Larrabee delayed: anyone know what's happening?
|
Prev: PEEEEEEP
Next: Texture units as a general function
From: nmm1 on 6 Dec 2009 12:39 In article <db0caa7f-6e7f-4fe2-8f99-8e5cb0edf075(a)v37g2000vbb.googlegroups.com>, Michael S <already5chosen(a)yahoo.com> wrote: > >Nick, SCC and Larrabee are different species. Both have plenty of >relatively simple x86 cores on a single chips but that's about only >thing they have in common. > >1. Larrabee cores are cache-coherent, SCC cores are not. >2. Larrabee interconnects have ring topology, SCC is a mesh >3. Larrabee cores are about vector performance (512-bit SIMD) and SMT >(4 hardware threads per core). SCC cores are supposed to be stronger >than Larrabee on scalar code and much much weaker on vector code. Thanks for the correction. .. I have been fully occupied with other matters, and so seem to have missed some developments. Do you have a pointer to any technical information? >4. Larrabee was originally intended for consumers, both as high-end 3D >graphics engine and as sort-of-GPGPU. Graphics as target for 1st >generation chip is canceled, but it still possible that it would be >shipped to paying customers as GPGPU. SCC, on the other hand, is >purely experimental. Now, there I beg to disagree. I have never seen anything reliable indicating that Larrabee has ever been intended for consumers, EXCEPT as a 'black-box' GPU programmed by 'Intel partners'. And some of that information came from semi-authoritative sources in Intel. Do you have a reference to an conflicting statement from someone in Intel? Regards, Nick Maclaren.
From: "Andy "Krazy" Glew" on 7 Dec 2009 09:28 Mayan Moudgill wrote: > All I've come across is the announcement that Larrabee has been delayed, > with the initial consumer version cancelled. Anyone know something more > substantive? I can guess. Part of my guess is that this is related to Pat Gelsinger's departure. Gelsinger was (a) ambitious, intent on becoming Intel CEO (said so in his book), (b) publicly very much behind Larrabee. I'm guessing that Gelsinger was trying to ride Larrabee as his ticket to the next level of executive power. And when Larrabee did not pan out as, Hicc well as he might have liked, he left. And/or conversely: when Gelsinger left, Larrabee lost its biggest executive proponent. Although my guess is that it was technology wagging the executive career tail: no amount of executive positioning can make a technology shippable when it isn't ready. However, I would not count Larrabee out yet. Hiccups happen. Although I remain an advocate of GPU style coherent threading microarchitectures - I think they are likely to be more power efficient than simple MIMD, whether SMT/HT or MCMT - the pull of X86 will be powerful. Eventually we will have X86 MIMD/SMT/HT in-order vs X86 MCMT. Hetero almost guaranteed. Only question will be heteroOOO/lO, or hetero X86 MCMT/GPU. Could be hetero X86 OOO & X86 W. GPU style Coherent Threading. The latter could even be CT/OOO. But these "Could be"s have no sightings.
From: "Andy "Krazy" Glew" on 7 Dec 2009 09:51 nmm1(a)cam.ac.uk wrote: > Now, there I beg to disagree. I have never seen anything reliable > indicating that Larrabee has ever been intended for consumers, > EXCEPT as a 'black-box' GPU programmed by 'Intel partners'. And > some of that information came from semi-authoritative sources in > Intel. Do you have a reference to an conflicting statement from > someone in Intel? http://software.intel.com/en-us/blogs/2008/08/11/siggraph-larrabee-and-the-future-of-computing/ Just a blog, not official, although of course anything blogged at Intel is semi-blest (believe me, I know the flip side.)
From: Del Cecchi on 7 Dec 2009 13:00 "Andy "Krazy" Glew" <ag-news(a)patten-glew.net> wrote in message news:4B1D1685.7080307(a)patten-glew.net... > nmm1(a)cam.ac.uk wrote: > >> Now, there I beg to disagree. I have never seen anything reliable >> indicating that Larrabee has ever been intended for consumers, >> EXCEPT as a 'black-box' GPU programmed by 'Intel partners'. And >> some of that information came from semi-authoritative sources in >> Intel. Do you have a reference to an conflicting statement from >> someone in Intel? > > http://software.intel.com/en-us/blogs/2008/08/11/siggraph-larrabee-and-the-future-of-computing/ > > Just a blog, not official, although of course anything blogged at > Intel is semi-blest (believe me, I know the flip side.) Does this mean Larrabee won't be the engine for the PS4? We were assured that it was not long ago. del
From: Robert Myers on 7 Dec 2009 13:25
On Dec 7, 9:51 am, "Andy \"Krazy\" Glew" <ag-n...(a)patten-glew.net> wrote: > n...(a)cam.ac.uk wrote: > > Now, there I beg to disagree. I have never seen anything reliable > > indicating that Larrabee has ever been intended for consumers, > > EXCEPT as a 'black-box' GPU programmed by 'Intel partners'. And > > some of that information came from semi-authoritative sources in > > Intel. Do you have a reference to an conflicting statement from > > someone in Intel? > > http://software.intel.com/en-us/blogs/2008/08/11/siggraph-larrabee-an... > > Just a blog, not official, although of course anything blogged at Intel > is semi-blest (believe me, I know the flip side.) The blog post reminded me. I have assumed, for years, that Intel planned on putting many (>>4) x86 cores on a single-die. I'm sure I can find Intel presentations from the nineties that seem to make that clear if I dig hard enough. From the very beginning, Larrabee seemed to be a technology of destiny in search of a mission, and the first, most obvious mission for any kind of massive parallelism is graphics. Thus, Intel explaining why it would introduce Larrabee at Siggraph always seemed a case of offering an explanation where none would be needed if the explanation weren't something they weren't sure they believed themselves (or that anyone else would). It just seemed like the least implausible mission for hardware that had been designed to a concept rather than to a mission. A more plausible claim that they were aiming at HPC probably wouldn't have seemed like a very attractive business proposition for a company the size of Intel. Also from the beginning, I wondered if Intel seriously expected to be able to compete at the high end with dedicated graphics engines using x86 cores. Either there was something about the technology I was missing completely, it was just another Intel bluff, or the "x86" cores that ultimately appeared on a graphics chips for market would be to an x86 as we know it as, say, a, lady bug is to a dalmatian. Robert. |