www.sicortex.com
|
From: David Kanter on 3 Apr 2008 13:09 On Apr 3, 1:01 am, n...(a)cus.cam.ac.uk (Nick Maclaren) wrote: > In article <DuqdnavI0b-KEWnanZ2dnUVZ8uidn...(a)giganews.com>,Terje Mathisen <terje.mathi...(a)hda.hydro.com> writes: > > |> > |> > Also, the great unknown is whether you can get a higher proportion > |> > of its theoretical peak out of it than you can on dual-socket, > |> > quad-core Intel systems. That is the key to whether it is a good > |> > buy or a bad one - plus its price, of course. > |> > |> Price is the really important consideration: > > Agreed. > > |> Currently StatoilHydro has around 20 TF (afair) in its seismic clusters, > |> and the machine rooms have been through the needed refit to handle the > |> cooling issue. > > Not everyone has that option. I was one of the first people to hit > this hard, but a lot of sites are limited by cooling and/or power or > space. They quite simply CAN'T upgrade for less than a cost that would > make any plausible computers look cheap. For example, some of the > finance houses located in central London, New York etc. are like that; > and their available budget makes salesmen drool :-) > > So there is still a significant market even without the price factor. > But I agree that performance/Euro will sell more than performance/watt, > performance/sq.m. or performance/Kg. > > Regards, > Nick Maclaren. Yup. The interesting question is the performance of SiCortex relative to a heavily power optimized x86 system. For instance, using 50W 2.5GHz cpus instead of 3.2GHz 120W cpus and using DDR2 instead of FB- DIMMs. Usually an appropriate x86 box (one that exists) beats most of these non-traditional solutions or comes near enough that stuff like SiCortex never makes sense in the long run (i.e. there may be a brief window where an alternative solution is optimal, but rarely for an extended period of time). DK
From: Greg Lindahl on 3 Apr 2008 13:18 In article <ft2uek$987$1(a)USTR-NEWS.TR.UNISYS.COM>, Tim McCaffrey <timcaffrey(a)aol.com> wrote: >And to followup, Google Earth shows they are in or right next to The Mill. They are in the Mill, however, they are the lost tribe of Thinking Machines. Some DEC people, yes. These are the same guys who bought the PathScale compiler group. -- greg
From: Nick Maclaren on 3 Apr 2008 13:32 In article <47f5117b$1(a)news.meer.net>, lindahl(a)pbm.com (Greg Lindahl) writes: |> In article <ft2uek$987$1(a)USTR-NEWS.TR.UNISYS.COM>, |> Tim McCaffrey <timcaffrey(a)aol.com> wrote: |> |> >And to followup, Google Earth shows they are in or right next to The Mill. |> |> They are in the Mill, however, they are the lost tribe of Thinking |> Machines. Some DEC people, yes. Ah! Them. That figures - it has their style. Regards, Nick Maclaren.
From: Paul Gotch on 3 Apr 2008 14:02 David Kanter <dkanter(a)gmail.com> wrote: > Yup. The interesting question is the performance of SiCortex relative > to a heavily power optimized x86 system. For instance, using 50W > 2.5GHz cpus instead of 3.2GHz 120W cpus and using DDR2 instead of FB- > DIMMs. They are talking about 600mw 500MHz processors. They say they've licensed the MIPS64 architecture however I wouldn't be at all supprised if the processor was actually a MIPS 20Kc or a tweak thereof since they claim they've taking the approach of licensing resuable IP to build the machine with a small team. This is the same approach that IBM have done, all be it in a much more specialised way, with the BlueGene project. > Usually an appropriate x86 box (one that exists) beats most of these > non-traditional solutions or comes near enough that stuff like > SiCortex never makes sense in the long run (i.e. there may be a brief > window where an alternative solution is optimal, but rarely for an > extended period of time). I remain unconvinced, although we'll see if Intel can come out with a 64 bit x86 SOC processor which has a worst case power of < 1W at between 500MHz to 1GHz and can deliver the same or greater number of FLOPS. -p -- "Unix is user friendly, it's just picky about who its friends are." - Anonymous --------------------------------------------------------------------
From: Jeff Kenton on 3 Apr 2008 18:55
Tim McCaffrey wrote: > In article <ft0fhe$7qg$1(a)gemini.csx.cam.ac.uk>, nmm1(a)cus.cam.ac.uk says... >> >> It looks interesting, but will it succeed? Any serious or even >> humorous comments on it appreciated. >> > Considering the location, is this where one of the lost tribes of DEC ended > up? > > - Tim > At least some of them came from the BBN Butterfly group. Don't know about any DECies, but they are in the old DEC mill building. -- --------------------------------------------------------------------- = Jeff Kenton http://home.comcast.net/~jeffrey.kenton = --------------------------------------------------------------------- |