Friday, March 02, 2007
AMD quad core based Petaflops supercomputer in 2008 !
During summer 2005, Cray undertook a months-long evaluation of whether it would continue with AMD or switch to Intel as its strategic processor supplier through 2010. After reviewing both companies' road maps, Cray judged AMD's technically supperior and thus decided to stay put.
Cray's Baker supercomputer upgrade at ORNL in late 2008: Quad core processor used will do 4 flops/cycle/core . Clock is 2.8 Ghz. Thus, pretty slow rise of clock at AMD quad core processors up to the end of 2008. AMD processor used is obviously Shanghai (6MB L3, 45nm) . Don't miss this 320 Tflops machine envisioned, based on 30508 of 2,6Ghz Dual cores.
And, what is supposed performance ratio between 2.8 Ghz Shanghai quad core and that 2,6 Dual core?
Three times !
Just as the difference in size of their caches. ! Cache size is not a measure of processor speed, but might be easily its limiting factor.
But, where is Intel's Petaflops babe? Not still ready? I remember well who first broke the Teraflops limit in 1996. It was Intel. As Cray was the first who broke 1 Gflops limit in 1989.
However, government computing scientists seems are now somewhat reluctant ( I don't know why) on Intel's performance superiority figures based solely on Intel's own benchmarks.
Seems not without justified reason. Perception vs. reality .
Whz the hell performance are so important at all? Well. look at this Japanese 10 petaflops building, that will start running in 5 years, exactly. Its size depends heavily on processor performance,as the system basic buliding block.
"We were serious about switching to Intel if that made more sense, [but] we really like what AMD is doing," said Scott. "We are very happy with the AMD processor cores and systems interfaces. They have been leading Intel for a few years, and we see that likely to continue."
That word comes just as Intel has disclosed its Core microarchitecture in a bid to close the gap with AMD on performance and power. So, AMD PetaFlos computers will come in 2 years. Until then ,Intel will have no CSI nor anything else comparable with AMD processor interbus connection.
Cray's Baker supercomputer upgrade at ORNL in late 2008: Quad core processor used will do 4 flops/cycle/core . Clock is 2.8 Ghz. Thus, pretty slow rise of clock at AMD quad core processors up to the end of 2008. AMD processor used is obviously Shanghai (6MB L3, 45nm) . Don't miss this 320 Tflops machine envisioned, based on 30508 of 2,6Ghz Dual cores.
And, what is supposed performance ratio between 2.8 Ghz Shanghai quad core and that 2,6 Dual core?
Three times !
Just as the difference in size of their caches. ! Cache size is not a measure of processor speed, but might be easily its limiting factor.
But, where is Intel's Petaflops babe? Not still ready? I remember well who first broke the Teraflops limit in 1996. It was Intel. As Cray was the first who broke 1 Gflops limit in 1989.
However, government computing scientists seems are now somewhat reluctant ( I don't know why) on Intel's performance superiority figures based solely on Intel's own benchmarks.
Seems not without justified reason. Perception vs. reality .
Whz the hell performance are so important at all? Well. look at this Japanese 10 petaflops building, that will start running in 5 years, exactly. Its size depends heavily on processor performance,as the system basic buliding block.