GTX 580 release

Looks like the first reviews of the GTX580 are coming out!
The initial one is pretty exhaustive.

As expected, GF110 used by GTX580 is not based on the GF104 architecture, it’s GF100 based. Compared to the GTX480, shader and memory clock rates are up about 10%, SMs are up to 512 from 448, and power use is down about 10%, and the GTX580 has an even better cooler on it making it quieter than the GTX480. This all matches the rumors and speculation from the past two weeks.

The only uncertainty is whether GF110 will support the Tesla features of ECC and full throughput DP. One clue is that there are slightly fewer transistors in GF110 (3.0B as opposed to GF100 3.2B), but this could be from anything including simply better layout. (My personal guess is that ECC was removed, and we won’t see a GF110 Tesla, but this could easily be wrong.)

For CUDA programmers, GF110 is certainly superior to GF100, but won’t make any change to the way you code or use CUDA… it’s just a nicer card with about 15-20% more horsepower.

Looks like the first reviews of the GTX580 are coming out!
The initial one is pretty exhaustive.

As expected, GF110 used by GTX580 is not based on the GF104 architecture, it’s GF100 based. Compared to the GTX480, shader and memory clock rates are up about 10%, SMs are up to 512 from 448, and power use is down about 10%, and the GTX580 has an even better cooler on it making it quieter than the GTX480. This all matches the rumors and speculation from the past two weeks.

The only uncertainty is whether GF110 will support the Tesla features of ECC and full throughput DP. One clue is that there are slightly fewer transistors in GF110 (3.0B as opposed to GF100 3.2B), but this could be from anything including simply better layout. (My personal guess is that ECC was removed, and we won’t see a GF110 Tesla, but this could easily be wrong.)

For CUDA programmers, GF110 is certainly superior to GF100, but won’t make any change to the way you code or use CUDA… it’s just a nicer card with about 15-20% more horsepower.

A little shaky video showing the new “Alien vs. Triangles” tech demo, and a few slides about the new vapor chamber cooler.

A little shaky video showing the new “Alien vs. Triangles” tech demo, and a few slides about the new vapor chamber cooler.

I’m 99.99% sure we won’t be told anything directly from Nvidia for many many months and the only way we will find out is from people buying them and posting in this forum in several weeks time.

I’m 99.99% sure we won’t be told anything directly from Nvidia for many many months and the only way we will find out is from people buying them and posting in this forum in several weeks time.

They removed the review and the corresponding forum threads again (probably because they violated the NDA).

They removed the review and the corresponding forum threads again (probably because they violated the NDA).

Does it mean it has compute capability 2.0 (i.e. 32 cuda cores per multiprocessor)?

Does it mean it has compute capability 2.0 (i.e. 32 cuda cores per multiprocessor)?

These cards are now available on Newegg priced ~$100 over the GTX 480 ($560 vs $450-470). I assume that after the initial launch price surge, these cards will slowly squeeze out the GTX 480 as existing supplies run out.

These cards are now available on Newegg priced ~$100 over the GTX 480 ($560 vs $450-470). I assume that after the initial launch price surge, these cards will slowly squeeze out the GTX 480 as existing supplies run out.

Anandtech’s review.

A lot more technical info in this one, especially talking about GF110 versus GF100 and GF104.

The big difference is GF110 has one of GF104’s improvements in improved fp16 texture sampling rate.
It does not have GF104’s superscalar improvements, or GF104’s improved texture:SP ratio.

GF110 has one new boost to (graphics-only) Z-culling, useful for dense tessellation triangle rendering.

No word on what compute level this will be… I suspect 2.0, since it’s not superscalar like 2.1 so there’s no advantage to hint the compiler.

Anandtech’s review.

A lot more technical info in this one, especially talking about GF110 versus GF100 and GF104.

The big difference is GF110 has one of GF104’s improvements in improved fp16 texture sampling rate.
It does not have GF104’s superscalar improvements, or GF104’s improved texture:SP ratio.

GF110 has one new boost to (graphics-only) Z-culling, useful for dense tessellation triangle rendering.

No word on what compute level this will be… I suspect 2.0, since it’s not superscalar like 2.1 so there’s no advantage to hint the compiler.

And Tech Report’s.

They state that GF110 still has the full Fermi features of DP throughput and ECC. So we should see some new Teslas and Quadros. (The power savings is especially important for HPC!)
Like GF100, they’re disabled for the consumer cards like the GTX580.

And Tech Report’s.

They state that GF110 still has the full Fermi features of DP throughput and ECC. So we should see some new Teslas and Quadros. (The power savings is especially important for HPC!)
Like GF100, they’re disabled for the consumer cards like the GTX580.

GF110 is Compute 2.0.

GF110 is Compute 2.0.

Aliens versus triangles tech demo now available… fun eye candy.

Aliens versus triangles tech demo now available… fun eye candy.