- Oct 14, 2003
- 8,686
- 3,786
- 136
http://www.theregister.co.uk/2009/11/17/sc09_rattner_keynote/page2.html
From a mere barely able to demonstrate functional silicon 2 month ago at Fall IDF this year, Rattner at Intel showed off Larrabee's Floating Point capabilities at SuperComputing 09.
Original - http://www.heise.de/newsticker/meld...rt-Larrabee-mit-ueber-1-Teraflops-862305.html
Translated - http://translate.google.ca/translat...rabee-mit-ueber-1-Teraflops-862305.html&hl=en
80 core - Workstation
64 core - High end consumer GPU
32 core - Mid range consumer GPU
Just speculating...
From a mere barely able to demonstrate functional silicon 2 month ago at Fall IDF this year, Rattner at Intel showed off Larrabee's Floating Point capabilities at SuperComputing 09.
"On the SGEMM single precision, dense matrix multiply test, Rattner showed Larrabee running at a peak of 417 gigaflops with half of its 80 cores activated; and with all of the cores turned on, it was able to hit 805 gigaflops. As the keynote was winding down, Rattner told the techies to overclock it, and was able to push a single Larrabee chip up to just over 1 teraflops, which is the design goal for the initial Larrabee co-processors."
Original - http://www.heise.de/newsticker/meld...rt-Larrabee-mit-ueber-1-Teraflops-862305.html
Translated - http://translate.google.ca/translat...rabee-mit-ueber-1-Teraflops-862305.html&hl=en
At the opening of this year's supercomputing conference in Portland, led Justin Rattner ago from Intel Larrabee one card in the overclocked mode in the matrix multiplication in single precision (SGEMM) the one-teraflops mark slightly surpassed. GPU while waiting in part to compete with much higher theoretical values, but the real measurable SGEMM rate of GPUs is often much less, with Nvidia Tesla C1060 such as about 320 GFlops.
80 core - Workstation
64 core - High end consumer GPU
32 core - Mid range consumer GPU
Just speculating...
Last edited: