Strange...

Strange...

not so strange

see your sub

Completely different. 20000+ is different than 30000+. I just lower the freq so the prog would be slower -> higher bandwith. (according to my tests.)

same done by me, several test until reach it

PS probably (or obviously?) this is not a good benchmark for a division.....

Year 2017

People still wondering about the validity of this benchmark ???

It's proven more than enough , in the past ... that you can manipulate the score of this benchmark.

I believe that it shouldn't even be in hwbot list.

If we needed a last proof that maxmem is eol and should be made pointless, here we go - not your fault, I did chose it on purpose to see if new bugs turn up apart from the 1000 we know already - and yes, it did turn up

His previous score, same frequencies and 1 minutes diference : http://hwbot.org/submission/3588455_

@Niuulh not same frequencies

as said, due to severeal test, lowering only cpu freq (by multiplier) and not ram freq, bandwidth increases;

looking at DDR3 MaxxMem Read Bandwidth's ranking there are a lot of subs with lower cpu freq and high bandwidth

The best, is to redo the bench, to verify and close the debate

Fairplay for ever ^^

I like like you defend yourself, you double your score by downclocking and try to make it legit... Seriously.... You force a run bug and have no shame to sumbit it..... o/

Th problem with this bench is that even if we redo the bench how would we know that e.g the 17000 would be legit and it doesn't mean 15000? We will never know for sure. And as we all know the program has a lot bugs.

Actually it isn't a bug run. By lowering the mhz you give it more time to "pass" mb/s.