Posts: 7.414
Threads: 66
Joined: Nov 2008
Reputation:
122
Location: Germany
I've seen one really bad superpi 1M result myself, and then another which was really good.
I fear for AMD if they really just do 23s @ 3.2Ghz... Isn't that lower than PhenomII even?
Posts: 6.069
Threads: 68
Joined: May 2010
Reputation:
167
Location: Grenoble, France
It could also be a hardware bug. I mean you don't do a new revision which cost min 100,000 to 500,000 dollars (could be much more actually) to gain few megahertz. Or it could be a bad design choice.
Posts: 3.559
Threads: 21
Joined: Jul 2010
Reputation:
61
Location: Australia
hmm, the 23s result was actually done with L caches disabled (apparently)
the recent test done by the czech should have had the caches enabled
it wouldn't be impossible to enable the caches if its just a microcode disabling them.