Nvidia’s GeForce GTX 570 graphics processor

The torrent of new hardware pouring into Damage Labs here at the end of 2010 is absolutely stunning. Just as we were putting the finishing touches on our massive multi-GPU mash-up last week, a package arrived containing yet another video card to review. Yet that card was not what we might have expected: a new Radeon based on the forthcoming Cayman GPU. Instead, we were greeted with a black-and-green box containing a pre-emptive strike from Nvidia in the form of the GeForce GTX 570.

The GTX 570 is based on a de-tuned version of the GF110 graphics chip that powered the GeForce GTX 580 into first place in the single-GPU performance sweeps last month. We pretty much expected a slightly slower variant of the GF110 to hit the market sooner or later, but I have to admit that I wasn’t entirely excited about the prospect. We already seemed to have our value leaders in the form of the Radeon HD 6850/6870 and higher-clocked variants of the GeForce GTX 460, and we had our performance leader in the GTX 580.

Heck, though, I just wasn’t thinking it through. Here’s the basic proposition for the GTX 570: performance equivalent to—or a little better than—the GeForce GTX 480, only with power draw and a price range similar to the GTX 470’s. If you’re familiar with the current GPU landscape, you’ll know that’s a very solid proposition indeed. We’re talking about one of the fastest graphics cards around for well under 400 bucks. Obviously, Nvidia wanted to get this puppy to market in time to greet the upcoming Radeons. Those Radeons could now have some very formidable competition with which to contend.

	GPU clock (MHz)	Shader ALUs	FP16 textures filtered/ clock	ROP pixels/ clock	Memory transfer rate	Memory interface width (bits)
GeForce GTX 470	607	448	28	40	3.4 Gbps	320
GeForce GTX 480	700	480	30	48	3.7 Gbps	384
GeForce GTX 570	732	480	60	40	3.8 Gbps	320
GeForce GTX 580	772	512	64	48	4.0 Gbps	384

Let’s face it: video cards are multiplying like cockroaches these days, with different models coming out seemingly every week. When I was younger, smarter, and more wildly enthusiastic about, well, everything, I could quote specifications for each and every CPU and video card available almost instantly. These days, I’m lucky to remember to brush my teeth, so tools like the table above are a must. I’ve included both the GTX 400- and GTX 500-series cards because the silicon on which they’re based is incredibly similar. The GF110 chip in the 500 series has exactly the same number and configuration of functional units as the GF100 GPU in the 400 series. The major differences are lower power draw and doubled FP16 texture filtering rates in the GF110.

Nvidia has selectively hobbled the GF110 in creating the GTX 570, but the reductions aren’t terribly drastic. Only one of its 15 shader multiprocessor partitions is disabled, as is one of its six memory controller/ROP partition combos. The GPU’s core clock speed is down 50MHz from its elder sibling, as is the base clock frequency of its 1280MB of GDDR5 memory. The result, as I’ve mentioned, is a video card whose basic capacities match those of the GeForce GTX 480 pretty closely. We’ll explore the particulars a little further in the following pages.

One happy consequence of these cuts is the fact that the GTX 570 only requires dual six-pin auxiliary power inputs, whereas the GTX 580 requires an extra-strength PSU with an eight-pin connector. Otherwise, the 570 looks very similar to its sibling, right down to the 10.5″ board length and dual-slot profile. Under that angular cooling shroud is a vapor chamber cooler similar to the one on the 580, but Nvidia tells us the 570’s cooler isn’t quite as beefy.

Aww, isn’t that cute? Our reference card from Nvidia has made friends with a retail GTX 570 from Zotac. The Zotac card arrived yesterday morning, a bit late in the review process, but we found time to drop it into an SLI pairing to run through a subset of our usual tests. You’ll see partial results for this config on the following pages, to give you at least a taste of the GTX 570’s multi-GPU performance.

This Zotac card runs at the same base clock frequencies as the reference model, and Nvidia expects cards like it to sell for $349.99 at online vendors, with availability starting today. We may see higher-clocked versions of the GTX 570 selling at a bit of a premium, as well.

Our testing methods
Many of our performance tests are scripted and repeatable, but for some of the games, including Battlefield: Bad Company 2, we used the Fraps utility to record frame rates while playing a 60-second sequence from the game. Although capturing frame rates while playing isn’t precisely repeatable, we tried to make each run as similar as possible to all of the others. We raised our sample size, testing each Fraps sequence five times per video card, in order to counteract any variability. We’ve included second-by-second frame rate results from Fraps for those games, and in that case, you’re seeing the results from a single, representative pass through the test sequence.

As ever, we did our best to deliver clean benchmark numbers. Tests were run at least three times, and we’ve reported the median result.

Our test systems were configured like so:

Thanks to Intel, Corsair, Western Digital, Gigabyte, and PC Power & Cooling for helping to outfit our test rigs with some of the finest hardware available. AMD, Nvidia, and the makers of the various products supplied the graphics cards for testing, as well.

Unless otherwise specified, image quality settings for the graphics cards were left at the control panel defaults. Vertical refresh sync (vsync) was disabled for all tests.

We used the following test applications:

• D3D RightMark beta 4
• Unigine Heaven 2.1
• ShaderToyMark 0.1.0
• TessMark 0.2.2.
• 3DMark Vantage 1.0.2
• Aliens vs. Predator benchmark
• Battlefield: Bad Company 2
• HAWX 2
• DiRT 2
• Left 4 Dead 2
• Sid Meier’s Civilization V
• Metro 2033
• StarCraft II
• Fraps 3.2.3
• GPU-Z 0.4.7

Some further notes on our methods:

• We measured total system power consumption at the wall socket using a Yokogawa WT210 digital power meter. The monitor was plugged into a separate outlet, so its power draw was not part of our measurement. The cards were plugged into a motherboard on an open test bench.

  The idle measurements were taken at the Windows desktop with the Aero theme enabled. The cards were tested under load running Left 4 Dead 2 at a 1920×1080 resolution with 4X AA and 16X anisotropic filtering. We test power with Left 4 Dead 2 because we’ve found that the Source engine’s fairly simple shaders tend to cause GPUs to draw quite a bit of power, so we think it’s a solidly representative peak gaming workload.

• We measured noise levels on our test system, sitting on an open test bench, using an Extech 407738 digital sound level meter. The meter was mounted on a tripod approximately 10″ from the test system at a height even with the top of the video card.

  You can think of these noise level measurements much like our system power consumption tests, because the entire systems’ noise levels were measured. Of course, noise levels will vary greatly in the real world along with the acoustic properties of the PC enclosure used, whether the enclosure provides adequate cooling to avoid a card’s highest fan speeds, placement of the enclosure in the room, and a whole range of other variables. These results should give a reasonably good picture of comparative fan noise, though.

• We used GPU-Z to log GPU temperatures during our load testing.

The tests and methods we employ are generally publicly available and reproducible. If you have questions about our methods, hit our forums to talk with us about them.

Pixel fill and texturing performance

	Peak pixel fill rate (Gpixels/s)	Peak bilinear integer texel filtering rate (Gtexels/s)	Peak bilinear FP16 texel filtering rate (Gtexels/s)	Peak memory bandwidth (GB/s)
GeForce GTX 460 768MB	16.8	39.2	39.2	88.3
GeForce GTX 460 1GB 810MHz	25.9	47.6	47.6	124.8
GeForce GTX 470 GC	25.0	35.0	17.5	133.9
GeForce GTX 480	33.6	42.0	21.0	177.4
GeForce GTX 570	29.3	43.9	43.9	152.0
GeForce GTX 580	37.1	49.4	49.4	192.0
Radeon HD 5870	27.2	68.0	34.0	153.6
Radeon HD 6850	25.3	37.9	19.0	128.0
Radeon HD 6870	28.8	50.4	25.2	134.4
Radeon HD 5970	46.4	116.0	58.0	256.0

Before we go off to the races, we should set a bit of context for our evaluation of the GeForce GTX 570. Right now, there’s really no video card from AMD that competes directly with the GTX 570 at $350. The closest rival of note is probably the Radeon HD 6870, but the 6870 is $100 cheaper and is based on a much smaller graphics chip. The 6870 doesn’t match up too poorly against the GTX 570 in terms of some of the peak rates in the table above, but that’s theory. In reality, Nvidia’s GPU architectures tend to achieve performance closer to their theoretical peaks most of the time.

Another possible competitor might be a couple of Radeon HD 6850 cards paired up in a CrossFireX config. Two of those would only cost a little more than a single GTX 570, so they might be considered a viable alternative, with the proper caveats about slot real-estate, power draw, and multi-GPU compatibility issues kept in mind. You could, in theory at least, double the 6850’s rates above for a CrossFire setup. That setup should outgun a single GTX 570 in every category except for FP16 texture filtering.

However, the real threat from AMD in the GTX 570’s price range will surely be a member of the soon-to-arrive Radeon HD 6900 series, so we won’t focus too closely on the AMD-versus-Nvidia angle today. That showdown is coming soon enough.

You may have noticed in the table above that the GeForce GTX 480 has one really noteworthy advantage over the 570: memory bandwidth. 3DMark’s color fill rate test tends to be bandwidth-bound more than anything, and so we have an expected result: the 570 can’t quite match the 480 in this test.

3DMark’s texture fill test doesn’t involve any sort of texture filtering. That’s unfortunate, since texture filtering rates are almost certainly more important than sampling rates in the grand scheme of things. Still, this is a decent test of FP16 texture sampling rates, so we’ll use it to consider that aspect of GPU performance. Texture storage is, after all, essentially the way GPUs access memory, and unfiltered access speeds will matter to routines that store data and retrieve it without filtering.

The GTX 570 edges out the 480 here, which is also according to script. AMD’s sampling rates, even on its smaller GPUs, are generally higher than Nvidia’s, though, as is evident.

The GTX 570 trails the Radeon HD 6870 in our simplest bilinear filtering test, but as the complexity of the filtering method increases and the texture format jumps up to 16 bits per color channel, the 570 rises through the ranks. The results of the FP16 filtering test are telling. At 810MHz, the much less expensive GeForce GTX 460 1GB card has a higher theoretical peak rate than the GTX 570. Yet the 570’s measured performance is substantially better, likely due to its greater memory bandwidth and the larger amount (128KB more) of L2 cache associated with its five memory controllers.

Shader and geometry processing performance

	Peak shader arithmetic (GFLOPS)	Peak rasterization rate (Mtris/s)	Peak memory bandwidth (GB/s)
GeForce GTX 460 768MB	941	1400	88.3
GeForce GTX 460 1GB 810MHz	1089	1620	124.8
GeForce GTX 470 GC	1120	2500	133.9
GeForce GTX 480	1345	2800	177.4
GeForce GTX 570	1405	2928	152.0
GeForce GTX 580	1581	3088	192.0
Radeon HD 5870	2720	850	153.6
Radeon HD 6850	1517	790	128.0
Radeon HD 6870	2016	900	134.4
Radeon HD 5970	4640	1450	256.0

By virtue of its higher clock speeds, the GTX 570 has somewhat better peak shader arithmetic and triangle rasterization rates than the GeForce GTX 480. As always, the vast SIMD arrays in the Radeon GPUs yield some eye-popping numbers for peak arithmetic rates. At the same time, Nvidia’s DX11 GPUs tend to have higher geometry throughput, as represented here by rasterization rates (though there’s really more to it than just those.)

The first tool we can use to measure delivered pixel shader performance is ShaderToyMark, a pixel shader test based on six different effects taken from the nifty ShaderToy utility. The pixel shaders used are fascinating abstract effects created by demoscene participants, all of whom are credited on the ShaderToyMark homepage. Running all six of these pixel shaders simultaneously easily stresses today’s fastest GPUs, even at the benchmark’s relatively low 960×540 default resolution.

Up next is a compute shader benchmark built into Civilization V. This test measures the GPU’s ability to decompress textures used for the graphically detailed leader characters depicted in the game. The decompression routine is based on a DirectX 11 compute shader. The benchmark reports individual results for a long list of leaders; we’ve averaged those scores to give you the results you see below.

Finally, we have the shader tests from 3DMark Vantage.

Clockwise from top left: Parallax occlusion mapping, Perlin noise,
GPU cloth, and GPU particles

Overall, the GTX 570 looks quite strong, as expected, trailing only the GeForce GTX 580 among the single-GPU configs. Obviously, the Radeons sweep the first two 3DMark Vantage shader tests, which are focused on pixel shaders and seem to map well to the wide SIMD machines AMD produces. Otherwise, though, Nvidia’s largest GPUs tend to outperform today’s smaller Radeons.

Geometry processing throughput
The most obvious area of divergence between the current GPU architectures from AMD and Nvidia is geometry processing, which has become a point of emphasis with the advent of DirectX 11’s tessellation feature. We can measure geometry processing speeds pretty straightforwardly with a couple of tools. The first is the Unigine Heaven demo. This demo doesn’t really make good use of additional polygons to increase image quality at its highest tessellation levels, but it does push enough polys to serve as a decent synthetic benchmark.

We can push into even higher degrees of tessellation using TessMark’s multiple detail levels.

GPUs based on Nvidia’s DirectX 11-class Fermi architecture tend to hold up well under the most demanding geometry processing loads, and the GTX 570 in particular pretty much aces these tests. Few modern games make use of tessellation to the degree that these synthetic benchmarks do, however.

HAWX 2
We already commented pretty extensively on the controversy surrounding tessellation and polygon use in HAWX 2, so we won’t go into that again. I’d encourage you to read what we wrote earlier, if you haven’t yet, in order to better understand the issues. We have included scores from the HAWX 2 benchmark in our tests below for your information, but be aware that this test’s results are the subject of some dispute. We’re keeping this one around mostly in anticipation of AMD’s Cayman architecture potentially making things interesting.

Lost Planet 2
Our next stop is another game with a built-in benchmark that makes extensive use of tessellation, believe it or not. We figured this and HAWX 2 would make a nice bridge from our synthetic tessellation benchmark and the rest of our game tests. This one isn’t quite so controversial, thank goodness.

This benchmark emphasizes the game’s DX11 effects, as the camera spends nearly all of its time locked onto the tessellated giant slug. We tested at two different tessellation levels to see whether it made any notable difference in performance. The difference in image quality between the two is, well, subtle.

Given the results from our directed tests and, uh, whatever you make of HAWX 2, the results of our first undisputed game test aren’t exactly surprising. The noteworthy outcomes include the fact that the GTX 570 really is substantially faster than the GTX 460 1GB, even the 810MHz version we tested, and the Radeon HD 6870. If you’re stepping up to the 570, you really are stepping up. Also, going dual with a pair of 6850s will net you higher performance than the GTX 570 for nearly the same price—not really news given that we’ve found dual 6850s will outperform a GTX 580, as well.

Civilization V
In addition to the compute shader test we’ve already covered, Civ V has several other built-in benchmarking modes, including two we think are useful for testing video cards. One of them concentrates on the world leaders presented in the game, which is interesting because the game’s developers have spent quite a bit of effort on generating very high quality images in those scenes, complete with some rather convincing material shaders to accent the hair, clothes, and skin of the characters. This benchmark isn’t necessarily representative of Civ V‘s core gameplay, but it does measure performance in one of the most graphically striking parts of the game. As with the earlier compute shader test, we chose to average the results from the individual leaders.

We’ve not often seen the Radeons’ large theoretical edge in shader arithmetic performance translate into higher frame rates in real games, but I’d wager that’s what’s happening here. The GTX 570 ends up falling behind the Radeon HD 6870 as a result.

Another benchmark in Civ V focuses, rightly, on the most taxing part of the core gameplay, when you’re deep into a map and have hundreds of units and structures populating the space. This is when an underpowered GPU can slow down and cause the game to run poorly. This test outputs a generic score that can be a little hard to interpret, so we’ve converted the results into frames per second to make them more readable.

Here’s our first look at two GTX 570 cards in SLI, by the way. In this particular game, AMD’s multi-GPU scaling appears to be superior to Nvidia’s. Even though a single 6870 isn’t nearly as quick as a GTX 570, two 6870s in CrossFireX are easily the fastest overall solution tested.

StarCraft II
Up next is a little game you may have heard of called StarCraft II. We tested SC2 by playing back a match from a recent tournament using the game’s replay feature. This particular match was about 10 minutes in duration, and we captured frame rates over that time using the Fraps utility. Thanks to the relatively long time window involved, we decided not to repeat this test multiple times, like we usually do when testing games with Fraps.

We tested at the settings shown above, with the notable exception that we also enabled 4X antialiasing via these cards’ respective driver control panels. SC2 doesn’t support AA natively, but we think this class of card can produce playable frame rates with AA enabled—and the game looks better that way.

This game is also, comparatively speaking, tough sledding for the GeForce cards. The GTX 570 doesn’t distinguish itself against the much older Radeon HD 5870, which is based on a smaller chip.

Battlefield: Bad Company 2
BC2 uses DirectX 11, but according to this interview, DX11 is mainly used to speed up soft shadow filtering. The DirectX 10 rendering path produces the same images.

We turned up nearly all of the image quality settings in the game. Our test sessions took place in the first 60 seconds of the “Heart of Darkness” level.

Boy, those older Radeons are hard to shake. The two 5870s we tested again bracket the GTX 570.

Metro 2033
We decided to test Metro 2033 at multiple image quality levels rather than multiple resolutions, because there’s quite a bit of opportunity to burden these GPUs simply using this game’s more complex shader effects. We used three different quality presets built into the game’s benchmark utility, with the performance-destroying advanced depth-of-field shader disabled and tessellation enabled in each case.

Among the single-GPU options, the GTX 570 comes out looking pretty good here, especially at the two higher image quality levels. The GTX 570 brings a marked improvement over a whole host of cheaper cards based on smaller chips, including the Radeon HD 5870 and 6870 and the GeForce GTX 460 1GB at 810MHz. The 470 arguably fits into that same group of slower solutions, although it’s not based on a smaller chip.

Aliens vs. Predator
AvP uses several DirectX 11 features to improve image quality and performance, including tessellation, advanced shadow sampling, and DX11-enhanced multisampled anti-aliasing. Naturally, we were pleased when the game’s developers put together an easily scriptable benchmark tool. This benchmark cycles through a range of scenes in the game, including one spot where a horde of tessellated aliens comes crawling down the floor, ceiling, and walls of a corridor.

For these tests, we turned up all of the image quality options to the max, with two exceptions. We held the line at 2X antialiasing and 8X anisotropic filtering simply to keep frame rates in a playable range with most of these graphics cards.

The performance of the 570 is more or less up to expectations here, and the dual-GPU SLI rig scales up almost perfectly, as well.

DiRT 2: DX9
This excellent racer packs a scriptable performance test. We tested at DiRT 2‘s “ultra” quality presets in both DirectX 9 and Direct X 11. The big difference between the two is that the DX11 mode includes tessellation on the crowd and water. Otherwise, they’re hardly distinguishable.

DiRT 2: DX11

Our final game test is DiRT 2, and I had really expected to see a larger gap here between the GTX 570 and GTX 480 thanks to the GF110’s ability to filter FP16 texture formats at twice the rate of the GF100. Most likely, the GTX 570’s lower memory bandwidth holds it back a bit, though.

Power consumption
We’ll kick off our testing with power and noise. Notice that the cards marked with asterisks in the results below have custom cooling solutions that may perform differently than the GPU maker’s reference solution.

Remember what I said about the GTX 570’s power consumption being similar to the 470’s? Yeah, here’s why. Happily, these results are pretty decent, given the performance we’ve seen from the 570. We’re talking about a reduction of over 50W of total system power draw compared to the GeForce GTX 480, with extremely similar overall performance.

Noise levels and GPU temperatures

Nvidia’s custom cooler dissipates the heat produced by the GTX 570 relatively quietly while delivering middle-of-the-pack GPU temperatures. Much like the GTX 580, the cooler on the 570 sounds even better to my ears than its numbers on the sound level meter might suggest. The noise produced has a smoother, less unpleasant texture than what emanates from most other coolers in this class.

The value proposition
Now that we’ve stuffed you full of benchmark results, we’ll try to help you make some sense of the bigger picture. We’ll start by compiling an overall average performance index, based on the highest quality settings and resolutions tested for each of our games, with the notable exception of the disputed HAWX 2. We’ve excluded directed performance tests from this index, and for Civ V, we included only the “late game view” results.

With this performance index established, we can consider overall performance per dollar by factoring price into the mix. Rather than relying on list prices all around, we grabbed our prices off of Newegg where possible. The one exception was the GTX 570 itself, where we had to take Nvidia at its word about the card’s $349.99 suggested price. Here’s hoping that’s accurate!

Generally, for graphics cards with reference clock speeds, we simply picked the lowest priced variant of a particular card available. For instance, that’s what we did for the GTX 580. For the cards with custom speeds, such as the Asus GTX 460 768MB and 6850, we used the price of that exact model as our reference.

AMD card	Price	Nvidia card
	$149.99	GeForce GTX 460 768MB
Radeon HD 6850	$189.99
	$214.99	GeForce GTX 460 1GB 810MHz
Radeon HD 6870	$249.99
	$259.99	GeForce GTX 470
Radeon HD 5870	$279.99
	$349.99	GeForce GTX 570
	$429.99	GeForce GTX 480
Radeon HD 5870 2GB	$499.99
Radeon HD 5970	$499.99
	$509.99	GeForce GTX 580

A simple mash-up of price and performance produces these results:

The lower-priced solutions tend to bubble to the top whenever you look at raw price and performance like that.

We can get a better sense of the overall picture by plotting price and performance on a scatter plot. On this plot, the better values will be closer to the top left corner, where performance is high and price is low. Worse values will gravitate toward the bottom right, where low frame rates meet high prices.

Nvidia’s newest is actually pretty well positioned on the scatter plot, with only the mid-range multi-GPU solutions occupying obviously better real-estate. Note that the GTX 570 is a very straightforward improvement over the GeForce GTX 480, which has slightly lower performance yet costs quite a bit more.

Another way we can consider GPU value is in the context of a larger system purchase, which may shed a different light on what it makes sense to buy. The GTX 570 is definitely an enthusiast-type part, so we’ve paired it with a proposed system config that’s similar to the hardware in our testbed system but a little more economical.

CPU	Intel Core i7-950	$294.99
Cooler	Thermaltake V1	$51.99
Motherboard	Gigabyte GA-X58A-UD3R	$209.99
Memory	6GB Corsair XMS3 DDR3-1333	$74.99
Storage	Western Digital Caviar Black 1TB	$89.99
	Asus DRW-24B1ST	$19.99
Audio	Asus Xonar DG	$29.99
PSU	PC Power & Cooling Silencer Mk II 750W	$129.99
Enclosure	Corsair Graphite Series 600T	$159.99
Total		$1,061.91

That system price will be our base. We’ve added the cost of the video cards to the total, factored in performance, and voila:

Multi-GPU solutions occupy the top six slots in the bar chart once we factor in total system price, amazingly enough. In fact, the most expensive setup we tested, the GTX 580 SLI config, ties for the top spot. Clearly, we’ve weighted things in favor of performance more than price once we add a thousand-dollar system to the equation. In that context, the GTX 570 lands in the middle of the pack—just as it did when we factored in GPU price alone, and the other solutions shift positions almost completely.

The scatter plot is a little less fickle, and it tells a similar story to the last one. The GTX 570 isn’t an outstanding value, but it’s pretty good, especially among the single-GPU options.

These results would look very different with a more or less expensive system, so your mileage may vary.