Say hello to the best “Price : Performance” graphics card in today’s market. I personally feel that buying a graphics card is one of the most technical choices when picking out the parts you want for your next upgrade or complete new build. It is less personal and more about the raw technical specs of the component, and it so happens to be one of, if not the most expensive component in your machine; thus potentially being a great burden on your budget. This is where the GeForce GTX 660Ti from NVIDIA truly shines.
At first glance
Gainward has done a really good job with the packaging of this card. The box itself comes with a window at the front hidden under the cover, as well as a carry handle. Inside, the packaging feels very sturdy and safe with two large pieces of heavy duty plastic moulds holding the card suspended in the middle to prevent any damage from physical shock during transportation.
Out of the box
The contents of the package is what you’d typically expect from a graphics card purchase; the commonly seen “HDMI - DVI”, “DVI - VGA” and “Molex - 6pin” adapters, as well as a user manual. There is the omission of a driver install disk, but I would recommend downloading the latest drivers from NVIDIA’s support page.
Gainward’s GTX 660 Ti Phantom
The main differences between the Gainward Phantom GTX 660Ti and a reference one are their Phantom cooling system, and a factory overclock to 1006 base clock, 1084 boost clock, and 6108 memory clock. A detailed comparison list can be found below.
The Phantom Cooling System is one of the most stylish non-standard cooling features I’ve seen on any card. The grill on the face of the card is metal and very robust, this makes handling of the card very easy when compared to other aftermarket cooling solutions offered by other vendors that have their fans exposed and connected on flimsy plastic frames. Underneath the grill there are two fans that are blowing cool air onto an aluminium based cooler with three copper heat pipes. These heat pipes are relatively thick and would provide excellent heat dissipation throughout the heatsink.
Oh and, did I mention the super sexy all black PCB? After all, this is the part of the card which is visible after installation!
Gainward provides “QuattroPorts” which is a 4-in-1 universal HD connectivity which is comprised of two dual-link DVI ports, a standard display and HDMI ports, so you can easily have four displays connected and running simultaneously from one of these cards.
The card comes with two SLI bridges, so 3-way or even 4-way SLI is a possibility with the GTX 660Ti. NVIDIA officially only support 3-way SLI, but that is only a driver restriction. A minimum of 2x 6pin power connectors is required, and runs at about 140W on load.
On to testing the car, the specifications of the test rig that I used to run this card are as follows:
Since this card is aimed at gamers, I cannot think of a better game to test the capabilities of any card in today’s market than Battlefield 3. When running on Ultra settings, this game can be very intense on any GPU and can prove to be a tougher benchmarking program than most GPU stress-tools. So many time I’ve hit a stable overclock with previous cards on GPUtool, MSI Kombustor, and have the card crash later on when try running it through Battlefield 3.
The single player testing was undertaken in the same way most Battlefield 3 benchmarks are done; that is multiple run throughs of a 1-2 minute section of certain single player campaigns missions. However it should be noted that singleplayer benchmarks will almost always constitute higher values than multiplayer as you will see further in the testing. Here below you can see first the campaign run in ultra in-game presets, and then in high presets. All testing was done with 4x Anti-Aliasing and 16x Anisotropic Filtering.
One possible conclusion which can be drawn from the above results is that this particular card naturally runs better at the “high” in-game presets. As you can see above, the minimum and maximum frames per second don’t vary as much as the average FPS numbers. The similar minimum FPS results can be attributed to the fact that anti-aliasing and anisotropic filtering were held constant on their maximum possible settings, so the high particle explosions and high volume collisions would constitute similar dips in performance. Whereas the similarity in maximum FPS can perhaps be explained by the limited bandwidth of the GTX 660Ti, where lowering the video quality doesn’t necessarily cause a higher volume of data to be transferred through.
What is most interesting is the large increase in mean average FPS which is almost 10 frames higher in the high presets, this is the value you’re going to be seeing during most of your fragging hours, so it is of more importance than the maximum frames in my opinion. That being said, the most important figure here is the minimum FPS, and you can clearly note that the minimum dip in both presets are nothing that would render the game unplayable.
The multiplayer testing was comprised of running on two different settings, Ultra and High, both with 4x Anti-Aliasing and 16x Anisotropic Filtering. Each setting was run three times through a 30 minute play-through on the map “Caspian Border” 64-player with the server at about ¼ - ½ player capacity at late times in the night.
Results and recommendations
From the graphs above, you can see the GTX 660Ti performing admirably on ultra settings in Battlefield 3. The thing to note here is the mean minimum FPS which is 39.7 exactly which is a very decent number for minimum FPS and happens to be higher than its direct competitor, the HD7950 from other reported benchmarks.
Based on my experience with running the card in Battlefield 3, I would suggest using the High presets; as you can see its mean avg FPS is almost 10 FPS higher than Ultra, and you are not compromising too much with video quality since you are leaving Anti-Aliasing and Anisotropic Filtering at the maximum.
A couple of screenshots of high intensity explosions on both ultra and high settings shows just how little difference there is between the two when keeping AA and AF at the maximum - pay attention to the foliage and character model details on the gloves since they are easier to determine than unpredictable explosions.
A detailed account of the graphics settings NVIDIA recommends can be found here: http://www.geforce.com/optimize where you can see the average frames per second NVIDIA reportedly gets with each of their cards for different resolutions.
As for temperatures, the Gainward Phantom cooler proved itself to be very efficient. Idle temperatures were around 35 degrees celsius and load temperatures didn’t go above 62 degrees celsius. It being a hot night with ambient temperatures close to 20 degrees celsius, in a non-airconditioned room, I’d say those numbers are very attractive.
So what have we learned? Well, if a friend came to me tomorrow and asked what I’d recommend for a new computer he’s building mainly for gaming, or a budget upgrade on his/her current machine, I would tell them to look no further than the GTX 660Ti from NVIDIA; here is a company who truly care about their products, they do not just create works of technological art and push it out to the shelves, they also maintain their products with one of the best driver track record in the computing world. Looking just at the quality of drivers and future support NVIDIA is the only choice for me personally and the only recommendation I am willing to put forth. A comprehensive driver support feature can be accessed from NVIDIA’s own website here: http://www.geforce.com/drivers
As for this particular GTX 660Ti from Gainward, showcasing once again the effectiveness and efficiency of their Phantom cooling system and boasting very impressive idle and load temp in my testing above - if I were to buy a non-reference card, this would one of the very few aftermarket cooled GTX 660Ti I would consider.