Intel Burn Test Problem - Help!

jmp_78 · 25 Apr 2011 at 08:47

All CPUs I've had in the past I have tested using Intel Burn Test (amongst other things) when stressing. I always use the MAXIMUM Stress Level for 50 runs when testing an overclock.

My previous 3 Intel and AMD processors have always reported back with the expected 'Speed (GFlops)' and always take about 30secs/1 minute per 1 test.

However my latest i3 560 @ 3.8GHz takes 3-4 minutes per test?!?

Does RAM affect this? I was sent Tri-Channel memory instead of Dual-Channel by OcUK - could this be an issue?

Azuse05 · 25 Apr 2011 at 08:57

The larger the problem, the more memory it uses, the longer it takes. Time per test isn't relevant, Gflops is what matters.

cjgardens · 25 Apr 2011 at 09:16

If it randomly takes longer it usually means its not quite stable

like if it take 57 seconds at 4ghz then you clock to 4.1ghz and it take way longer like you say i normally give it a bit more volts and it sorts it out

jmp_78 · 25 Apr 2011 at 09:26

It gets about 23 GFlops and this takes about 210secs.

My old Athlon II X2 245 @ 3.6GHz got 21 GFlops, and took 70secs.

This isn't once or twice, I have tested LOADS.

cjgardens · 25 Apr 2011 at 09:36

What do you get stock compared to clocked

seconds and flops

OLDPHART · 25 Apr 2011 at 10:29

Hi, check to see if your memory volts is set to manufactures permited maximum, mine is 1.66v, try putting your CPU core volts up a notch at a time, IBT 20 runs is enough.

jmp_78 · 25 Apr 2011 at 12:38

cjgardens said:
What do you get stock compared to clocked

seconds and flops

Good question, don't remember! Will check shortly.

OLDPHART said:
Hi, check to see if your memory volts is set to manufactures permited maximum, mine is 1.66v, try putting your CPU core volts up a notch at a time, IBT 20 runs is enough.

I've now got the memory running at 1520 MHz with 1.58v. This was stable for 25 runs of IBT Max.

----------------------------

Highest GFlops was 25.1 which I reckon is fine (judging by comparisons), but it still takes ~190secs per run.

Mick-1965 · 25 Apr 2011 at 15:27

jmp_78 said:
It gets about 23 GFlops and this takes about 210secs.

My old Athlon II X2 245 @ 3.6GHz got 21 GFlops, and took 70secs.

This isn't once or twice, I have tested LOADS.

Did both these have the same amount of memory fitted ?

jmp_78 · 25 Apr 2011 at 19:52

Mick-1965 said:
Did both these have the same amount of memory fitted ?

Athlon II had 2GB DDR2, and i3 has 4GB DDR3.

Azuse05 · 25 Apr 2011 at 20:45

Try linx http://www.xtremesystems.org/forums/showthread.php?t=201670 and see what you get then, however make sure you go into the settings and change priority to high before you do any runs.

Edit: actually another thing you can try is using the latest version linx with the latest version of linpack http://www.mediafire.com/?b9e4zg7yg0078o2

Sayso · 25 Apr 2011 at 20:58

jmp_78 said:
Athlon II had 2GB DDR2, and i3 has 4GB DDR3.

More memory takes longer to test.

Mick-1965 · 25 Apr 2011 at 21:15

jmp_78 said:
Athlon II had 2GB DDR2, and i3 has 4GB DDR3.

The more memory used for the test, the calculation size is then larger. Hence taking longer.
The GFlops should stay the same as that is the speed its working out the calculation.

jmp_78 · 25 Apr 2011 at 22:12

I guess I was suprised because of the results from my desktop i5 750 too (all tested multiple times):

i5 750 @ 3.8GHz, 4GB DDR3 @ 1520MHz = ~55 GFlops in ~80 secs
i3 560 @ 3.8GHz, 4GB DDR3 @ 1520MHz = ~25 GFlops in ~190 secs
Athlon II X2 @ 3.6GHz, 2GB DDR2 @ 840MHz = ~21 GFlops in ~70 secs

I'm not saying this is wrong, just trying to understand (and check all is well).

Do you think that the i3 takes twice as long as the Athlon due to double memory?
And in turn, the i5 is twice as quick as the i3 because of the 2 extra pyhsical cores?

I should stop thinking about this! Thanks for all of the posts so far though

C64 · 25 Apr 2011 at 23:13

You should run it with the correct threads set for your cpu not auto and set the ram to the free memory that you have under task manager not avialable ram but free ram.

None of the auto settings are valid.

Mick-1965 · 25 Apr 2011 at 23:32

jmp_78 said:
I guess I was suprised because of the results from my desktop i5 750 too (all tested multiple times):

i5 750 @ 3.8GHz, 4GB DDR3 @ 1520MHz = ~55 GFlops in ~80 secs
i3 560 @ 3.8GHz, 4GB DDR3 @ 1520MHz = ~25 GFlops in ~190 secs
Athlon II X2 @ 3.6GHz, 2GB DDR2 @ 840MHz = ~21 GFlops in ~70 secs

I'm not saying this is wrong, just trying to understand (and check all is well).

Do you think that the i3 takes twice as long as the Athlon due to double memory?
And in turn, the i5 is twice as quick as the i3 because of the 2 extra pyhsical cores?

I should stop thinking about this! Thanks for all of the posts so far though

A very rough analogy for you case. Think of the memory as the distance to travel and the GFlops the max speed you are going.
So in answer to your questions...yes and yes

jmp_78 · 26 Apr 2011 at 08:51

C64 said:
You should run it with the correct threads set for your cpu not auto and set the ram to the free memory that you have under task manager not avialable ram but free ram.

None of the auto settings are valid.

Tried this, but exactly the same results (score and duration). Set the i3 to AUTO threads and then 4 threads.

dazzerd · 27 Apr 2011 at 15:57

I find when I run this I always get different gflps resultsbut I wouldn't worry about it you're only using it to test your overclock.

ps I always assumed that the more ram it used the slower the gflops as it is testing the number to a higher precision but that's just a guess!