Sign in with
Sign up | Sign in

GF110: Nvidia Gives Fermi A Facelift

GeForce GTX 580 And GF110: The Way Nvidia Meant It To Be Played
By

As I’ve mentioned, GF110 is decidedly evolutionary, and Nvidia already had a foundation to build on with GF100, so the number of changes made this time around is actually pretty small. We’re still looking a 40 nm chip consisting of roughly three billion transistors.

First off, the GPU itself is largely the same. This isn’t a GF100 to GF104 sort of change, where Shader Multiprocessors get reoriented to improve performance at mainstream price points (read: more texturing horsepower). The emphasis here remains compute muscle. Really, there are only two feature changes: full-speed FP16 filtering and improved Z-cull efficiency.

Finally, we don't have to grey out any of this block diagram...Finally, we don't have to grey out any of this block diagram...

GF110 can perform FP16 texture filtering in one clock cycle (similar to GF104), while GF100 required two cycles. In texturing-limited applications, this speed-up may translate into performance gains. The culling improvements give GF110 an advantage in titles that suffer lots of overdraw, helping maximize available memory bandwidth. On a clock-for-clock basis, Nvidia claims these enhancements have up to a 14% impact (or so).


GeForce GTX 580
GeForce GTX 480
GeForce GTX 470
Graphics Processing Clusters (GPCs)
4
4
4
Streaming Multiprocessors (SMs)
16
15
14
CUDA Cores
512
480
448
Texture Units
64
60
56
ROP Units
48
48
40
Graphics Clock
772 MHz
700 MHz
607 MHz
Shader Clock
1544 MHz
1401 MHz
1215 MHz
Memory Clock (Data Rate)
1002 MHz (4008 MT/s)
924 MHz (3696 MT/s)
837 MHz (3348 MT/s)
Memory Capacity
1.5 GB GDDR5
1.5 GB GDDR5
1.25 GB GDDR5
Memory Interface
384-bit
384-bit
320-bit
Memory Bandwidth
192.4 GB/s
177.4 GB/s
133.9 GB/s
Fillrate
49.4 GTexels/s
42.0 GTexels/s
34.0 GTexels/s
Manufacturing Process
40 nm TSMC
40 nm TSMC
40 nm TSMC
Form Factor
Dual-slot
Dual-slot
Dual-slot
Display Outputs
2 x DL-DVI, 1 x mini-HDMI2 x DL-DVI, 1 x mini-HDMI
2 x DL-DVI, 1 x mini-HDMI


Part of the shift from GF100 to GF110 involves a chip-level re-work. There are different types of transistors an architect can use to build an integrated circuit, depending on the properties he wants to impart. Nvidia’s engineers went back to their GF100 design and purportedly modified much of it, implementing slower, lower-leakage transistors in less timing-sensitive paths and faster, higher-leakage transistors in other areas.

The result was a significant enough power savings  to not only allow Nvidia to turn on the 16th Shader Multiprocessor originally disabled in its original design (adding 32 CUDA cores, four texture units, and a single PolyMorph geometry engine), but also ramp up clock rates. Whereas the GeForce GTX 480 sported core/shader/memory frequencies of 700/1401/924 MHz, GeForce GTX 580 employs a 772 MHz core clock, a 1544 MHz shader frequency, and a 1002 MHz memory clock (which translates to a 4008 MT/s data rate).  All told, GF110 offers specifications closer to what we were expecting earlier this year. It includes 512 functional CUDA cores, 64 texture units, and 16 PolyMorph engines.

The independent back-end looks the same on a block diagram, the main change being the faster memory clock. It still features six ROP partitions, each associated with a 64-bit memory interface (totaling 384-bit aggregate). Each partition is capable of outputting eight 32-bit integer pixels at a time, totaling 48 pixels per clock.

Telling That Tessellation Tale

Geometry is suddenly causing quite a stir. When Nvidia briefed us on its Fermi architecture and how it related to gaming, DirectX 11 and tessellation came up over and over. The fact that the company built one of its PolyMorph engines into each of its Shader Multiprocessors (x16 on the GF110 ASIC) would purportedly make a huge difference in titles that extensively utilized tessellation to help increase realism. The problem, of course, was that there weren't any games to show off at the time. Yeah, DiRT 2 and Aliens vs. Predator were out, but both "first-gen" DirectX 11 titles are very choosy in where geometry gets added (neither was able to back Nvidia's claims that more geometry was the future of gaming).

And then there was HAWX 2. The game doesn't actually launch until after the GeForce GTX 580, but a review copy of the game did arrive a couple of days ago. We plan to roll this one into the test suite, just as we used HAWX before it. The controversy, it seems, is that AMD feels HAWX 2 employs an unrealistically high tessellation factor, handicapping the performance of its cards. The company claims to be lobbying for a patch of some sort that'd enable a definable degree of geometry. But once the game starts shipping, with or without the patch, that's the experience available to gamers, and we'll report on it in whichever state it exists.

For now, we're using Unigine to measure the scaling performance of each architecture as geometry is ramped up.

Though AMD claims its approach to tessellation is the right one, it's pretty clear that as complexity increases, Nvidia's design maintains more of its performance. The extra SM (and PolyMorph engine) works in concert with higher clocks to give the GeForce GTX 580 a fairly commanding win over the 480 with tessellation set to Extreme in this test. The Radeon HD 5870 can't even keep up with Nvidia's GeForce GTX 470 once complexity is cranked up.

Also interesting is the fact that AMD's Radeon HD 6870 outperforms the technically more capable 5870. AMD claimed that tessellation performance was one of the attributes it worked on with the Barts GPU, and despite its shader deficiency, the newer card finishes ahead (and with a higher minimum frame rate).

Display all 185 comments.
This thread is closed for comments
Top Comments
  • 38 Hide
    TheRockMonsi , November 9, 2010 11:45 AM
    The price right now for this card is way over $500 on newegg. For that price NVIDIA better be giving me a bj as well.
  • 35 Hide
    KT_WASP , November 9, 2010 11:24 AM
    The last bit of the article is the most important I think. Anyone who drops $500 on this card right now, before Cayman releases, should have their head examined. With two companies releasing so close together, it would be in a person's best interest to see what the other is bringing to the table before shelling out such a large chunk of change.

    If the 6850 and 6870 have shown one thing.. they are much better then the last gen in many ways (power, noise and scaling) and the cayman is much more robust then the barts. So, before you start calling a winner here, wait and see. That is my advise.
  • 25 Hide
    awood28211 , November 9, 2010 11:27 AM
    Sound performance but the game here seems to be...double leap-frog. You can just release a product that competes with the competitors current offerings, you gotta compete with what he releases next... If AMD's next offering is significantly faster than it's current, then NVIDIA will still be playing catchup.
Other Comments
  • -8 Hide
    xurwin , November 9, 2010 11:18 AM
    its the beast(best)! no doubt nvidia is making a way to combat 6900's
  • 35 Hide
    KT_WASP , November 9, 2010 11:24 AM
    The last bit of the article is the most important I think. Anyone who drops $500 on this card right now, before Cayman releases, should have their head examined. With two companies releasing so close together, it would be in a person's best interest to see what the other is bringing to the table before shelling out such a large chunk of change.

    If the 6850 and 6870 have shown one thing.. they are much better then the last gen in many ways (power, noise and scaling) and the cayman is much more robust then the barts. So, before you start calling a winner here, wait and see. That is my advise.
  • 25 Hide
    awood28211 , November 9, 2010 11:27 AM
    Sound performance but the game here seems to be...double leap-frog. You can just release a product that competes with the competitors current offerings, you gotta compete with what he releases next... If AMD's next offering is significantly faster than it's current, then NVIDIA will still be playing catchup.
  • 13 Hide
    nevertell , November 9, 2010 11:44 AM
    So it's basically what the 480 should have been. Fair enough, I'll wait for the 470 version of the gf110 and buy that.
  • 38 Hide
    TheRockMonsi , November 9, 2010 11:45 AM
    The price right now for this card is way over $500 on newegg. For that price NVIDIA better be giving me a bj as well.
  • 6 Hide
    Anonymous , November 9, 2010 11:45 AM
    It'll certainly be interesting, even if i don't agree with NVIDIA playing catchup. The 480 had its flaws, but it still was the fastest single GPU around.

    We'll see what the 69xx have to offer. NVIDIA releasing now puts somewhat of a time constraint on AMD though. If it takes them too long to get something out the door, even some people waiting now may just get the 580 for christmas.
  • 20 Hide
    iamtheking123 , November 9, 2010 11:47 AM
    Looks to me that the 580 is somewhere between a 5870 and a 5970. Might have been more impressive if it was Q2 2010 and not Q4 2010.

    With ATI's meat-and-gravy bits of the 6000 series on the launchpad, you'd be an idiot to buy one of these at this price.
  • 12 Hide
    Blink , November 9, 2010 11:49 AM
    On Civ 5 benchmark the 5970 has the worst 'Zoomed Out' fps. Strange?
  • 1 Hide
    rottingsheep , November 9, 2010 11:52 AM
    i think 1 big inefficient chip trumps 2 efficient chips when it comes to performance taking into account the dual gpu scaling inefficiency.
  • 10 Hide
    rottingsheep , November 9, 2010 11:55 AM
    where's the data for 580s temp and power consumption?
  • 8 Hide
    mustbhacks , November 9, 2010 12:00 PM
    kevin1212Nvidia is embarrassed by the power draw of the gtx 580, haha. Improvement in performance but uses the same amt of power... still not a big enough improvement in efficiency, and no big leap in value either. AMD will wipe the floor with this card. By the way, i know you guys decided to drop crysis, and i can understand that, but given that this is a high end card, maybe you should have considered it, since frankly anyone buying a card like this would probably want it for crysis more than anything else. A 6870 is more than enough for the others.


    I'd say Metro more than fills Crysis's shoes.

    And to the guy above who was dissapointed by the 6850, you're insane.
  • 4 Hide
    notty22 , November 9, 2010 12:01 PM
    If the GTX 580 could talk it would probably say:
    "There can be only one "
    Fastest GPU on the planet.
  • 18 Hide
    rhino13 , November 9, 2010 12:09 PM
    Nice offering.
    I wish nVidia had had this success initially.
    But it is nice that they went back and patched things up.

    Looking forward to a comparison between the 580 and the 6970!
  • 23 Hide
    eklipz330 , November 9, 2010 12:09 PM
    how come other sites have power and heat representation?
  • -7 Hide
    dominatorix , November 9, 2010 12:14 PM
    just wait for Cayman.....but I want a card on 32 or 28 nm chip architecture, sow I' will wait even more.....
  • 4 Hide
    yzfr1guy , November 9, 2010 12:16 PM
    Power usage is in-effing-sane for higher end gpu's. What happened to the Go Green effort with pcs? :p 
Display more comments