Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
This makes me angry
- Thread starter The Dreaming
- Start date
- Sort by reaction score
While he does have some nearly valid arguments about the power of the chip, I am disgusted by how he is bashing V-Tech's. That super-cluster handed the asses of all similar cluster to them on a silver platter, and I think that people are just jealous.
<sarcasm>
I love his argument about that $5 million dollars not including the cost of pizza!! I'm sure that the volunteers ate so many MILLIONS OF DOLLARS OF PIZZA that the cluster was no longer economical!!
</sarcasm>
<sarcasm>
I love his argument about that $5 million dollars not including the cost of pizza!! I'm sure that the volunteers ate so many MILLIONS OF DOLLARS OF PIZZA that the cluster was no longer economical!!
</sarcasm>
I thought this quite interesting "Let me remind you that the very notion of Gigaflop was defined specifically for 64-bit precision." I'm not sure where he got this or that he is in a position to remind us. Many companies have been just as happy to report 32-bit numbers when their processors didn't handle 64-bit numbers directly, even those on that otherplatform. 
Is AltiVec's single precision performance good? Yes, better than many. Is it right for every situation? Of course not.
He stated that Apple's 2.0 GHz numbers were below IBM's 1.8 GHz numbers and then makes it seem that they bumped them up for marketing purposes. The SPEC benchmarks have been suspectible to tampering but they're still just benchmarks.
I do think that jealousy prevails a bit with the Virginia Tech. cluster. It took up a lot of space (as noted) and took extra cooling (as with any cluster) and knocked over twice as many PCs in one cluster down a notch.
Is AltiVec's single precision performance good? Yes, better than many. Is it right for every situation? Of course not.He stated that Apple's 2.0 GHz numbers were below IBM's 1.8 GHz numbers and then makes it seem that they bumped them up for marketing purposes. The SPEC benchmarks have been suspectible to tampering but they're still just benchmarks.
I do think that jealousy prevails a bit with the Virginia Tech. cluster. It took up a lot of space (as noted) and took extra cooling (as with any cluster) and knocked over twice as many PCs in one cluster down a notch.
Well once they replace those G5's with Xserves he won't be able to complain about them taking up so much space anymore! And then they'll be rackmounted, and probably cool better then too.
The cluster was rackmounted before, just in really big racks...
And as far as cost goes, I doubt that even including the cost of the building and cooling system would make it more expensive than the #2 cluster, it would have to be a $245 million dollar facility!
And as far as cost goes, I doubt that even including the cost of the building and cooling system would make it more expensive than the #2 cluster, it would have to be a $245 million dollar facility!
