- Saturday, 4 July 2009
- INQ Mobile
- RSS
The palest ink is better than the best memory - Chinese proverb
AMD announces petaflop supercomputer
AMD HAD some interesting things to say at its CES keynote, one of which signaled an industry paradigm shift. When AMD's CEO Dirk Meyer says that more transistors are no longer the way to push innovation, you have to pay attention.
While he didn't say it outright, the point is simple. More transistors have a diminishing rate of return in making a core, and more cores aren't really going to do much for the consumer at this point. There has to be another way to use the transistors, and to use what the fabs can pump out. That is where innovation will be in the future.
One thing that AMD thinks will drive things is film quality animation. The example it used was Spiderman 2, one of the first films with a totally digital actor. In the end when Doc Oc is falling into the pit, he is totally digital. The actor was captured in a lightstage, and from there you can animate him, slowly, but almost perfectly. Will Smith in Hancock was also shown using this technology.
But when frames take a day or more to render, that puts a big crimp on the usefulness of any technology. With GPGPU acceleration, companies are now able to manipulate that data in nearly real time instead of days. That is progress for you, but ever increasing data rates will catch up and overwhelm that performance capability soon enough.
Luckily, there is a solution. AMD announced a petaflop supercomputer for on-premises or cloud rendering.
Imagine a supercomputer like the current petaflop designs, but instead of being sized like a datacenter, it will fit in a single room. AMD is building a system with more than 1,000 R4870 class GPUs, each pumping out just over a teraflop, and the resulting complex will consume about 1/10th of the power of other petaflop-scale supercomputer installations.
The software that AMD runs on it is from Otoy, and the system is called the System Fusion Render Node.
Because AMD controls both the CPU and GPU designs, it can operate them in much closer synchronization, building a cohesive fabric of nodes instead of just a collection of cards in boxes like some other 'solutions'. The system can also be continuously upgraded, so that when the R5870 hits, you will be able to upgrade the machines.
With that, AMD will be taking the concept of petaflop machines from large government sponsored institutions and purpose-built facilities with customised power feeds and cooling down to a single room. Not only that, but it will be commercialized and available.
This technology will trickle down to the consumer in short order, and soon enough, people will be complaining that their petaflop phone is too damn slow. µ
Share this:
Business Intelligence Software
Customer Relationship Management (CRM)
Enterprise Accounting Software
Enterprise Resource Planning (ERP)
Enterprise Systems Management
I think the most interesting comment in this article was about the tight synchronization between the AMD CPU and the AMD GPU.
With Intel and Nvidia getting in each others space concerning both CPU's and GPU's, could this become an Archilles heel in Intel/Nvidia marchitecture especially concerning bus archicture,licensing and ultimately performance?
Also, could this comment (if it has any vailidity) portend the fact that AMD could become the next SGI...not in the sense of actually building extraordinarily pretty and overpriced graphics workstations and supercomputers...but in the sense of SGI at the time they owned MIPS and controlled and engineered both the CPU and GPU they flogged into their boxen? Also consider this....SGI came up with OpenGL. Now AMD comes up with OpenCL to "sychronize" CPU's and GPU's for higher levels of performance.
History may not repeat but AMD may be on the verge of rhyming quite a bit with the SGI of old.
Just some interesting thoughts.
Charlie must work for AMD. This is pure marketing! Man, how I hate such articles. Why don't you improve your new site with a filter for author? Blargh.
SGI just isn't SGI without their pretty workstations (so I'm ignoring their IRIS/4D series).
If AMD really has SGI in mind when planning their moves, I'll get around to sending an email to AMD to start making stylish, colourful, overpriced computers with names based on hydrocarbons pretty soon
1st things 1st, Charlie, please don't mix up product names/codes. You of all people should know its the HD4870 not the R4870 (the R belongs to the GPU code name, the RV770). And @The Sentinel, AMD didn't come up with Open CL, Apple kinda did that. But anyway, interesting article and amazing to see AMD getting so many GPUs to work together when CrossfireX is so pish, lol
Apologies Cool Tech. You are quite right concerning OpenCl. I had in mind CAL which AMD markets as STREAM which is their version of Nvidia's CUDA.
Of course, AMD supports OpenCL as well.
Thanks.
CUDA vs CAL is apples vs oranges. In one you must worry about the cache and mem, in the other you don't have to.
AFACT 4870 has better DP FP performance but GT200 has better SP FP performance.
Pick you problem and your method of solution, the choice of product will be clear.
Perhaps OpenCL will supplant CAL and CUDA at the API level. Nevertheless each company will offer products with better or worse performance.
Just like, say, POSIX threads on Intel vs AMD, OpenCL on AMD vs Nvidia will be a question of which company make better hardware for the API.
Heres NEW WIG for Mr. Myers, Its HOT, Hotter than AMD, Yet its Big, very big. Burning Elephant, You Ask? No, first, Hon Barack Esq, Has Pledged to 900,000 New Engineers for IBM, To help eco sytem, or economackie, as its expressed in Dollars: $45 Bil. Next, Right Out of Mkes Rogister is this article: http://www.theregister.co.uk/2009/01/05/windows_for_warships_hits_type_23s/
As You Can See HRM is Planning New Shebangun. & AMD Racing with Intel could be Key IN. Both Worlds, Empty World of Drak in Pathic Non Medical sector, comverges with Sluethe of Dirk, Hunting 'Em down, Hunting 'Em Down. This could be Vacumne Chamber of Decade. New, Newness that goes beyond Dual Layer o/s systems & into Blu. this could be way to: Catch Up. TS Drashek
Inventor of:Super Blu Ray: MILATARY.
but what is with the anti nvidia and pro AMD news reporting from Charlie. Surely you are meant to hide the fact that you work for a company much better than this.
no Shame, although ive yet to see an Nvidia "Scandal today". Im waiting to see whats next from the guys in green. Always entertaining !
the reality is that fusion has degenerated into a branding exercise - kind of sad.
amd should just bury it b/c when intel release a real fusion chip, amd would be caught with its pants down.
Have to wonder about the literacy levels of some of the above readers. Lawrence Livermore identified this design as the best way forward over 3 years ago. https://www.llnl.gov/str/November05/pdfs/11_05.3.pdf Intel has been asleep at the switch as they are way behind where this design is and can't make the cut in DARPA's research projects. HEC Phase III for 2010 is AMD and IBM http://www.nitrd.gov/Pubs/2009supplement/NITRD-09Supp_FINAL-hec-rd.pdf So it looks like Charlie got the story right as usual.