gtxx58 - DDR2 SDRAM - 13022 MByte/sec. MaxxMem Read Bandwidth

[M1RROR]

v3.0 version show higher score than classic V1.95 its some cheating, need to improve rules

[gtxx58]

I redo it if you want !

[gtxx58]

This is simple, beat it with v3.0 :)

[Remarc]

to M1RROR this bench is one big bug and very strange why has not it been removed yet

[TerraRaptor]

Just now, Remarc said:

this bench is one big bug

+100500. I was testing it with Am2 this weekend and get anything from 2500MB/s to 12000MB/s with same settings. No chance to get consistent results at all.

[unityofsaints]

So I'm guessing tRP didn't help the score? ;)

[M1RROR]

I will never use version 3.0, for what?

[gtxx58]

This is just a cpuz bug, trp is at 5

[gtxx58]

For me it’s pretty consistent from 11600 to 13000, if I considered it as a bug, I’m not submitted it, maybe it’s score higher but I don’t think by 1000. If I compare my results with you terraraptor, on the same setting I do about the same.

[TerraRaptor]

your result is quite in line with others (mine included) so noone blames this is cheat or undoable - I even liked it :) I really like your clocks here. 

The bench itself sucks. While superpi gives 0.5% spread from from to run this buggy bench gives me +/-xx% difference with intel and +/-xxx% difference with AMD between runs.

PS.  I decided to skip benching maxmemm2 after review of the following results:

13263 - https://hwbot.org/submission/1020832_ageom_maxxmem_ddr2_sdram_916.6_marks/

13447 - https://hwbot.org/submission/1022472_armannaddafi_maxxmem_ddr2_sdram_936.2_marks/

13092 - https://hwbot.org/submission/939077_endea_maxxmem_ddr2_sdram_788.6_marks/

 

[gtxx58]

I agree with you, on amd side it seem not predictable, the score done with striker, I think it legit, Nvidia chipset is Faster than p45, x48 on memory

[gtxx58]

Maybe other memory bench is more accuracy!

[Noxinite]

Don't forget that apparently if you run the CPU slower then the bench runs for longer and you get a higher memory score. This bench shoud ideally have points stripped from it.

[yosarianilives]

16 minutes ago, Noxinite said:

Don't forget that apparently if you run the CPU slower then the bench runs for longer and you get a higher memory score. This bench shoud ideally have points stripped from it.

Yeah let's drop this buggy mess