BabelTechReviews is at Nvidia’s GPU Technology Conference for the next four days. Nvidia’s CEO Jensen has just fully unveiled and hard launched for sale the new DX12 Level 1, fully-enabled, 8 billion transistor Maxwell flagship GTX TITAN X at $999. With 3072 CUDA cores and 12GB of VRAM, it promises to be significantly faster than the GTX 980, and much faster than AMD’s aging flagship, the R9 290X.
We are testing all of our competing cards on a clean installation of Windows 8.1 using resolutions of 1920×1080, 2560×1600 and at 4K’s 3840×2160. As befits testing top video cards, we use Intel’s enthusiast Z97 platform with Core i7 4790K turboed to 4.4GHz, and 16GB of Kingston’s 2133MHz Predator DDR3.
Before we give you the results of our performance testing, we want to briefly recap Maxwell architecture, as well as detail the specifications and features of the new GTX TITAN X.
Key Features of the Maxwell GTX TITAN X
The GTX 980 and the GTX 970 both launched last September as Maxwell GM204 architecture , and both of them are faster than AMD’s flagship, the R9 290X. The GTX 960 is a smaller chip also based on the Maxwell GM206 architecture and it was released in January at the 200 dollar price point. The GeForce GTX 980, 970 and GTX 960 GPUs support all-new graphics features formerly available only to them, and are now available on the GM220 GTX TITAN X. Nvidia’s Voxel Global Illumination (VXGI) technology allows the new GPUs to render fully dynamic global illumination at playable frame rates bringing more realism and immersion to gamers. Although it is not real-time ray tracing, it a good step in that direction.
PC games can also perform and look better with new anti-aliasing modes like Multi-Frame sampled AA (MFAA) which works particularly well with 4K resolution. MFAA combines multiple AA sample positions to produce a result that looks like higher quality anti-aliasing, but with better performance. From our testing of MFAA, it appears to produce an image that looks very similar to 4xMSAA at the performance cost of roughly 2xMSAA. Until the TITAN X launch now, MFAA was only available to the GTX 980, the GTX 970 and the GTX 960.
The new GeForce Maxwell GPUs also support Dynamic Super Resolution (DSR) which is similar to driver-based SuperSampling. DSR brings the crisp detail of 4K resolution to 1920×1080 displays. It looks great, but without a FCAT capture, screen shots cannot be shown with Fraps. These Maxwell GPUs retain and improve on features like ShadowPlay, which now support recording at resolutions up to 4K at 60 fps. And with the new G-SYNC displays, gamers no longer have to put up with tearing or stutter as part of the current common gaming experience.
The next generation of games will not only look better and run faster on the GeForce GTX 9X0s and on the TITAN X, they’ll also be more immersive thanks to virtual reality headsets like the Oculus Rift. With VR Direct, Nvidia has developed a number of advancements for virtual reality reducing latency, improving image quality, and bringing a whole range of new content to VR.
This editor has already experienced an Oculus Rift demo created from Unreal Engine 4’s Infiltrator assets at Nvidia’s 2014 Maxwell Press Event that is an awesome extension of S3D. VR frame rates need to be locked to a minimum of 75 fps for fluidity – and they need to be rendered twice, once for each eye. Eventually, as the resolution increases, the grainy look will disappear allowing for more realism, but at the price of requiring extreme graphics performance from the video card or cards like the TITAN X. We expect to check out the new VR demos at the GTC and we will report on them in our GTC wrap-up article.
The Maxwell GM204 GTX 980 versus the GM220 GTX TITAN X
First, take a look at the GM204 block diagram which represents the GTX 980 – and until today’s TITAN X launch – the world’s fastest GPU:
The GTX 980 GM204 has 64 Raster Operating Units, double Kepler GK110’s 32.
There are 5.2 billion transistors packed into the GM204’s die size of 398 mm2.
The GTX 980 has only a TDP of 165W and the GeForce GTX 9x0s have HDMI 2.0 support and support for 4 displays
Now check out the differences by looking at the GM220 block diagram of the fully enabled 8-billion transistor Maxwell GM220 with 24 SMX units and 3072 CUDA cores:
There is quite a difference between the TITAN X and the GTX 980 besides the size of the GPUs. The base clock of the GTX TITAN X is 1000MHz with a typical boost clock of at least 1075MHz. The TITAN X texture filtering is performed by 192 texture units leading to 192 gigatexels/second, or more than one third faster than the GTX 980. And GM200’s memory subsystem is also much more robust in the TITAN X by giving 335GB/second, or more than 50% more peak memory bandwidth over the GTX 980 by using a 384-bit memory interface (six 64-bit memory controllers) with 7GHz memory clocks.
One key difference is the 12GB of vRAM used in the GTX TITAN X, compared with 4GB used in the GTX 980. This means that the TITAN X will not run out of video memory at 4K even with multiple displays. And we found at least one game that will simply not run at the extreme settings on the GTX 980 at 4K, but will run fine on the GTX TITAN X.
The display/video engines are unchanged in TITAN X from the GTX 980 and the overall double precision instruction throughput is 1/32 the rate of single precision throughput in both GPUs. The GTX TITAN X is primarily built for extreme GeForce gaming and is by no means a budget card. Price conscious video gamers will most likely choose a pair of GTX 970s instead. For the extreme enthusiast, no doubt two, three, or four GTX TITAN X video cards may be used together in SLI to power multiple 4K displays, something the SLI’d GM204 GPUs cannot handle.
The GTX TITAN is a regular-sized 10.5″ video card with a TDP of 250W using one 6-pin and one 8-pin PCIe connectors. Display outputs include one Dual Link DVI, one HDMI 2.0 connector for use with the latest HDTVs, and three DisplayPort connectors. There is no backplate on the TITAN X as there is with the GTX 980 because the TITAN X is a much warmer-running card, and even a removable black plate cover is impractical on the reference design because of the screw length.
The TITAN X continues the tradition of Nvidia’s original TITAN Industrial Designer look, but in black matte painted aluminum. It uses a copper vapor chamber combined with a large aluminum dual heatsink to dissipate the GM200’s heat. Unlike its competitor, AMD’s R9 290X, the TITAN X is very quiet, reaching only about 3300 rpm under full load and topping out with a peak temperature of 87C under the very hot Summer-like conditions of our testing lab.
Southern California has been having a heatwave with record temperatures over the past weekend, and we have had to run our 290X late at night to prevent its throttling. We found no such throttling issues with the TITAN X, even when ambient temperatures surpassed 80F in our testing lab!
Here are the specifications for the GM204 GTX 980:
There were quite a few changes between the GTX 980 and the GTX TITAN X. Everything that applies to GM204 also applies to GM220 and it may be worth checking out our September GTX 980/970 launch article.
How does the GTX TITAN X compare with its rival, AMD’s R9 290X?
It doesn’t. Although the GTX 980 is mostly faster than the 290X, TITAN X is simply in a class above the 290X and even over the GTX 980. How far? By just looking at the specification, it looks like the GTX TITAN X will be about 30 to 50% faster than the GTX 980 and also significantly faster than the 290X. We are going to look at the performance of 30 games to compare the TITAN X with the GTX 980 and the R9 290X.
However, before we do performance testing, let’s take a closer look at the GTX TITAN X and check out overclocking and noise.