Yesterday I pointed to reporting that The Reg did on a match up between Istanbul and Nehalem on HPL; short story, Istanbul was better price performance.
As reported by Michael Feldman today, Advanced Clustering Technologies who ran the original benchmark has done it again, this time with STREAM. This time Nehalem wins.
Now the engineers at Advanced Clustering Technologies have pitted those same microprocessors against each other using the STREAM benchmark and have posted the results on their Web site. STREAM is part of the HPC Challenge suite and measures sustainable memory bandwidth — one of the most important attributes of high performance computing systems today.
…Even the slowest memory speed on a Xeon 5500 processor bests the fastest produced by the Opteron by as much as 20%; comparing the Opteron to the fastest Xeon, the Xeon outperforms by over 75%. The Xeon 5500 gets these much higher memory bandwidth results because of tri-channel instead of dual-channel memory, the increased clock speed of DDR3 (up to 1333MHz), and the fast point-to-point CPU interconnect provided by its Quick Path Interconnect.
One would probably intuitively expect this (DDR3 vs. DDR2 and more channels to memory), but it’s nice to see the numbers confirm it. And in this case, the Xeons win the price/performance match
When you add cost per machine into the mix, the results still show the Xeon 5500 series with a clear lead. The Xeon machine as configured has a price of approximately $3,800 while the Opteron is priced at $3,500. This gives the Xeon a rate of 9.8 megabytes per second per dollar vs. 5.9 megabytes per second per dollar for the Opteron: a 66% advantage for the Intel Xeon 5500 series.
as usual, intel and amd are trying to stay out of step with each other, so each has some advantage to crow about. overall, intel dominates the crow-ables, which isn’t really surprising. but amd can crow about more cores and (as far as we know) better SMP scalability. have there been any reports of 4 or 8-socket nehalem boxes? IMO, much of the point of istanbul is actually the numa improvement, which applies more to > 2-socket boxes.
in the end, intel is showing the advantages of more/faster ram channels, but also paying the price penalty. it’ll be interesting to see when amd chooses to release ddr3 for opterons – obviously there are some codes where the relatively low BW is hurting them. it’s a bit hard to understand why they’re waiting, since time only lets the impression of intel’s superiority solidify further.