Wednesday, May 25th 2011
ASUS Designs MARS II Dual-GTX 580 Graphics Card for GeForce Enthusiasts
When ASUS first set out to build the MARS, a dual-GeForce GTX 285 graphics card, people thought the designers were crazy. NVIDIA could barely get the two 55 nm G200b GPUs on GeForce GTX 295 to run at GTX 285 specifications of 240 shaders and 512-bit GDDR3 each, and instead ran them with 448-bit wide GDDR3 memory interface, and lower clocks. ASUS managed to back two GPUs with insanely strong VRM and cooling, to achieve a dual-GPU graphics card that was more than just a GTX 295 overclocked. ASUS is back on the drawing boards with MARS II, a new dual-GPU graphics card that runs two GF110 GPUs in the same exact config as on GeForce GTX 580, perhaps with clock speeds higher than those of GTX 580.
The GeForce GTX 590 runs two GF110 GPUs with all CUDA cores and the complete width of the memory bus enabled, but the GPUs and memory are clocked significantly lower at 607 MHz core, 1214 MHz CUDA cores, and 3.42 GHz memory; while the single-GPU GTX 580 runs at 772 MHz core, 1544 MHz CUDA cores, and 4.00 GHz memory. MARS II also aims to get over the various design pitfalls of GTX 590 that made voltage-assisted overclocking practically impossible on air-cooling. To do this, MARS II will make use of a massive 19-phase VRM with Super Alloy chokes. To cool the beast, ASUS will put to use the expertise it gathered over time with its DirectCu series video card coolers. The new cooler will provide uniform cooling to both GPUs. More details, particularly about the when the card will be released, are awaited.
Source:
DonanimHaber
The GeForce GTX 590 runs two GF110 GPUs with all CUDA cores and the complete width of the memory bus enabled, but the GPUs and memory are clocked significantly lower at 607 MHz core, 1214 MHz CUDA cores, and 3.42 GHz memory; while the single-GPU GTX 580 runs at 772 MHz core, 1544 MHz CUDA cores, and 4.00 GHz memory. MARS II also aims to get over the various design pitfalls of GTX 590 that made voltage-assisted overclocking practically impossible on air-cooling. To do this, MARS II will make use of a massive 19-phase VRM with Super Alloy chokes. To cool the beast, ASUS will put to use the expertise it gathered over time with its DirectCu series video card coolers. The new cooler will provide uniform cooling to both GPUs. More details, particularly about the when the card will be released, are awaited.
19 Comments on ASUS Designs MARS II Dual-GTX 580 Graphics Card for GeForce Enthusiasts
I wonder how the cooling is going to work out. Direct CU II coolers on higher-end cards really are amazing (I own one) so this card may be extremely low noise to boot.
It is much harder to cool dual-GPU cards than single GPU cards.
This card is just an e-peen item just like every Mars or Ares Asus ever released.More like $1200 at launch.
At $999 this card will be a bargain.
Anyone wanting to fit these into a small space should expect to have to water cool them.
One can safely assume that any dual 580 design will use nearly twice as much power as a 580. That's nearly 600W, when the stock 590 is set to stick within a 365W TDP (unless you manually exceed that).
Regarding E-ATX baords, some are already available, but for most peopl, even with high-end systems, ATX still makes most sense - triple GPU setups aren't that popular even amongst those that can afford them.
Back to your original remark, I have a 130W GPU (HD6850) in my main system and I choose to use a triple-slot cooler. I don't use any other expansion slots and I have a full ATX case, so the lower noise and temperatures make sense for me. My PC is almost entirely inaudible while GPU folding, for example. With the original reference design, which is praised by most reviews for its low noise, the noise levels when folding were intolerable.