On Mar 4, 2:26 am, Bug Filler <mozbug...@gmail.com> wrote: > On 3月4日, 上午4時01分, lkcl <luke.leigh...@gmail.com> wrote: > > > *sigh* yes i honestly do not know enough, here, but i am looking at > > perhaps MIPS 1074k which has MIPS-DSP (which has had gcc SIMD > > acceleration since ooo 2006? 2007?), but since i then discovered the > > stunning performance of loongon 3A, and the ICT's plans to put that > > into 28nm, it would be foolish not to look at that as well. > > So you are looking into the future... > > In the hotchips presentation[1], it's said that the CPU core used in > Loongson 2H > and 3B, i.e. GS464V, shall be able to decode H.264 1080P at over > 100FPS.
*whistles*. http://www.hotchips.org/uploads/archive22/HC22.24.630-1-Hu-GS464V-godson.pdf yyyeeee godds. wow. yeah i missed that in the first cursory glance, i didn't realise i'd downloaded that pdf already. i should pay more attention :) single core. 1ghz 1080p. wow. crucial question of course: if the 2G is 3 watts in 65nm @ 1ghz, and has a 64-bit FP unit, what's the 2H going to be like in 28nm @ 1ghz, with a 256-bit SIMD and all the other extensions that make it ROCK! :) 65nm down to 28nm, it's a straight jump to 1/4 of 3 watts - 0.75 watts, then - but all the extra stuff... did someone say that the 16- core version, 28nm, 1ghz, would still be only around 15 watts? that'd make a single-core roughly... yeah, just under 1 watt which would be ... well... damn good would be an understatement. l. -- You received this message because you are subscribed to the Google Groups "loongson-dev" group. To post to this group, send email to loongson-dev@googlegroups.com. To unsubscribe from this group, send email to loongson-dev+unsubscr...@googlegroups.com. For more options, visit this group at http://groups.google.com/group/loongson-dev?hl=en.
