The NVIDIA GeForce GTX 660 Review: GK106 Fills Out The Kepler Family
by Ryan Smith on September 13, 2012 9:00 AM ESTAs our regular readers are well aware, NVIDIA’s 28nm supply constraints have proven to be a constant thorn in the side of the company. Since Q2 the message in financial statements has been clear: NVIDIA could be selling more GPUs if they had access to more 28nm capacity. As a result of this capacity constraint they have had to prioritize the high-profit mainstream mobile and high-end desktop markets above other consumer markets, leaving holes in their product lineups. In the intervening time they have launched products like the GK104-based GeForce GTX 660 Ti to help bridge that gap, but even that still left a hole between $100 and $300.
Now nearly 6 months after the launch of the first Kepler GPUs – and 9 months after the launch of the first 28nm GPUs – NVIDIA’s situation has finally improved to the point where they can finish filling out the first iteration of the Kepler GPU family. With GK104 at the high-end and GK107 at the low-end, the task of filling out the middle falls to NVIDIA’s latest GPU: GK106.
As given away by the model number, GK106 is designed to fit in between GK104 and GK107. GK106 offers a more modest collection of functional blocks in exchange for a smaller die size and lower power consumption, making it a perfect fit for NVIDIA’s mainstream desktop products. Even so, we have to admit that until a month ago we weren’t quite sure whether there would even be a GK106 since NVIDIA has covered so much of their typical product lineup with GK104 and GK107, leaving open the possibility of using those GPUs to also cover the rest. So the arrival of GK106 comes as a pleasant surprise amidst what for the last 6 months has been a very small GPU family.
GK106’s launch vehicle will be the GeForce GTX 660, the central member of NVIDIA’s mainstream video card lineup. GTX 660 is designed to come in between GTX 660 Ti and GTX 650 (also launching today), bringing Kepler and its improved performance down to the same $230 price range that the GTX 460 launched at nearly two years ago. NVIDIA has had a tremendous amount of success with the GTX 560 and GTX 460 families, so they’re looking to maintain this momentum with the GTX 660.
| GTX 660 Ti | GTX 660 | GTX 650 | GT 640 | |
| Stream Processors | 1344 | 960 | 384 | 384 |
| Texture Units | 112 | 80 | 32 | 32 |
| ROPs | 24 | 24 | 16 | 16 |
| Core Clock | 915MHz | 980MHz | 1058MHz | 900MHz |
| Shader Clock | N/A | N/A | N/A | N/A |
| Boost Clock | 980MHz | 1033MHz | N/A | N/A |
| Memory Clock | 6.008GHz GDDR5 | 6.008GHz GDDR5 | 5GHz GDDR5 | 1.782GHz DDR3 |
| Memory Bus Width | 192-bit | 192-bit | 128-bit | 128-bit |
| VRAM | 2GB | 2GB | 1GB/2GB | 2GB |
| FP64 | 1/24 FP32 | 1/24 FP32 | 1/24 FP32 | 1/24 FP32 |
| TDP | 150W | 140W | 64W | 65W |
| GPU | GK104 | GK106 | GK107 | GK107 |
| Transistor Count | 3.5B | 2.54B | 1.3B | 1.3B |
| Manufacturing Process | TSMC 28nm | TSMC 28nm | TSMC 28nm | TSMC 28nm |
| Launch Price | $299 | $229 | $109 | $99 |
Diving right into the guts of things, the GeForce GTX 660 will be utilizing a fully enabled GK106 GPU. A fully enabled GK106 in turn is composed of 5 SMXes – arranged in an asymmetric 3 GPC configuration – along with 24 ROPs, 3 64bit memory controllers, and 384KB of L2 cache. Design-wise this basically splits the difference between the 8 SMX + 32 ROP GK104 and the 2 SMX + 16 ROP GK107. This also means that GTX 660 ends up looking a great deal like a GTX 660 Ti with fewer SMXes.
Meanwhile the reduction in functional units has had the expected impact on die size and transistor count, with GK106 packing 2.54B transistors into 214mm2. This also means that GK106 is only 2mm2 larger than AMD’s Pitcairn GPU, which sets up a very obvious product showdown.
In breaking down GK106, it’s interesting to note that this is the first time since 2008’s G9x family of GPUs that NVIDIA’s consumer GPU has had this level of consistency. The 200 series was split between 3 different architectures (G9x, GT200, and GT21x), and the 400/500 series was split between Big Fermi (GF1x0) and Little Fermi (GF1x4/1x6/1x8). The 600 series on the other hand is architecturally consistent from top to bottom in all respects, which is why NVIDIA’s split of the GTX 660 series between GK104 and GK106 makes no practical difference. As a result GK104, GK106, and GK107 all offer the same Kepler family features – such as the NVENC hardware H.264 encoder, VP5 video decoder, FastHDMI support, TXAA anti-aliasing, and PCIe 3.0 connectivity – with only the number of functional units differing.
As GK106’s launch vehicle, GTX 660 will be the highest performing implementation of GK106 that we expect to see. NVIDIA is setting the reference clocks for the GTX 660 at 980MHz for the core and 6GHz for the memory, the second to only the GTX 680 in core clockspeed and still the same common 6GHz memory clockspeed we’ve seen across all of NVIDIA’s GDDR5 desktop Kepler parts this far. Compared to GTX 660 Ti this means that on paper GTX 660 has around 76% of the shading and texturing performance of the GTX 660 Ti, 80% of the rasterization performance, 100% of the memory bandwidth, and a full 107% of the ROP performance.
These figures mean that the performance of the GTX 660 relative to the GTX 660 Ti is going to be heavily dependent on shading and rasterization. Shader-heavy games will suffer the most while memory bandwidth-bound and ROP-bound games are likely to perform very similarly between the two video cards. Interestingly enough this is effectively opposite the difference between the GTX 670 and GTX 660 Ti, where the differences between the two of those cards were all in memory bandwidth and ROPs. So in scenarios where GTX 660 Ti’s configuration exacerbated GK104’s memory bandwidth limitations GTX 660 should emerge relatively unscathed.
On the power front, GTX 660 has power target of 115W with a TDP of 140W. Once again drawing a GTX 660 Ti comparison, this puts the TDP of the GTX 660 at only 10W lower than its larger sibling, but the power target is a full 19W lower. In practice power consumption on the GTX 600 series has been much more closely tracking the power target than it has the TDP, so as we’ll see the GTX 660 is often pulling 20W+ less than the GTX 660 Ti. This lower level of power consumption also means that the GTX 660 is the first GTX 600 product to only require 1 supplementary PCIe power connection.
Moving on, for today’s launch NVIDIA is once again going all virtual, with partners being left to their own designs. However given that this is the first GK106 part and that partners have had relatively little time with the GPU, in practice partners are using NVIDIA’s PCB designs with their own coolers – many of which have been lifted from their GTX 660 Ti designs – meaning that all of the cards being launched today are merely semi-custom as opposed to some fully custom designs like we saw with the GTX 660 Ti. This means that though there’s going to be a wide range designs with respect to cooling, all of today’s launch cards will be extremely consistent with regard to clockspeeds and power delivery.
Like the GTX 660 Ti launch, partners have the option of going with either 2GB or 3GB of RAM, with the former once more taking advantage of NVIDIA’s asymmetrical memory controller functionality. For partners that do offer cards in both memory capacities we’re expecting most partners to charge $30-$40 more for the extra 1GB of RAM.
NVIDIA has set the MSRP on the GTX 660 at $229, which NVIDIA’s partners will be adhering to almost to a fault. Of the 3 cards we’re looking at in our upcoming companion GTX 660 launch roundup article, every last card is going for $229 despite the fact that every last card is also factory overclocked. Because NVIDIA does not provide an exhaustive list of cards and prices it’s not possible to say for sure just what the retail market will look like ahead of time, but at this point it looks like most $229 cards will be shipping with some kind of factory overclock. This is very similar to how the GTX 560 launch played out, though if it parallels the GTX 560 launch close enough then reference-clocked cards will still be plentiful in time.
At $229 the GTX 660 is going to be coming in just under AMD’s Radeon HD 7870. AMD’s official MSRP on the 7870 is $249, but at this point in time the 7870 is commonly available for $10 cheaper at $239 after rebate. Meanwhile the 2GB 7850 will be boxing in the GTX 660 in from the other side, with the 7850 regularly found at $199. Like we saw with the GTX 660 Ti launch, these prices are no mistake by AMD, with AMD once again having preemptively cut prices so that NVIDIA doesn’t undercut them at launch. It’s also worth noting that NVIDIA will not be extending their Borderlands 2 promotion to the GTX 660, so this is $229 without any bundled games, whereas AMD’s Sleeping Dogs promotion is still active for the 7870.
Finally, along with the GTX 660 the GK107-based GTX 650 is also launching today at $109. For the full details of that launch please see our GTX 650 companion article. Supplies of both cards are expected to be plentiful.
| Summer 2012 GPU Pricing Comparison | |||||
| AMD | Price | NVIDIA | |||
| Radeon HD 7950 | $329 | ||||
| $299 | GeForce GTX 660 Ti | ||||
| Radeon HD 7870 | $239 | ||||
| $229 | GeForce GTX 660 | ||||
| Radeon HD 7850 | $199 | ||||
| Radeon HD 7770 | $109 | GeForce GTX 650 | |||
| Radeon HD 7750 | $99 | GeForce GT 640 | |||
Facebook
Twitter
RSS
147 Comments
View All Comments
chizow - Monday, September 17, 2012 - link
Yes you chose to interject in this discussion and made a reference to the rebate in particular, continuing on as if the GTX 280 price was unwarranted. I think corrected you by showing the GTX 280's price *WAS* warranted relative to last-gen unlike the 7970, but even still, Nvidia cut prices and did right by their customer by issuing rebates. So, win-win for GTX 260/280 buyers, unlike this case of lose-lose for 7970/7950/7870 buyers."One comment about the rebate on the gtx 280, it's quite different from now. The 549$ radeon 7970 lost to a 499$ gtx 680 3 months after it's launch.
The 650$ gtx 280 was on average 10% better and sometimes 10% worse than the 300$ radeon 4870 one month after it's launch..."
And there you go again saying Nvidia was wrong to price the GTX 680 at $500, so you think it should be priced at $600 since it outperformed the $550 7970? And I guess the GTX 780 should be priced at $750 ad infinitum? This is what happens when you lack the perspective or understanding for a reasonable valuation or basis...I've already laid it out for you, this is why we use historical price and performance expectations....
Calling someone an idiot isn't disrespectful when they continually demonstrate a low level of intelligence and continually argue from a position of ignorance.
Galidou - Monday, September 17, 2012 - link
I never said they were wrong in their price, did I? you take words I never used, do you see in the sentence: ''Nvidia was wrong to price it so high at launch''. I was just using this to show the situation is different thus interjecting you about the fact AMD should issue rebate for the 7970 buyers and this discussion went so far that I lost the beginning of it. The only reason I mentioned those 2 facts was to show the % relavite to THIS gen comparing performance and price that's IT. So to give a short answer, the price of the 7970 wasn't so bad even after the launch of the gtx 280 as to the opposite of the gtx 280 price when the radeon 4870 launched 1 month after for less than half the price of it from a % of price difference and % of performance difference....Sorry if I'm not making myself clear at all time but this discussion is becoming so long and my english isn't as perfect as yours, french is my main language so I tried to stay as clear as possible even if I know I made mistakes when explaining my OPINION. Not the facts, I won'T say these are facts even if I took them from reliabe websites because to be a FACT I'd have to be sure a 100% of the EARTH beleives it the VERY SAME way I do thanks.
And to this day you never told me you bought an AMD/ATI card and never refuted you're not an nvidia fanboy thus proving you are. We all know when you have a choosen side, facts can be interpreted like you are doing, because the words you use are not the ones I hear from EVERYONE on earth and you can't prove everyone THINKS the way you do. It's not as simple as 2 + 2 = 4. If someone thought that every card above 500$ whatever the last gen was is wrong the the pricing of the reason of the gtx 280 pricing might not be as FACTUALLY good to everyone as you might think even before the 4870..... as for the 7970.....
Galidou - Monday, September 17, 2012 - link
''7970 wasn't so bad even after the launch of the gtx 280 ''I meant even after the launch of the gtx 680.
chizow - Tuesday, September 18, 2012 - link
You did say Nvidia was wrong to price at $500, which again is not accurate because if anything Nvidia's pricing was still too high relative to its softer than expected increase in performance relative to GTX 580.The only reason Nvidia was able to get away with this tiny increase was due to the lackluster performance of Tahiti along with its ridiculous pricing, allowing Nvidia to beat AMD this round in both price and performance with only their 2nd tier midrange ASIC GK104.
You don't seem to think this or people's buying decisions has an impact on you, but it does, and it already has. It just means you pay more for performance today or you have to wait longer for that level of performance to trickle down to your pricepoint. I've already seen it, as has every single person who bought an AMD GPU since launch. The prices today are what they should've been at launch now that the market has corrected itself (due to Kepler's launches).
As for my buying decisions again...I have owned AMD in the past a 9700pro and a 5850 for my gf. There's some integral features Nvidia offers that I know AMD is deficient in and that gap has only grown over the years to the point AMD products no longer satisfy my base expectations for graphics card purchases.
So while the two may technically compete in the same market, the products differ so much at this point for me that AMD is really no longer an option.
Some examples, since I'm sure you will ask, are features as basic as game-specific profiles and custom SLI and AA bit control. And no, AMD doesn't offer this, they just do what RadeonPro did for years by adding additional profiles without exposing the AA/SLI bits. Then there is 3D Vision support among many other less important features (PhysX, driver FXAA/AO, Vsync, better game bundles, better game support etc).
Btw, I had to end up selling the 5850 because it lacked support for something as simple as SM2.0 fur and native MSAA in Sims 3 Pets, bugs with AMD cards which my gf picked up on. That's when I threw the GTX 280 in that machine and she didn't even notice a difference (other than the new pet fur and AA). She's run GW2, Diablo3, Skyrim, and a bunch of newer games without AA at 1080p and they run great, think I could say the same for a 4870 4 years later?
Galidou - Monday, September 17, 2012 - link
"End of the discussion, you're a disrespectful Nvidia fanboy."Sorry but saying you're a fanboy wasn'T meant to disrespect you even if it was said in a harsh way and for that I'm sorry but you calling others idiots defending their honor... Nvidia fanboy in my language mean you have a choosen side and for that, you have I know you can't say you're not an Nvidia fanboy and you haven't refuted it either. Now that I did I guess it will be easier for you to say in your answer: ''Well I'm not an nvidia fanboy because of X reasons''. but if you can't say it, it will mean I was right about choosen side and interpretation of the things yuo call ''facts'' because they're interpreted by the same eyes that favor this side.
I can't read what you're saying on any website in the same exact words you're use so it has been in fact interpreted by your brain like my opinions are.
Your question is irrelevant, we were speaking of the pricing scheme at launch of their competitive parts to see if the asking price would have to force the company to issue rebates.
I'll answer you the way I remember I judged the card from my buyer perspective because I can't judge for everyone else not knowing what was in their head(I'll stop speaking like you do and say THIS IS A FACT while I don't know what OTHERS people thought in the whole world). Many of my friends back then had 8800gt because they got them dirt cheap(180$ CAD) and some of them had sli 8800gt running perfectly.
Seeing the gtx 280 at 650$ performance I was really shocked. We were in an era where Sli was becoming real popular as well as double gpu cards. And knowing you could already get easily the performance of a ''new and amazing card'' equalled on many levels by other CHEAPER solutions, I wasn't impressed but the price was totally out of what I pay for a card anyway.
Same for the 7970 I really think of both as not very good solutions. When I saw the benchmarks, I wasn't impressed at all, the only reason. I understand your point of view, the pricing of the new gen 28nm would normally drive the price back of all the generations before it while the 7xxx series instead just placed itself around to the price points corresponding at it's performance. There was NO deal but there was no CROOK either, 650$ video cards and 800$ video cards(thinking about geforce 2 and 3 series) just never made any sens to me, 550$ for a radeon 7970 don't make sense but I KNEW it had to go up someday because of AMD driving the price down WAY too much. It's just unbearable to see people whine when AMD drives the prices down too much telling they made a mistake and then whine when they drive the prices up back to normal putting all the fault on their shoulder again.....
You have to see the whole story sometimes and stop focusing on only one side of the medal IT HAD TO HAPPEN, while the 7970 wasn't priced right, 550$ to me, at launch WHATEVER the performance relative to the last gen is more acceptable than anything priced 600$ and above for gaming usage end of the line, per dollar performance was always and still remain TO ME in the 150-300$ range.
BTW you keep comparing the gtx 280 to it's last gen counter part ''8800gtx'' which it was(considering the 9xxx series was a refresh). But you keep comparing performance and price to the REFRESH of the last gen fron ATI because instead of just remaking the video cards and giving them new names, they made new more powerful parts(6950 and 6970).
If you compare the REAL last gen not refreshed parts, the 7970 would have to compare to the 5870 which it almost doubled the performance from. 300$ 4870, 380$ 5870, 550$ 7970 a return to ''normal things'' sorry if it did harm your eyes to the point you couldn'T stop remembering everyone about the 4870 SO bad PRICING and hoping they kepp it for the radeon 7970 BECAUSE when you make a mistake you cannot go back to normal after HEY? Without having some fanboys freaking out, HEY?
chizow - Monday, September 17, 2012 - link
I don't need to refute anything, I'm a fan of "whatever is better" and Nvidia products are consistently better at meeting my needs and expectations.As much as I'd love to go over all of that with you, I'm sure its just a huge waste of time, but needless to say some people don't just look at FPS charts and sticker prices for buying guidance. Luckily for me however, Nvidia is still bound by this guidance in pricing their products, otherwise I might really be paying dearly for their parts. :)
In any case, if you can't even admit 7970 pricing was far worst than GTX 280 pricing at launch, there is no point in continuing this discussion with you. I won't even bother calling you a fanboy because honestly, it has nothing do with fanboyism and everything to do with intellect, or lack thereof. These really are very simple metrics that everyone should use to make an informed buying decision.
Finally, you are right about the generational comparisons, but you can just as easily plug in the 5870 and see the 7970 is only 50% faster, ~40% faster than the 6970. Either way you can see the 7970 offers the worst increase in performance for the biggest increase in price of any new AMD or Nvidia generation or process in the last 10 years, and you really don't need to be a fanboy of either company to understand this. ;)
Galidou - Monday, September 17, 2012 - link
''Calling someone an idiot isn't disrespectful when they continually demonstrate a low level of intelligence and continually argue from a position of ignorance. ''You're not even knowing me personally, english is not my main language and you tell me I have a low level of intelligence. Who got a choosen side, who's fit to speak of both side of the medal for having PERSONALLY experimenting with BOTH companies. I even apologized for calling you a fanboy in a harsh way and you have to push the insult farther.
''7970 offers the worst increase in performance for the biggest increase in price''
That's right but that doesn't justify, compared to the competition OUT NOW, the reason to issue rebates like for the 4870 case that's all I meant from the freaking beginning... gosh it's hard. We just spoke why this is happening, AMD back in the 4870 days had to regain populatiry for being many years behind, WAY behind the pack. Right they could of priced it higher but THEY DIDN'T and good thing it put them back on the track, bad thing for now because they have to spike the prices back to normal, gosh...
Galidou - Monday, September 17, 2012 - link
''5870 and see the 7970 is only 50% faster''Well that's a little more than that, I see from 40% to 110% faster but I'll go with your 50%, not bad considering that gtx 680 is 20-25% faster than gtx580....
http://www.techpowerup.com/reviews/NVIDIA/GeForce_...
But we all know these comparisons are useless because people mostly upgrades jumping 2-3 generations, cept for heavy gamers but no need to speak or discuss for them, they already know what they want.
chizow - Tuesday, September 18, 2012 - link
I guess it depends on what review you prefer but yes, even your own preferred review shows ~50% just as I stated. So 150% performance for 150% pricing, terrible I know.Also curious as to why you single out 680 performance over GTX 580, certainly 120-125% performance for 100% of the price is better than what AMD was asking....112% of the performance for 110% of the price.
Galidou - Thursday, September 20, 2012 - link
Wait wait, AMD's 7970 price at launch was bad, but the gtx 680 will keep it's price for a while. This was a TOCK in intel's language, usually giving a huge increase in performance over last gen. The 7970 was the worst increase of performance for the worst, but Nvidia's 20-25% improve over last gen is the worst ever in history improve in performance for a TOCK in history, not speaking about price wise, just increase in performance. They went TOCK gtx 480, tick gtx 580, tick gtx 680, and we can guess next one will be a TOCK with big improve in performance or another tick with a refresh of the 600 series. And add to that the series with an automatic overclock on all their cards, still it gave out 20-25% more than last gen.... Nothing amazing there either, sure the price seems better because it's the same than last gen.