yesQuote:
Originally Posted by skirms_fr
then that dissipate the cold temp on all CPU, in watercooling you can remove it, but not under -5 -10°C, if you do it, the bloc can broken the core, and is not better contact on all over the CPU...
Printable View
yesQuote:
Originally Posted by skirms_fr
then that dissipate the cold temp on all CPU, in watercooling you can remove it, but not under -5 -10°C, if you do it, the bloc can broken the core, and is not better contact on all over the CPU...
now cmon ;) 4.1 :banana:
great results being shown here well done all
Cpull spa un PayDay :D
GOod job for your new WR ;)
thx :)Quote:
Originally Posted by PRTeamJapan
done ;)
Do you can do more MHz? ;)
nice Job there cpull, you can pull some more i can feel it ^^
Go for it again ;)
greetz,
Jort
Yeah Baby, we wanna see the 4.1Ghz :toast:
Why do you only use 1.76V???? Do more volts cause instability? :confused:
I have an official 1M WR with the superpi mod 1.4 version at 3814MHz and 14 multiplier
at 3825MHz = cold bug
cpull discover a new cold named cascade :p
Great Job CPULL! :toast:
I'm just waiting to see 20's from you.... ;)
20's will be very very very hard i think
Heh, I know....just wanted to challenge they guy a bit ;)Quote:
Originally Posted by skirms_fr
he s a crazy bencher :p
impossible I think for the moment
the time dont advance
http://membres.lycos.fr/cpulloverclo...que/21.891.gif
I can do this bench around 3740MHz without error
http://membres.lycos.fr/cpulloverclo...ique/32.67.gif
What do you mean? You're able to get the same 1m time @ lower speeds?Quote:
Originally Posted by cpulloverclock
higher i think
I mean I have 24.9 at 3258Quote:
Originally Posted by s7e9h3n
21.9 around 3795
3s for 537MHz!!!!
normally I should have 5.2s for 537MHz
at this speed I see 20.9 at 4GHz
1MB L2 is not enough
You of all people should know that the results won't scale linearly..... ;)Quote:
Originally Posted by cpulloverclock
it was linear until 3260Quote:
Originally Posted by s7e9h3n
after, I dont know
Were you using the same multiplier up to 3260? Then after that, I assume you experimented with diff mult @ the higher speeds? What were the settings on the two different 1m runs you were ciomparing?Quote:
Originally Posted by cpulloverclock
same settings and same frequency ramQuote:
Originally Posted by s7e9h3n
i think you can achieve lower timming with higher mem speeds at the tighest timmings SPI 1mb runs are quite affected by mem timings and speed
hope to see sub 20 with that baby great OC great cpu congratz from venezuela
So if the ram was runnning the same @ 3258 and @ 3795, then that would mean the only thing you're changing is the multiplier. That's the reason why you're times aren't improving linearly. As you probably already know, as your multiplier increases, so does your memory bandwidth efficiency. If for example you were running 12x271.5 = 3258, then your bandwidth efficiency is prob in the range of 92-93%. If you then bench @ 14x271.5 = 3801, your efficiency should increase to ~94-96%. Any benches using the same mult should scale linearly but when you change the mult, then it's gonna be a different story. Of course this is only my opinnion........ ;)Quote:
Originally Posted by cpulloverclock
i think you comment about the 1mb l2 cache is the reason this cpu is stuck.Quote:
Originally Posted by cpulloverclock
if you had 2mb l2 like a pentium dothan core then you would scale better.
________
LovelyWendie99