooler compare in price to a water block? Some blocks don't have interchangeable mounting brackets, I would steer clear of them. TBH, I'm thinking of going back to a core only universal GPU block on my next GPU.



So will either of my 775 or 1366 block brackets fit 1150?

Oh, no sorry.
1156 was the new one, they kept it with 1155, and are meant to be keeping it with 1150.

Some boards however do have the 775 slots for 1156 boards.

For top air coolers, you're looking at say the Silver Arrow at about 55 quid, there's more expensive ones.

But you can put down 55 quid on a top end CPU block, and still spend more if you want.

But then you can also put down sub 50 and get a Raystorm and lose pretty much nothing in the way of cooling performance over a more expensive block.

 

[Jason2]

Do you reckon the Corsair h80 allow me to achieve a 5ghz+ overclock? Just curious.

 

[Martini1991]

Do you reckon the Corsair h80 allow me to achieve a 5ghz+ overclock? Just curious.

No.
The all in ones aren't as good as something like a silver arrow.