Intel's Larrabee to also be presented at Hot Chips

[NV55]

Larrabee to also be presented at Hot Chips
Friday 13th June 2008, 11:24:00 PM, written by Arun

Since the rest of the internet is still at the stage where they're all
excited about Larrabee being presented at Siggraph (hint: you guys are
ten days late), we thought we'd let you know it will also be presented
at Hot Chips, presumably with more of a hardware perspective.

We hope Intel will actually dare to make their strategy clear at these
two events, especially when it comes to rasterisation vs raytracing,
developer evangelism, and DirectX 11. Let's make one thing clear:
there's no real difference between the current ray tracing stratagems
of Intel and NVIDIA, or what will come out when the end games of both
are presented.

The implementation details of how they want to make raytracing fast
may vary, but both see it is as a very important research project that
should not, however, be applied too much too fast. It is amusing how
it seems that NVIDIA thinks Intel takes raytracing more seriously than
they really do, while Intel thinks the same for NVIDIA with
rasterisation. As is true about many parts of the semiconductor
industry and life in general, the truth is often in the middle.

http://www.beyond3d.com/content/news/655

 

[Bill Davidsen]

Larrabee to also be presented at Hot Chips
Friday 13th June 2008, 11:24:00 PM, written by Arun

Since the rest of the internet is still at the stage where they're all
excited about Larrabee being presented at Siggraph (hint: you guys are
ten days late), we thought we'd let you know it will also be presented
at Hot Chips, presumably with more of a hardware perspective.

We hope Intel will actually dare to make their strategy clear at these
two events, especially when it comes to rasterisation vs raytracing,
developer evangelism, and DirectX 11. Let's make one thing clear:
there's no real difference between the current ray tracing stratagems
of Intel and NVIDIA, or what will come out when the end games of both
are presented.

The implementation details of how they want to make raytracing fast
may vary, but both see it is as a very important research project that
should not, however, be applied too much too fast. It is amusing how
it seems that NVIDIA thinks Intel takes raytracing more seriously than
they really do, while Intel thinks the same for NVIDIA with
rasterisation. As is true about many parts of the semiconductor
industry and life in general, the truth is often in the middle.

http://www.beyond3d.com/content/news/655

In the context of graphics, AMD released a half-teraflop GPU chip,
report a few days ago. Clearly it should work with games, video, and
scientific computing. I think this is the GPU IBM used for their
petaflop system, but I can't find the article quickly.

http://www.engadget.com/2008/06/16/amds-firestream-9250-first-processor-to-top-1-teraflop/

 

[NV55]

In the context of graphics, AMD released a half-teraflop GPU chip,
report a few days ago. Clearly it should work with games, video, and
scientific computing. I think this is the GPU IBM used for their
petaflop system, but I can't find the article quickly.

http://www.engadget.com/2008/06/16/amds-firestream-9250-first-process...

AMD released a *one* -full- TeraFLOP GPU, the RV770, which powers the
Radeon HD 48xx series.
Even the $200 4850 is 1 TeraFLOP.
The faster $300 4870 coming in July is 1.2 TeraFLOP
The $500 (I'll bet it'll be $600 tho) R700: 4870X2 with two RV770
GPUs will provide 2 TeraFLOPS per card when it gets released in August
(or September).

IBM did not use GPUs for their PetaFLOP supercomputer. They used an
enhanced, double-precision tuned version of the CELL CPU, the IBM
PowerXCell 8i, 12,960 of them, to reach that milestone.

 

[NV55]

I forgot to mention, that with CrossFireX, four 4870 cards will
offer about 5 TeraFLOPs performance.

Roughly the same will be possible with two 4870X2 cards.


Of course this is all for graphics processing and GPGPU applications.

 

[Rev. 11D Meow!]

I forgot to mention, that with CrossFireX, four 4870 cards will
offer about 5 TeraFLOPs performance.

Roughly the same will be possible with two 4870X2 cards.


Of course this is all for graphics processing and GPGPU applications.

You drive a Chevy Nova, don't you?

 

[Yousuf Khan]

In the context of graphics, AMD released a half-teraflop GPU chip,
report a few days ago. Clearly it should work with games, video, and
scientific computing. I think this is the GPU IBM used for their
petaflop system, but I can't find the article quickly.

http://www.engadget.com/2008/06/16/amds-firestream-9250-first-processor-to-top-1-teraflop/

IBM paired up some Opterons with their own Cell processors to get the
petaflop Roadrunner system. It used 6912 1.8Ghz Opteron 2210 dual-core
processors. It then pairs them up with 12960 PowerXCell 8i 3.2Ghz
processors. It uses blades which hold two Opterons, and four PowerXCells.

IBM Roadrunner - Wikipedia, the free encyclopedia
http://en.wikipedia.org/wiki/IBM_Roadrunner#Processors

However, I'm sure there is some great potential AMD to get to 1 teraflop
all by itself with a combination of Opterons and their new HD4000-series
GPUs. The 4850 by itself gets to 1 teraflop in single-precision, so a
thousand will take it to 1 petaflop. However, the 4850 only does 200
gigaflop in double-precision; so if it's going to match the 1 petaflop
mark at double-precision, it requires 5000 of those chips.

Yousuf Khan

 

[nv55]

You drive a Chevy Nova, don't you?

No, a hydrogen fuel cell powered go-kart.