Asus GTX 660Ti Direct CU II TOP review

Presenting Asus GTX 660Ti Direct CU II TOP



Now, before i dive into benchmarks and stuff let’s have a look at the thing we have in hand.The card we would be reviewing is the Direct CU II TOP version from Asus; currently the most aggressively binned card in the market.Let’s have a look at the specs to start things off

Graphics EngineNVIDIA GeForce GTX 660 Ti
Bus Standard PCI Express 3.0
Video MemoryGDDR5 2GB
CUDA Core1344
Shader ClockGPU Boost Clock : 1137 MHzGPU
Base Clock : 1059 MHz
Memory Clock6008 MHz (1502 MHz GDDR5)
Memory Interface192-bit
Resolution DVI Max Resolution : 2560x1600
Interface DVI Output : Yes x 1 (DVI-I), Yes x 1 (DVI-D)
HDMI Output : Yes x 1
Display Port : Yes x 1 (Regular DP)
HDCP Support : Yes
Dimensions10.7 " x 5.4 " x 1.7 " Inch



The main Difference between the reference 660TI and this card is that this card is clocked around 150Mhz higher than the referenced version with a better cooler.

Kepler is the newest GPU architecture from Nvidia and packs some nifty features. It’s vastly different from the last generation of chips from Nvidia in many ways. The foremost thing that's comes into light is that Nvidia made is a tradeoff between clock speed and core counts that all CPU and GPU makers are wrestling with every day. Power consumption rises with the clock speed, so reducing the frequency can lower the power consumption. And the second big change is a shrink in the die from 40nm to 28nm so we can expect lower power consumption and temps. There are two different Kepler GPUs in development. The Kepler1 chip, also known as GK104, is aimed at Enthusiast graphics cards and Tesla GPU coprocessors, where single-precision floating point math is most important. Nvidia has not said much about the Kepler2 GPUs – also known as GK110 internally – except that they will be tuned for double-precision floating point math and will support more GDDR5 memory, will have ECC scrubbing on that memory, will have different packaging aimed at servers, and will cost more money than Tesla cards based on the Kepler1 units. And we may see some Gaming cards based on that card(rumors off course).
Now back to the Kepler architecture here is a die shot and a block diagram of the die for the interested persons

Nvidia Kepler Die shot


Nvidia Kepler block diagram

Now if we go back in time...and take a look at Fermi architecture we will see that The Fermi GPU had 512 cores, and grouped 32 cores into something called a streaming multiprocessor, or SM. The SM had 64KB of L1 cache and a 768KB L2 cache shared across the multiple SMs. The Fermis were the first GPUs that had cache memory,making the SMs look a lot more like standard CPUs in terms of their memory hierarchy. A Fermi GPU had 16 of these SMs and either 3GB or 6GB of GDDR5 memory that they all shared.The initial Fermis only shipped with 448 cores activated in the top-end models, but as yields improved at Taiwan Semiconductor Manufacturing Corp on its 40 nanometer process, Nvidia was able to ship chips with all 512 cores running.The Fermis burned between 225 watts and 250 watts in a discrete graphics card and Tesla coprocessor cards; they originally ran at 1.15GHz with the 448 core version and were boosted to 1.3GHz with the 512 core variant. Now with the Kepler architecture Nvidia moved into something called SMX(streaming multiprocessor extreme architecture)




SM vs SMX Architecture
Nvidia is putting 192 cores into a streaming multiprocessor group with slightly modified CUDA cores. Eight of these SMX units are on a single GPU chip for a total of 1,536 cores.(normally). So basically we have a 3X power performance ratio increase on paper. Here's some prediction slides made by Nvidia

So there goes some of features of the new architecture.Lets focus on the card itself..shall we?

Showcase
Not much to say here,let the pictures do the talking;i will fill in the details
. i
it comes in the standard CU II Version box with scratches and all,the one i have has been reinforced.


Inside the box we have the custom black cardboard box by Asus which in itself looks pretty cool and classy.And yes the kiddy eraser has a huge importance in this writeup

.
Front view of the card


Back view of the card


the display connectors


the card uses 2 six pin power connectors so make sure you have enough power Showcase continues


the SLI connectors


Power LED's One of the cool features that i liked..they light up according to the state of power;Green for all systems go Red if there is some problems. here's how the work if everything is all right:

A few more pics before we dive into performance info.





Test Setups and results

First up my test setup
CPU Core i5 2500K
Motherboard P8Z68 Deluxe
GPU Asus GTX 660Ti Direct CU II TOP(driver version 306.04)
Ram Gskill Ripjaws X 1600 Mhz 9-9-9-24-1T
PSU Corsair AX 1200WSo lets carry on to the benchmarks..shall we?
First up some Synthetic benchmarks like 3DMark Suite and UNIGINE HEAVEN benchmark
3DMark3DMark is a computer benchmarking tool created and developed by Futuremark Corporation (formerly MadOnion.com and initially Futuremark) to determine the performance of a computer's 3D graphic rendering and CPU workload processing capabilities.The higher the score the better the system.As each 3DMark is based on a specific version of the DirectX API scores cannot be compared across the different releases
.
Since am an overclocker i was interested to see how much gain is seen in benchmarking suites.Overall the scores are pretty healthy.
UNIGINE HEAVEN benchmarkHeaven Benchmark is a DirectX 11 GPU benchmark based on advanced UNIGINEâ„¢ engine. It reveals the enchanting magic of floating islands with a tiny village hidden in the cloudy skies.It is the first DirectX 11 benchmark in the world, the original version was released at the moment of Microsoft Windows 7 launch in October, 2009.the version we are using is used in HWBOT.


Game benchmarks
Crysis
Crysis is a FPS series developed by German developer Crytek and published by Electronic Arts. The series revolves around a group of military protagonists with "nanosuits," technologically advanced suits of armor that allow them to gain enhanced physical strength, speed, defense, and cloaking abilities. The protagonists face off against hostile North Korean soldiers, heavily-armed mercenaries, and a race of technologically advanced aliens known as the Ceph, who arrived on Earth millions of years ago for unclear reasons, and have recently been awakened.the game uses "cryEngine" which very demanding on the GPU and had made itself an benchmark from the time of its release.

Setting:maximum playable setting with 4x AA.
Maps: Island for Crysis and Times Square for Crysis 2.



DiRT 3
DiRT 3 is a rallying video game and the third in the Dirt series of the Colin McRae Rally series, developed and published by Codemasters based on the EGO 2.0 engine.The Ego engine was developed to render more detailed damage and physics as well as render large-scale environments.

Setting: maximum playable setting with 4X AA.
Map: Aspen

Metro 2033
Metro 2033 is a first-person shooter video game with survival horror elements, based on the novel Metro 2033 by Russian author Dmitry Glukhovsky. It was developed by 4A Games in Ukraine and released in March 2010.The game is played from the perspective of Artyom, the player-character. The story takes place in post-apocalyptic Moscow.The game uses 4A Engine which supports Direct3D APIs 9, 10, and 11, along with NVidia's PhysX and also NVidia's 3D Vision.

Setting: maximum playable setting with 4X MSAA.
Map: Frontline (in game benchmark)

Sniper Elite V2
Sniper Elite V2 is a 2012 tactical shooter video game developed by Rebellion Developments.It's a remake of Rebellion's 2005 game Sniper Elite. The game takes place in the same timeframe and location—the Battle of Berlin in April–May 1945.The game uses the ASURA engine
For benchmarking we used the benchmarking demo released by rebellion.

Setting:All maxed out.
Map:Kaiser-Friedrich Museum(part of it)

The reason for the detailed graph is that i wanted to show the effect of "Super Sampling" on present generation cards.As its evident that even a card like GTX 660ti will bow down to the effect of Super Sampling.

Next up is a mix of 3 games and demo's:

  • Stalker:Call of Pripyat
  • PLA game benchmark based on Unreal 3 Engine
  • Battlefield 3


S.T.A.L.K.E.R.: Call of Pripyat
is an open world first-person shooter video game based on the X-Ray Engine.The X-Ray Engine is a game engine, supporting DirectX 8.1/9.0c/10/10.1/11 and Shader Model 5.0. Up to a million polygons can be on-screen at any one time. The engine features HDR rendering, parallax and normal mapping, soft shadows, motion blur, widescreen support, weather effects and day/night cycles. As with other engines that utilize deferred shading, the X-Ray Engine does not support anti-aliasing and motion blur with enhanced dynamic lighting modes enabled. However, a "fake" form of anti-aliasing can be enabled with the static lighting option; this format utilizes a technique to blur the image to give the false impression of anti-aliasing.

PLA game benchmark
is based on a game named Passion Leads Army using the Unreal 3 Engine.The third and current generation of the Unreal Engine (UE3) is designed for DirectX (versions 9-11 for Windows and Xbox 360), as well as systems using OpenGL, including the PlayStation 3, Mac OS X, iOS, Android, Stage 3D for Adobe Flash Player 11, PlayStation Vita and Wii U[10]. Its renderer supports many advanced techniques including HDRR, per-pixel lighting, and dynamic shadows.

Battlefield 3
is a first-person shooter video game developed by EA Digital Illusions CE and published by Electronic Arts. It is a direct sequel to 2005's Battlefield 2, and the twelfth installment in the Battlefield franchise.The game uses Frostbite 2 which is the next generation of Frostbite Engine. It takes full advantage of the DirectX 11 API and 64-bit processors,with no support for DirectX 9 so NO XP.It also features enhanced in-game destruction with Destruction 3.0[citation needed], creating more refined physics than its predecessor and quasi-realtime radiosity using Geomerics' Enlighten technology.



So that takes care of the benchmarks part.From a gamers point of view its a very good card and a strong response from the GREEN side to the RED league.

So,its time for the final verdict...In one hand we have Nvidia's latest Keplar GPU on the other hand you have Asus's perfect engineering so with the mix of that we have a perfect gaming GPU.Pro's

  • Top Notch performance.
  • Factory overclocked so you dont really need to get your hands dirty
  • Silent so noise is not going to bother you.

Con's

  • Tried a lot but couldn't find any..sorry.The guys over at Asus really did a good job with this;although i have some complaints against Nvidia but that's not within the scope of this review.

From this review onwards we are adopting to an award system for rating all the products that we are reviewing.So expect a rating in every hardware review we do from now on.Am very happy to announce that the GTX 660Ti Direct CU II TOP version gets a 5 Star rating from the ocfreaks team after seeing its performance all across the board.
So...that's all for today...time for some BF3 fragging for me.signing out...Sumon Pathak
 
Excellent review sumonpathak ..... Just a few querries (If I can pose them here)

What is the pricing of the 660Ti & how well does it compare to the plain 660 ....... [Asus Geforce GTX660 2GB DDR5 - GTX660-DC2-2GD5]?

If a price-to-performance ratio is considered then which hold's better ?
The "Asus Geforce GTX660 2GB DDR5 - GTX660-DC2-2GD5" is priced at Rs 18800/- @prime and Rs 19920/- @theITdepot

If someone is interested in the Kepler and not the 560Ti (Fermi) - which is priced max at 12-16K and performs at least 45% below the 660....then pushing one's budget & grabbing the Kepler 660 would better sense, is'nt it ?

I am inclined towards favouring the 660 Asus giving the best bang-for-the buck for the following reasons
- PCIe 3.0 based is 2X faster than the previous gen & better bandwidth
- Future-proof (way better than 560Ti)
- More CUDA cores/ texture units / 2GB GDDR5

and the 660Ti ...which I believe starts from 23K is great for "hard-core" gamers & oc'ing ...but the major price diff between 660 vs 660Ti ???

Lastly - what is the difference between the GK104 GPU core and GK106 & is this correct - that the 660 has GK106 whereas the 660Ti has GK104 & that the latter is better ? Why did Nvidia have 2 variants for the same 660 series ??

(sorry if some of my querries seem out of the scope of this review....but the 660 is in the nEWS today)

Thanks
Terry
 
1.What is the pricing of the 660Ti & how well does it compare to the plain 660 ....... [Asus Geforce GTX660 2GB DDR5 - GTX660-DC2-2GD5]?

2.If a price-to-performance ratio is considered then which hold's better ?
The "Asus Geforce GTX660 2GB DDR5 - GTX660-DC2-2GD5" is priced at Rs 18800/- @prime and Rs 19920/- @theITdepot

3.If someone is interested in the Kepler and not the 560Ti (Fermi) - which is priced max at 12-16K and performs at least 45% below the 660....then pushing one's budget & grabbing the Kepler 660 would better sense, is'nt it ?

4.I am inclined towards favouring the 660 Asus giving the best bang-for-the buck for the following reasons
- PCIe 3.0 based is 2X faster than the previous gen & better bandwidth
- Future-proof (way better than 560Ti)
- More CUDA cores/ texture units / 2GB GDDR5

5.and the 660Ti ...which I believe starts from 23K is great for "hard-core" gamers & oc'ing ...but the major price diff between 660 vs 660Ti ???

6.Lastly - what is the difference between the GK104 GPU core and GK106 & is this correct - that the 660 has GK106 whereas the 660Ti has GK104 & that the latter is better ? Why did Nvidia have 2 variants for the same 660 series ??
Thanks
Terry

1.the price AFAIK i have known is around 21K as stated by tech2..i need to confirm about it though.
2.in price to performance consideration i would take the 660ti any-day if the price is around 22.5K all even if i have to push my budget(on theory though)and it will depend on my usage.
3.in this case pushing and getting a 660 would be more sense..although depending on playing resolution one can go for the HD7850.
4.pci 3.0 is still a hype coz most single gpu setups cannot even saturate pci 2.0 bandwidth.apart from that it is future proof and packs more puch so can forget the 560ti :p
5.the diff. is because its for two different tiers of market.660 is an answer for 7850 while 660ti is put against the 7870..two entirely different performance points.
6.the main diff is in the no of transistor 2.5B GK 106 vs 3.5B in GK 104 hence is diff in performance...since the competition for this two cards are diff they decided to use diff chip(also its a tradition for them:p)

hope this helps...
cheers....
 
I really like the "precise & consise" way in which you answered my querries @sumonpathak.

One further querry (regarding Nvidia & GTX only & not ATI/AMD) ..... I've nothing personal against ATI, it's just that I need ONLY Nvidia/CUDA for specific software usage purposes :yes:

- What constitutes a "reference" card & who releases it ?
Is it always a "bare-bones" version ..... not factory-OCed; everything at "stock" levels/settings; heatsink & cooling capability is very simple & mediocre & a WHOLE LOT CHEAPER & thus not great for "stability/performance/durability"

- Does Nvidia actually make available these reference cards to the general masses ?

Do Asus/Giga/Zotac/MSI also release "reference" cards ? If YES, then how do we identify it (do they mention it specifically) ?

Sometimes, CRAZY offer's are made by certain sellers ....eg "BUY a 660 at Rs16K ONLY .... Limitted period offer" .... so are these usually reference cards ?

Is Asus Geforce GTX660-DC2-2GD5 - a reference card ?

Lastly, I would never be interested in buying a reference card - if stability/durability/cooling-capability is questionable ?
@ALPHA17 .... your inputs too.

Thanks Terry
 
Last edited by a moderator:
I really like the "precise & consise" way in which you answered my querries sumonpathak.

One further querry (regarding Nvidia & GTX only & not ATI/AMD) ..... I've nothing personal against ATI, it's just that I need ONLY Nvidia/CUDA for specific software usage purposes :yes:

- What constitutes a "reference" card & who releases it ?
Is it always a "bare-bones" version ..... not factory-OCed; everything at "stock" levels/settings; heatsink & cooling capability is very simple & mediocre & a WHOLE LOT CHEAPER & thus not great for "stability/performance/durability"

- Does Nvidia actually make available these reference cards to the general masses ?

Do Asus/Giga/Zotac/MSI also release "reference" cards ? If YES, then how do we identify it (do they mention it specifically) ?

Sometimes, CRAZY offer's are made by certain sellers ....eg "BUY a 660 at Rs16K ONLY .... Limitted period offer" .... so are these usually reference cards ?

Is Asus Geforce GTX660-DC2-2GD5 - a reference card ?

Lastly, I would never be interested in buying a reference card - if stability/durability/cooling-capability is questionable ?
ALPHA17 .... your inputs too.

Thanks Terry
1.Reference card is made by the chip manufacturers..in this case Nvidia.It is the bare bone version as u have mentioned.It is stable just dont overclock it too much:p
2.For GTX 66ti i haven't seen any reference card made for the public.
3.They release only if its allowed by the chip makers,usually they dont have any fancy coolers and stuff...
4.Crazy offers are made if only the seller has ample stocks...
5.NO
6.If you are one of those fire and forget users then ref. design should pose no problem to you.




CU II is available here for 20999 HOME | GolchhaIT.com
CU II top is different ?
CU II TOP is different.
 
I have to disagree with Sumon on the ref card part. Ref cards are the original design that tge gpu maker releases to the manufacturers and through them to the public at large. Most ref cards are as good as the oc'Ed versions. In some cases the ref card is better, and we have seen manufacturers skimping on components for thier own version cards.

thst said there are quite a few well known models that tend to oc better than ref cards.
 
hokey...so i have doing some reading.....
apparently....some..if not all special edition card offer more feature-set than the reference ones..in some area's they are better than the reference design...
Case in point the card in question...i have been reading on it...
this card is better than the reference 660ti..
its about 18-20% cooler..and the noise is lower than the reference one...
the biggest difference comes in terms of overclocking features...its boosted around 157mhz above the normal boost speed provide by Nvidia...also the the components(Gpu PWM controller and POSCAPS) are bigger and better.
and with the inclusion of DIGI+ vrm its expected to obtain more stable overclock and perform better.When a company brings out a special edition card like the TOP or Direct CU II u can be assured that its gonna perform better. @terence_fdes
 
Last edited by a moderator:
hokey...so i have doing some reading.....
apparently....some..if not all special edition card offer more feature-set than the reference ones..in some area's they are better than the reference design...
Case in point the card in question...i have been reading on it...
this card is better than the reference 660ti..
its about 18-20% cooler..and the noise is lower than the reference one...
the biggest difference comes in terms of overclocking features...its boosted around 157mhz above the normal boost speed provide by Nvidia...also the the components(Gpu PWM controller and POSCAPS) are bigger and better.
and with the inclusion of DIGI+ vrm its expected to obtain more stable overclock and perform better.When a company brings out a special edition card like the TOP or Direct CU II u can be assured that its gonna perform better. terence_fdes

Did the same here & after some reading ......... so mav2000 ..... methinks "reference" cards at least for us in India may be made available ONLY for reviewing/benchmarking purposes and to manufacturer's offering fully branded systems........
.........concluded that "reference" GPU cards are bare-boned editions and everything is at the stock or basic levels - including "heatsinks". ATI/Radeon & Nvidia do not directly make available their cards in India (just as AMD's RAM is hard to come by in India). Instead, many of these (reference cards) are directly found in Branded Laptops/Desktops - thus offering retailers (Dell/HP/Sony....) better profit margin's than them having to add a performance card from say Asus/Giga/Zotac/Sapphire etc.

I was initially confused that the Asus GTX660-DC2-2GD5 was a reference card (owing to it being the lowest priced 660 in their line-up). Now I feel assured that it's not.

Meanwhile Asus has 3 variants in the GTX 660 lineup and 5 variants in the 660Ti lineup. Thus end user's have better options in chosing as per their needs & budget. The sad truth is ALL of these would never be made available at the same time. Same is the case with Gigabyte/Zotac etc.

Asus India has still not listed the GTX 660 although cards are locally available !!!
 
Did the same here & after some reading ......... so mav2000 ..... methinks "reference" cards at least for us in India may be made available ONLY for reviewing/benchmarking purposes and to manufacturer's offering fully branded systems........
.........concluded that "reference" GPU cards are bare-boned editions and everything is at the stock or basic levels - including "heatsinks". ATI/Radeon & Nvidia do not directly make available their cards in India (just as AMD's RAM is hard to come by in India). Instead, many of these (reference cards) are directly found in Branded Laptops/Desktops - thus offering retailers (Dell/HP/Sony....) better profit margin's than them having to add a performance card from say Asus/Giga/Zotac/Sapphire etc.

I was initially confused that the Asus GTX660-DC2-2GD5 was a reference card (owing to it being the lowest priced 660 in their line-up). Now I feel assured that it's not.

Meanwhile Asus has 3 variants in the GTX 660 lineup and 5 variants in the 660Ti lineup. Thus end user's have better options in chosing as per their needs & budget. The sad truth is ALL of these would never be made available at the same time. Same is the case with Gigabyte/Zotac etc.

There are no reference cards for the GTX660Ti (the release date was a soft launch -- the partner companies already had the GPU core --> GK104 same as the GTX670 and GTX680 and had stocked themselves for release in advance), almost every company is doing their own spin on the card. The closest you can get a reference to is via the eVGA card and most reviews have noted that it overheats a lot thanks to its nVidia reference cooler design.

Reference cards are not to be confused with OEM cards, the stuff you see used in a DeLL Alienware, SONY Viao. These cards are a completely different line which is made on a contractual MoU between the two companies.For example AMD Performance RAM and Value RAM are manufactured and handled by Patriot Memory Inc. (which is not present in India) hence the availability issue.

Hope this helps, Cheerio!
 
Nice review sumon, waiting to see more. By the way, a comparison chart with the 560Ti would have been much better. WE would just know exactly about the improvements over the previous gen.
 
Great review for an old card. Can it play modern games buttery smooth? I think nvidia is now in their 9xx series, so 6xx seems quite old
 
Great review for an old card. Can it play modern games buttery smooth? I think nvidia is now in their 9xx series, so 6xx seems quite old

The series have been renamed, so stop trolling.

The 9*** series was the last 4-digit series in the nVidia line (9[SUP]th[/SUP] generation of GPU's).

After the introduction of GTX2** series (10[SUP]th[/SUP] generation of GPU's), the naming convention changed to tri-digit's and we have had the following families --
GTX 4** ('Fermi' 1[SUP]st[/SUP] generation)
GTX 5** ('Fermi' 2[SUP]nd[/SUP] generation)
GTX 6** ('Kepler', the series this card comes from)

And yes it can play most 'modern' games butter-smooth.

?????????????? Old? Compared to what, today's ebola?

:rofl: :rofl: :rofl:
 
Back
Top