• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

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

If you're going to use IBT as a stress tester you need to assign more memory so it is busy for longer periods, 10-80 seconds is hardly enough time for the processor to warm up and then between each run there is about 30-60sec of idle time before the output displays.
 
Too much vcore and the Gflops drop rapidly.
You start out in the 90's and end up at 75 or lower.
Although the clock speed according to cpuz doesn't drop.

I am leaning towards experimenting with FSB and memory clocks to achieve 100 rather than raw clock speed which necessitates too much vcore.

It is interesting to see the performance fall off when you push the chips.
It appears PD doesn't really crash when you push too far but instead slows itself down some how.
 
mmj_uk said:
If you're going to use IBT as a stress tester you need to assign more memory so it is busy for longer periods, 10-80 seconds is hardly enough time for the processor to warm up and then between each run there is about 30-60sec of idle time before the output displays.
Click to expand...

Really i never knew that how informative thank you.

Anyhow point of this exersize is to achieve 100 Gflops over 10 runs.
Doesn't matter how you do it.

Any problem size is acceptable.
Ok standard is not, but high and above is fine, custom whatever
 
Last edited:
PCZ said:
Too much vcore and the Gflops drop rapidly.
You start out in the 90's and end up at 75 or lower.
Although the clock speed according to cpuz doesn't drop.

I am leaning towards experimenting with FSB and memory clocks to achieve 100 rather than raw clock speed which necessitates too much vcore.

It is interesting to see the performance fall off when you push the chips.
It appears PD doesn't really crash when you push too far but instead slows itself down some how.
Click to expand...

That doesn't happen in any other benchmark I've tried, and when encoding, so maybe it's some kind of excessive stress protection like GPUs have for Furmark?

Also, it's worth testing with the old LinPack data to see if it's related.
 
Can you compare these IBT fairly with a 3570k then? the highest I've got mine(3570k) to do was 125 Gflop on 4.7Ghz. I've just done it again on 4.6Ghz my everyday OC and got 120 Gflop. With 8 threads I thought the linpack output of the 8350 would be much higher.
 
HankScorpio said:
Can you compare these IBT fairly with a 3570k then? the highest I've got mine(3570k) to do was 125 Gflop on 4.7Ghz. I've just done it again on 4.6Ghz my everyday OC and got 120 Gflop. With 8 threads I thought the linpack output of the 8350 would be much higher.
Click to expand...

It depends on what Instruction sets it is or isn't using.

Intel Burn Test is an app for Intel CPU's made compatible for AMD CPU's.
 
nkata said:

At 4.8GHz, 1.41V
Click to expand...

Nice and consistent Gflops that time.
Makes you wonder if clocking higher with elevated vcore isn't actually slowing these chips down.

These tests are heavy though.

Hankscorpio
The reults here are not directly comparable to Intel.
IBT for Intel is using AVX these tests are i believe FMA.
 
PCZ said:
Nice and consistent Gflops that time.
Makes you wonder if clocking higher with elevated vcore isn't actually slowing these chips down.

These tests are heavy though.

Hankscorpio
The reults here are not directly comparable to Intel.
IBT for Intel is using AVX these tests are i believe FMA.
Click to expand...


I did get much closer, a few decimal places off 100, then Linpack packed in so was not too stable. Since then I have not got much further. Will give up for tonight. I believe that it is possible though.
 
Yes it's possible :)

100.png
 
Yes you can run lower if you want.
There is no right or wrong way to run them.

You can stick to the 200MHz x multi if you want there BE after all.
Best performance comes with higher FSB though.

You should buy one then you could find out for yourself.
 
PCZ said:
Yes you can run lower if you want.
There is no right or wrong way to run them.

You can stick to the 200MHz x multi if you want there BE after all.
Best performance comes with higher FSB though.

You should buy one then you could find out for yourself.
Click to expand...

I hear you, But, do i spend £150 and upgrade my aging Thuban by a bit, or do i spend £300 on Haswel + MOBO to upgrade my Thuban by a lot, i'm at that indecisive impasse again.

Martini1991 said:
What FSB? ;)

I have never ever been able to get CPUZ to bring up FSB for me, since FSB is long dead.
On AMD it's always reported correctly as HTL.
Click to expand...

"Rated FSB 2528Mhz" i'm just going by what CPU-Z prints.
 
I can see what it says, I've never ever had CPUZ report the FSB, but I've seen plenty of screenshots of it. Are people running some weird version of CPUZ or something?

Your Thuban screenshot says it correctly as HT Link.
 
Martini1991 said:
I can see what it says, I've never ever had CPUZ report the FSB, but I've seen plenty of screenshots of it. Are people running some weird version of CPUZ or something?

Your Thuban screenshot says it correctly as HT Link.
Click to expand...

I know, there is no such thing as a FSB with these chips, it has an integrated memory controller.....

Did it report FSB with Bulldozer or is this just a CPU-Z Piledriver thing?
 
You must log in or register to reply here.
Back
Top Bottom