What will Microsoft use its ARM license for?
|
Prev: Effects of Memory Latency and Bandwidth onSupercomputer,Application Performance
Next: Effects of Memory Latency and Bandwidth on Supercomputer,Application Performance
From: Torben �gidius Mogensen on 2 Aug 2010 08:17 gnirre <gnirre(a)gmail.com> writes: > Will Microsofts design an ARM processor? If so, what operating system > will it be intended for? What architectural features would Microsoft > be looking for in its own ARM processor? I doubt it is for a mobile unit, as that would be served well enough with an existing core (Cortex A8 or A9). Some have mentioned future Xboxes as a possibility, which is possible, but even there a multicore A9 would serve. My own guess is that MS will make a cheap and lean server processor for their Azure cloud and for webservices like Bing. If you want to compete with Google, you need huge server parks, but the processors do not need to be very powerful, as long as they have good memory interfaces. The cost of designing their own core is easily gained in lower per-processor cost and lower power use (which implies less heating). I think Google may have similar plans, though they are more likely to use an existing core and just add extras around this. Torben
From: Brett Davis on 3 Aug 2010 00:44 In article <7zocdlp5th.fsf(a)ask.diku.dk>, torbenm(a)diku.dk (Torben ?gidius Mogensen) wrote: > > Will Microsofts design an ARM processor? If so, what operating system > > will it be intended for? What architectural features would Microsoft > > be looking for in its own ARM processor? > > My own guess is that MS will make a cheap and lean server processor for > their Azure cloud and for webservices like Bing. If you want to compete > with Google, you need huge server parks, but the processors do not need > to be very powerful, as long as they have good memory interfaces. The > cost of designing their own core is easily gained in lower per-processor > cost and lower power use (which implies less heating). Even one of the biggest cloud computing centers in the world is only going to cost you a billion dollars. Chump change for Microsoft or any other big company. http://www.appleinsider.com/articles/09/08/18/expert_speculates_apples_new_data_center_to_be_for_cloud_computing.html Its going to cost Microsoft a couple of hundred million to roll their own ARM chip with accessories. The math no worky. The idea of a low power ARM server sounds good on paper, but lets look at the facts: No gigabit ethernet support. No 10 gigabit ethernet support, at all. No server class 10 gigabit ethernet. (Fast DMA.) A feeble 32 bit memory interface, not 64, (low end everyone) not 96, (Intel) not 128. (AMD) RAM support for obsolete DDR, or at most DDR2, without the memory controller support to actually stream at those rates. A 32 bit address space, not 64, too small to be called a desktop much less a server today. No support for large memory configurations. I have 12 gigs of RAM in my desktop, and only half the slots are full. A 32 bit address space can only talk to 2 gigs of RAM. So you are talking about ARM "servers" with 2 gigs RAM and 100 meg ethernet competing against 1U servers with 256 gigs of DDR3 RAM and dual port 10 gigabit ethernet, or better. ;) This is comp.arch, not the Amiga users newsgroup. ;) Brett PS - I see the 40Gbps Dual port InfiniBand adapters are below $1000 now. Well worth the cost if you can replace 4 of those old slow 10 gigabit ethernet servers with one of these new fast Infiniband servers. ;) These numbers are from my first Goggle hit: http://www.thinkmate.com/System/RAX_QS3-4110 My first draft of this post had only 64 gigs RAM. ;)
From: Paul Gotch on 8 Aug 2010 06:34 Brett Davis <ggtgp(a)yahoo.com> wrote: <snip list of stuff> ARM licenses cores, or the right to write your on ARM architecture compliant core, which partners then build into SoCs. Except for the address space all the other things you mentioned are implementation details of a particular SoC not of the core itself. -p -- Paul Gotch --------------------------------------------------------------------
From: Terje Mathisen "terje.mathisen at on 9 Aug 2010 02:23 Owen Shepherd wrote: > Certainly AVR32 and Thumb show between them that many of the > features can be kept; its just a matter of value - note that > neither Thumb or AVR32 are highly orthogonal architectures like > more traditional designs, as they're much more aimed towards > compilers. Huh? So are you saying that Thunb/AVR32 are non-orthogonal, because they are aimed at compiled code, or that 'more traditional designs' have that aim? The latter does make sense, but not to my non-native reading of your english. :-( Terje -- - <Terje.Mathisen at tmsw.no> "almost all programming can be viewed as an exercise in caching"
From: Torben �gidius Mogensen on 9 Aug 2010 04:56
Owen Shepherd <owen.shepherd(a)e43.eu> writes: > Much of the focus these days is on the Thumb2 instruction set. > This is still fully predicated, but does this by using special > instructions (Called "IT," for some reason beyond my knowledge). IT is for "if-then". It is followed by up to three Ts and Es ("then" and "else") describing which of the following instructions are executed when the condition is true or false. Hence, 1-4 instructions can be made conditional. In old-style ARM code it was common to have several adjacent conditional instructions and usually all with the same or the negated condition. The IT instruction is a compact form of encoding this pattern. Torben |