Lab Testing Nvidia’s Holiday Gift: The GTX 570
The first feature-reduced version of the GTX 580 arrives, rendering the GTX 480 obsolete and body-slamming the Radeon HD 5870.
This is the silly season for PR presents. Technology writers and product reviewers receive boxes in the mail, sometimes elaborately giftwrapped, from public relations people in the industry. Usually, what we find inside are fruit, chocolate, calendars with generic photographs and assorted pastries. So when we got a gift box from Nvidia, we naturally thought it was one of the usual holiday PR gimmicks.

We were wrong. When we got around to opening the box, we found this.

This is the follow-up to the GeForce GTX 580. Unsurprisingly, it’s called the GTX 570. As with the earlier GTX 470, it’s a cut-down version of the mother chip, offering 480 compute cores instead of the GTX 580’s 512 cores. Other features have been scaled back as well.

What’s obvious by comparing specs is that the GTX 570 is pretty much a juiced up GTX 480, with a couple of minor differences. Memory interface width is narrower and the 570 has fewer ROPs. Plus, the GTX 570 ships with 1.2GB of GDDR5, which runs at a higher clock speed than the GTX 480.
Nvidia’s suggested price for reference level GTX 570s will be $349. Factory overclocked cards will likely cost a little more. Better warranties also tend to push prices up slightly. That also puts the GTX 570 squarely in the price range occupied by AMD’s Radeon HD 5870.
From the specs, it looks like the GTX 570 makes the GTX 480 obsolete. Let’s see if the reality lines up with the theoretical.
Performance
After installing the GTX 570 into our reference graphics system, we fired it up, installed the driver, and ran our entire benchmark suite. All tests were run at 1920x1200, with 4x AA enabled. Our test system consisted of a Core i7 975 at 3.3GHz, with 6GB of DDR3/1333 memory, running on an Asus P6X58D Premium motherboard, with a Seagate 7200.12 1TB drive, an LG Blu-ray ROM drive, a Corsair TX850w 850W PSU, and Windows 7 Ultimate 64-bit.
What we found was a category killer. At the $349 price point – assuming you can actually buy cards at that price – the GTX 570 simply crushes everything else. The only cards that beat it was the GTX 580 and the dual GPU Radeon HD 5970. But nothing within shouting range of $350 could touch it.
3DMark Vantage and Unigine Heaven
Let’s take a quick perusal at a couple of synthetic benchmarks. We don’t put much weight to these results, but it’s interesting to check them out.


With a new 3DMark now on the scene, this is probably the last time we’ll report 3D Mark Vantage scores. While the hard-to-find Radeon HD 5970 comes out on top, the GTX 570 is the only card under $400 that breaks 20,000 for the final score.
The GTX 570 is also in a dead heat with the GTX 480 in Unigine – not really that much of the surprise, given the relative similarity between the two cards.
DirectX 10 Gaming Performance
The games we tested for DX10 performance include Far Cry 2 (two different scenes), Just Cause 2 (the Concrete Jungle benchmark), Tom Clancy’s HAWX and the aging, but still gorgeous, Crysis.





While the GTX 570 loses to the older GTX 480 in a couple of benchmarks, it’s performance envelope is really quite similar. The card also leaves all the AMD cards gasping in its wake.
DirectX 11 Gaming Performance
The DX11 games we tested are a mixed bag. Some, like the recently released Tom Clancy’s HAWX 2 and Metro 2033, make heavy use of DX11 features. HAWX2, in particular, uses DX11 hardware tessellation very heavily. Others, like BattleForge, DiRT2, Aliens vs Predator or Call of Pripyat, use DX11 features a little more judiciously.






HAWX 2 uses tessellation in an extreme way, but the result is gorgeously rendered, near-photorealistic landscapes. Fermi’s ability to tessellate and render down to very small meshes plays very well in this test.
Metro 2033 was also interesting, mostly because of how poorly the single GPU Radeon HD 5870 fared. This result was repeatable, and we’re not quite sure what’s going on, since the newer HD 6870 managed a reasonable, if low score.
At the suggested $350 price point, the GTX 570 once again leaves its competitors in the dust.
Power
Nvidia’s new card ships with a pair of 6-pin power connectors, so we expect it to be a little more efficient than the GTX 570. And it is – a little.

At idle, the GTX 570 actually consumes a little less power than XFX’s Radeon HD 5870 XXX edition. However, under a heavy rendering load, power consumption climbs well beyond AMD’s cards, but does fall a little shy of the older GTX 480 and the high end GTX 570.
Final Thoughts: Almost in the Sweet Spot
At $350, the GTX 570 occupies what we might call the lower part of the high end graphics card segment. That price still makes many users pause, and the real sweet spot is still in the $250 category, which is currently ruled by the AMD Radeon HD 6870.
That said, the GTX 570 is considerably cheaper than the GTX 580, and should even come in lower than many currently shipping GTX 480 cards. It’s pretty much on par in terms of performance with the GTX 480, and is a bit more power efficient to boot. While Radeon HD 5870 cards cost slightly less, the difference will be minimal, and there’s no reason you shouldn’t get a GTX 570 over an HD 5870. The GTX 570 is even a little smaller, at a true 10.5 inches, and will fit comfortably in many midrange towers.
The real question is availability and pricing. It’s unlikely we’ll see a broad swath of these cards shipping before the holidays. And it’s likely that any cards that do ship will quickly be snapped up. How many of those products will actually be available at the $350 price point is open to question. Yes, the GTX 570 is a fast card, but if actual prices get too close to the $400 mark, then it becomes much less attractive. Let’s hope that Nvidia’s real holiday surprise is the ability to buy a GTX 570 at the suggested price. That would be a welcome stocking stuffer.
![]()
ASHOK BISHNOI
December 31, 2010 at 6:55am
The battle between AMD and Nvidia is getting fiercier...and both are trying to bring new cards to get market share and clinch each other...This one is nice Graphic Card but it has certain issues...Still I am hoping for a best PERFORMING card.
![]()
ASHOK BISHNOI
December 31, 2010 at 6:52am
The battle between AMD and Nvidia is getting fiercier...and both are trying to bring new cards to get market share and clinch each other...This one is nice Graphic Card but it has certain issues...Still I am hoping for a best PERFORMING card.
![]()
ASHOK BISHNOI
December 31, 2010 at 6:52am
The battle between AMD and Nvidia is getting fiercier...and both are trying to bring new cards to get market share and clinch each other...This one is nice Graphic Card but it has certain issues...Still I am hoping for a best PERFORMING card.
![]()
darkliquids
December 31, 2010 at 5:55am
Look at it! It does not start the values at 0 but at 58. look at the numbers 65 looks like it is half of 73. Excessive gap. Visually, I find it deceptive. All other graphs are fairly represented.
![]()
Biceps
December 14, 2010 at 9:19am
Got me one of these, and it's purty! Switched in a single 570 in place of 2 8800GTs in SLI, and man, what a difference. The first thing I noticed was how quiet the fan is - can hardly hear the 570 after the double-jet-engine-scream of the dual 8800s. The performance is fantastic, too.
Happiness is a warm Nvidia card.
![]()
aferrara50
December 12, 2010 at 5:29pm
Now what really matters is how these overclock under water, not stock settings. There really aren't many people that buy high end cards and don't watercool / aftermarket aircool them. They are way too noisy and hot. Not doing that would be like using he stock cooler for ur cpu... it's just wrong... wrong in so many ways. They need to benchmark how these scale in tri sli since that is the real sweetspot. The main negative of the 570 is that it puts out less performance per mhz than the 480. Heat and noise should be a nonissue for high end cards. They should really all be sold with high end aftermarket coolers/ blocks already installed. Would be a cold day in hell before I toss 3 of those in my rig without wb on them.
![]()
Trooper_One
December 08, 2010 at 12:36pm
I wish in future reviews, you (actually many other reviewers) would put the result of a benchmark from at least one card from a previous generation or two, just for comparison's sake.
Not everyone can keep up to date from one gen to the next. Some are actually one, two or more years behind. It would be nice not having to dig up previous articles just to see how the cards used to preform.
![]()
Nimrod
December 08, 2010 at 2:54am
The comentary about the results and spects couldnt possibly more far off. This is one of the most biased and fake reviews ive seen from MaxPC in a long time.
![]()
Keith E. Whisman
December 08, 2010 at 10:59am
Your name says it all.. Dude they don't bake the benchmarks. Just look at the hardware layout compared to the GTX480. It's pretty much the same card but with a few cuts here and a few additions there to improve performance. Don't be a tool or a freaking Nimrod. You probably made accusations like this when MaximumPC showed how much more powerful the Intel 980X if compared to AMD's fastest 6 core processor. You are the first AMD fan boy to scream about obvious benchmark results and comparisons. Back in the day fan boys from both sides would scream about benchmarks when either side had a difference of just a few FPS. Now times have changed and we are talking more than 15FPS difference and that is significant especially considering the prices involved.
![]()
Keith E. Whisman
December 07, 2010 at 4:13pm
Can you run the GTX570 in three way SLI? I was expecting good performance on par with the 480 but I didn't expect it to utterly destroy the 480 like it did. Well I'm exaggerating a little bit, but the scores don't lie. I wan't three of these cards in a box with an 980X CPU.
So Can it do 3way sli and when is Intel going to release a 8 or more core processor? Is intel waiting for AMD to unleash it's bulldozer?
![]()
aferrara50
December 12, 2010 at 5:38pm
intel 8 cores will be out in Q3/4 of 2011. They are Tri sli compatible, but I would suggest never using AMD cpus with nVIDIA cards. They just don't work as well together and the benchmarks prove that, generally on vantage and heaven. Take a look at some. The amount of cores of the cpu doesn't really matter, it is the speed of each core that does. When playing a game you will see better performance with a 920 clocked at 4.6ghz than a 980x @ 4.6ghz because each core is clocked much higher. This will be true until games use more than 4 cores at a time, most only use 1 or 2. Anything faster than a 580 is pointless due to bottleneck which you will have unless you drop $2500 on a ballin wc setup. I would personally never trade my 3 480s for 3 570s since they aren't as great of overclockers. 580s on the other hand I def would.
![]()
americaeh
December 07, 2010 at 3:49pm
throughout the whole thing you refere to a "gtx 6870" when its really a amd radeon hd 6870.....
![]()
vistageek
December 07, 2010 at 4:04pm
I was just noticing that and I was about to comment about it. You beat me to it.
![]()
QuadraQ
December 07, 2010 at 2:36pm
This is a great card, but I'm waiting for the release of the GTX 560 so I can run two of them in SLI for the new rig I'm building.
![]()
Neufeldt2002
December 07, 2010 at 1:06pm
My only question: What are the temps that these cards run at?
![]()
jackal49
December 07, 2010 at 8:18pm
Go to Tom's Hardware. I love MaxPC, but they usually leave out numbers like this. http://www.tomshardware.com/reviews/geforce-gtx-570-gf110-performance,2806.html
![]()
Keith E. Whisman
December 08, 2010 at 11:25am
What are you talking about? Toms Hardware benchmarks pretty much reflect whats in this article. Hell in most benchmarks the 570GTX beat the 6850Crossfire setup. I'm sure you didn't mean it but it seems your post was trying to show up MaxPC as trying to hide something that after I read the toms hardware article you linked doesn't show any benchmark scores that differ from those above. You do realize that different organizations test things differently.
I think it's Toms Hardware or is it HardOCP that refuses to do an Apples to Apples comparison of it's hardware while benchmarking. They generally make the settings the best for that video card and then benchmark it and compare the numbers to other video cards that they themselves have their setting and resolution set for optimal performance.












