AMD fx-6200 low score

[boobteg]

Couple of questions. My score was 3017. My 2d score is much lower then my 3d scrore, dont know why this is. Also, memory has low score even though its latency is only 66. I'm at 4300mhz (21.5 x 200) and core voltage is 1.432. Any lower voltage it fails testing. If I raise my multiplier to 22 to make it 4400mhz, it requires a jump above 1.48v to be stable. If I leave the multiplier at 21.5 and raise fsb to 205 and bump up the cpu NB voltage very little. I get over 4400 mhz stable. But my scores from passmark are worse with any choice other choice then my original 4300mhz. I thought raising the fsb was supposed to increase performance of all aspects, ( with voltage twinkering of course), Any suggestions or tried and trued settings? Oh btw 4500mhz voltage 1.512 BSOD. The highest temperature i reached was 54c so ive got some headroom.

[David (PassMark)]

My 2d score is much lower then my 3d scrore

You can't compare the 2D and 3D score to each other. So having one higher or lower than the other isn't of any significance. The more meaningful comparison is comparing you 3D score to the 3D score from other machines and the same goes for 2D.

memory has low score even though its latency is only 66

Latency only makes up part of the overall memory score.
A latency of 66ns isn't really all that good. (20 to 25ns would be a good value)

There is a list of typical latency value on AMD system here,
http://www.memorybenchmark.net/latency_ddr3_amd.html
and on Intel here,
http://www.memorybenchmark.net/latency_ddr3_intel.html

You are obviously trying to measure very small differences in performance. Which can be hard to do. The margin for error tends to be greater than the difference in performance that you are trying to measure. So make you are doing multiple runs each time and take the max performance as a way of increasing the precision of the measurement.

[boobteg]

This still didnt answer my question on why raising the the fsb instead of the multiplier produced a lower score. I know my cpu score is good its 7288 compared to the 6178 they say it runs on the website under benchmarks. And no not small margains in performance, i dont consider 1.44 volts to 1.482v for 100 mhz a small difference in voltage for stability

[boobteg]

Also suggestions would be handy, like maybe memory timings mine now are factory set at 11-11-11-28-39. Is this producing the high latency? My board only supports 1866mhz ram.

[David (PassMark)]

This still didnt answer my question on why raising the the fsb instead of the multiplier produced a lower score.

Higher voltage doesn't equal higher (or lower) performance. So really the voltage level is irrelevant.

You didn't post the actual results. So I am speculating, based on the limited information, that margin for error is greater than the difference in performance that you are trying to measure.

[David (PassMark)]

Also suggestions would be handy, like maybe memory timings mine now are factory set at 11-11-11-28-39. Is this producing the high latency? My board only supports 1866mhz ram.

The AMD CPU / boards are all high latency. Even if you put in lower latency RAM you won't see significant benefit on an AMD platform. Sorry.

You might be able to get lower wait states by deceasing the clock speed if the only thing that worries you is latency.

[boobteg]

Sorry I attached three photo's from pasmark, don't know why they did not attach. New member here to this forum. Unless for some reason i can see the attached thumbnails at the top if you cant.
Passmark total score was 3,017, Cpu score was 7,288, memory score was 1,283. Thank you for your imput.

[David (PassMark)]

Yes the photos are attached I don't know why you can't see them.
But they don't show the differences between the fast and slow run.

[boobteg]

Could only put three attactchments, i will do another run and post others.

[DigDeep]

This still didnt answer my question on why raising the the fsb instead of the multiplier produced a lower score. I know my cpu score is good its 7288 compared to the 6178 they say it runs on the website under benchmarks. And no not small margains in performance, i dont consider 1.44 volts to 1.482v for 100 mhz a small difference in voltage for stability
the 6178 they say it runs on the website under benchmarks. And no not small margains in performance, i dont consider 1.44 volts to 1.482v for 100 mhz a small difference in voltage for stability

if you raise FSB you also raise the ram, HT and NB freuquency
thats why you are getting higher scores.

[boobteg]

Since this post I replaced my memory with corsair vengeance pro 1600mhz 9.9.9.24 timings. Improved a lot. I think I didnt express correctly what I meant. I tried raising just the multiplier to achieve (X) amount of mhz and benchmarked, then I tried not raising the multiplier and raised the FSB to achieve the same amount of (x) mhz, adjusting the voltage accordingly for ram, HT and cpu nb, but acheived a higher score by just raising the muliplier. I have found out since this post that AMD FX processors do not benefit from ram above 1600mhz, as per the AMD website. So I may have just been hindering my score anyways by raising the FSB and also raising voltages which did not need to be raised to accomodate the higher RAM, ht, and nb. I think if you install RAM of 1600mhz with good timings, with an AMD processor your best bet is just raising the multiplier, not to mention I could acheive stability at a lower vcore by just raising the multiplier vs the FSB. I think really there is not much performance gain from overclocking memory at all, at least the numbers don't show it. Just buy the right memory the first time, that doesn't need overclocking, it solved my problems.