Review Index:

NVIDIA GeForce GTX 580 1.5GB Review and SLI Testing - GF110 brings full Fermi

Manufacturer: NVIDIA

GF110 is a new GPU

What should have been

It might be cliche to say it as this point, but despite the launch of NVIDIA's GeForce GTX 480 and 470 cards offering the best performing single GPUs available at the time, there was a lot about the GF100 GPU that didn't go as planned.  First, it was late and left NVIDIA with a 6+ month deficit in DX11 market share.  Secondly, and most pestering to consumers, the GTX 480/470 GPU was hot and noisy.

The GPU itself never actually met the specifications that NVIDIA set out to reach 4-5 months prior to its launch either.  The GPU was capable of including as many as 512 shader cores according to NVIDIA's own information, but when the GTX 480 launch in only enabled 480 of them in order for the company to meet temperature and yield demands.  The oft-discussed "full" Fermi GPU was never seen.  Until today. 

Today's release of the GeForce GTX 580 graphics card introduces the GF110 GPU, a revision to the high-end offering from NVIDIA that brings us the first "full" Fermi offering.

The GF110 GPU

If this diagram looks familiar, it should.  This is the same block setup we showed you in the GF100/Fermi GPU release and the GF110 stays true to this design.  The key this time is that all 512 CUDA cores, and all 16 SMs, are enabled and active on the GF110 where as only 15 SMs and 480 CUDA cores were active on the GTX 480 cards. 

As you would expect, not only do we get additional CUDA/shader cores in the GF110 GPUs but also additional texture and tessellation performance include with the addition of that extra SM.  Texture unit count goes from 60 on the GTX 480 to 64 on this new GF110 GPU and this new iteration sees an additional bump thanks to more Polymorph Engine power.  The ROP count remains the same though with a count of 48 since our memory bus also remains the same at 384-bits wide. 

The GF110 is still a 3.0 billion transistor GPU (despite rumors to the contrary) and has not dropped features like ECC to reach these performance levels.  The secret to NVIDIA's improved performance and yield comes from a monitoring and rearranging of the transistor paths that caused the most leakage on GF100 in real-world environments.  NVIDIA's engineers then rearranged many segments of the die to better improve efficiency and power consumption finally allowing them to enable the full power of the architecture as GF110.

There have been some other architectural enhancements made on the GF110 GPU as well including a full speed FP16 texture filtering addition and improved Z-cull efficiency.  These clock-for-clock improvements in the GPU design offer performance improvements of anywhere from 5% in Metro 2033 to 13% in Dirt 2 based on NVIDIA's own testing. 

GF110 improvements don't stop there though; the clock speeds on the GPU have also been increased along with the above mentioned architectural changes.  The new reference speed of the GPU is 772 MHz core clock / 1544 MHz shader clock and the GDDR5 memory clock goes from 924 MHz to 1002 MHz, 4008 MHz effective. 

According to NVIDIA's chart above, the GTX 580 will be anywhere from 15-30% faster than the GTX 480 thanks to the combination of faster clocks, more processing cores and architectural changes.  We'll have to see how these claims work out in our testing of course. 

Good news for those of you worried about heat and power - the GTX 580 is rated at nearly identical TDP levels as the GTX 480 but with an improved cooler design noise levels should be much improved.  Physically, the GeForce GTX 580 card is nearly identical to the GTX 480 and is actually pin-compatible with the GF100 GPU.
An interesting addition to the GTX 580 hardware is a hardware-based current monitor that keeps track of the voltage and current coming IN through the 12V rails on the 8-pin, 6-pin and PCIe bus.  The goal here was to keep the GPU within the power specifications shown above in all cases, even when programs like the oft-despised FurMark are run that heat up the GPU "more than any real-world application ever could" according to NVIDIA.  The logic on board will adjust the clocks and performance of the GPU when these types of current draws are detected thus limiting the benefit of such applications for temperature and cooling testing.

While I have no problem with NVIDIA protecting their hardware in this way I am curious to see how long it will take for a "real-world" application to hit this current draw limits in some way and initiate these performance drops on its own.  FurMark doesn't really do anything special to the GPU - it simply attempts to work as much of the GPU core as possible all at one time.  In many people's view, this is only testing the top efficiency possible out of the GPU - something that software developers should really be striving for anyway. 

Obviously with improved performance with additional CUDA cores and architectural improvements and no real rise in the TDP of the GTX 580 compared to the GTX 480, performance per watt should rise as well.  Considering this is an area that AMD has continued to dominate NVIDIA in with the Radeon HD 5000 and now 6000 series, we'll be interested to see how that pans out after our own testing.
Thanks to an improved cooler that utilizes a vapor chamber cooler, the GTX 580 GPU is actually going to be quieter than the GTX 285 based on NVIDIA's testing.  Let's get to the card itself and see what it has to offer high-end gamers.

No comments posted yet.

Post new comment

The content of this field is kept private and will not be shown publicly.
  • Lines and paragraphs break automatically.
  • Allowed HTML tags: <a> <em> <strong> <cite> <code> <ul> <ol> <li> <dl> <dt> <dd> <blockquote><p><br>
  • Web page addresses and e-mail addresses turn into links automatically.

More information about formatting options

This question is for testing whether you are a human visitor and to prevent automated spam submissions.