43660_Featured

Top 5 GPUs to power your 3D, VFX or game dev computer

‌‌
Over the past few years, more and more software programs have started taking advantage of the GPU on your computer. This is especially true in the world of 3D, where you can harness the power of your GPU to speed up processor-heavy tasks such as rendering or simulating. For example, mental ray has been around for over two decades as a CPU-based renderer, but lately NVIDIA has started to implement features that use the GPU. And since NVIDIA owns mental ray, it only makes sense that NVIDIA will focus its energy on developing technology specific to its own cards. Those are important things to keep in mind when you're looking to buy a new graphics card. Depending on the tool you’re using, sometimes you may need to tell it to use your GPU. In RealFlow you’ll need to go into your simulation options to enable OpenCL before it’ll calculate with your GPU. On the other hand, tools such as Octane and FurryBall don’t require any adjustment in the settings since they only use the GPU anyway.

GPU vs CPU

Although you’re not going to see any differences in the result of a render done on the GPU or the CPU, you will see a difference in the time it takes to get those results. This is simply because of a few key distinctions between the way your CPU and GPU are designed to work.

Shared processes

From your operating system to your browser, all of your computer programs are run through your CPU. You can open up your Task Manager and go to the Processes tab to see a list of the things your CPU is processing right now. So when you tell your 3D program to render or crank out a simulation, if that render or simulation is running on your CPU the system has to balance the processing of the render alongside all of the other programs that are running. Since other programs don’t run in the GPU, if you’re rendering or simulating using your GPU it’s able to focus all of its power on the task at hand without other programs needing to be processed.

Dedicated memory

Another way your GPU can help speed up your workflow is because the RAM on your graphics card is typically dedicated to anything you’re rendering out on your GPU. This is a lot like shared processes; your 3D program doesn’t have to share system RAM with other applications when it’s rendering on the GPU. Keeping both the processes and RAM dedicated to calculating your renders, simulations or whatever else you’re calculating on the GPU will also free up your CPU and system RAM for the other programs on your system.

More cores means more power

If you’ve looked into buying CPUs, you’ll know they generally have multiple cores. For example, the Intel i7 series chip has two cores with four threads. On the other hand, most current-generation graphics cards have thousands of cores. For NVIDIA cards, they use CUDA cores while ATI cards uses STREAM technology. While there’s the obvious benefit of more, dedicated cores to process information, in addition to that, most 3D software is written to take advantage of these GPU cores to optimize the rendering or simulation process.

Top 5 GPUs from the community

Those are just a few benefits of using your GPU over your CPU for graphics-heavy calculations. There’s more, such as sequential processing vs parallel processing, but I think you get the idea. With all of this in mind, which GPU you use for your creative projects is very important. It can mean the difference between a render taking hours or minutes which, in turn, can mean a project’s success or failure. For that reason, we asked our community to vote on their favorite GPUs. Here are the top five results, ordered from the most popular to least.

NVIDIA GeForce GTX TITAN X

43660_Inline_01 While it’s hardly the cheapest option, the TITAN X was by far the favorite GPU. With its eight billion transistors, 12 GB of dedicated RAM and over 3,000 CUDA cores it’s easy to see why. If you’re a tech geek, you’ll start drooling over some of the specs on the TITAN X. A majority of the votes we received for the TITAN X came from game developers and 3D artists. One level designer noted that the TITAN X allows them to have multiple things open at once, such as a 3D program, Photoshop, a real-time rendering program (such as Marmoset Toolbag), Visual Studio, over 30 tabs in Chrome and still run Unreal Engine 4 with maximum graphics and getting 100+ FPS. Even with a hefty price tag of around $1,000, the TITAN X shows it’s worth the cost to 3D artists with large-scale projects. As one 3D artist stated, “The 12GB of VRAM puts almost any scene complexity in reach for Octane render and also TurbulenceFD loves the RAM. To get that capacity anywhere else you need to jump to the meteoric pricing found with the pro cards (Quadro/Tesla).”

NVIDIA GeForce GTX 970

43660_Inline_02 The second most popular card in our survey was the GeForce GTX 970. While the specifications for the GTX 970 aren’t as powerful as the TITAN X, the cost isn’t nearly as high either. Most of the votes for the GTX 970 came from 3D and VFX artists who complimented the 970 for being fast and cheap. With the lower price point, a lot of the 3D artists who nominated the GTX 970 also made mention of an SLI option to harness two GTX 970s together for even more power.

NVIDIA GeForce GTX 980

43660_Inline_03 Interestingly the GeForce GTX 980, which as you might imagine is slightly more powerful – and expensive – than the GTX 970. While the GTX 970 has the same amount of RAM as the GTX 980, the 980 has 2,048 CUDA cores compared to 1,664 on the 970. Those extra cores, as we talked about earlier, are enough to provide an impressive boost to your GPU rendering or simulating projects. Most of the nominations for the GTX 980 came from 3D generalists, game artists or motion designers. And like the nominations for the 970, there were quite a few artists who preferred a GTX 980 SLI over just one 980.

NVIDIA GeForce GTX 750 Ti

43660_Inline_04 At around $150, the GTX 750 Ti is the cheapest card in the top five. An important distinction to make with this card is that it’s not the base GTX 750, which only has 512 CUDA cores and 1 GB of RAM. The GTX 750 Ti, on the other hand, has 640 CUDA cores and 2 GB of RAM. Although the difference may seem slight, the 3D generalists who nominated this card clearly saw enough difference in its performance to warrant the distinction. Another great aspect of the GTX 750 Ti that makes it somewhat unique among the powerful cards in this list is that it doesn’t require nearly as much power. For example, the TITAN X requires 600 W of power whereas the GTX 750 Ti only needs 300 W. That can save even more money when building your system because you probably won’t need to buy a specialty power supply just to help support your graphics card.

NVIDIA GeForce GTX 780

43660_Inline_05 Rounding out the top five cards is the GTX 780, which offers 2,304 CUDA cores along with 3 GB of RAM. Like the 900 series cards in this list, a lot of the nominations for the GTX 780 mentioned an SLI for even more power. Most of the nominations were from 3D and VFX artists. As one of the 3D artists who voted in our survey pointed out, the 2,304 CUDA cores match the rendering speed of more expensive cards at a fraction of the cost. If you’ve noticed, NVIDIA cards dominated the top five. With the top ATI card being the FirePro W7000, which came in 10th place, it’s clear that creatives around the world prefer NVIDIA cards. And no wonder, as studios such as Framestore, ILM, Pixomonodo and WETA Digital power their hardware with NVIDIA technology as well. Here's the top twenty cards from our survey.
  1. NVIDIA GeForce GTX TITAN X
  2. NVIDIA GeForce GTX 970
  3. NVIDIA GeForce GTX 980
  4. NVIDIA GeForce GTX 750 Ti
  5. NVIDIA GeForce GTX 780
  6. NVIDIA GeForce GTX TITAN
  7. NVIDIA GeForce GTX 680
  8. NVIDIA Quadro K4000
  9. NVIDIA GeForce GTX 770
  10. AMD FirePro W7000
  11. NVIDIA GeForce GTX 780 Ti
  12. NVIDIA GeForce GTX 550
  13. AMD Radeon R9 280X
  14. NVIDIA GeForce 9400 GT
  15. NVIDIA GeForce GTX 660M
  16. NVIDIA Quadro K4200
  17. NVIDIA Quadro K5200
  18. NVIDIA Quadro M6000
  19. NVIDIA Quadro K6000
  20. NVIDIA GeForce GTX 675
As technology is always changing and more and more software starts to take advantage of GPU, there is no right or wrong graphics card to use. It all boils down to what you need it for, how much power you need to get out of it and what your budget is. Now that you've seen what our community nominated, you can vote for your overall favorite. Since we're really just focusing on the top five, use the poll below to cast your vote: [gravityform id="22" name="2015-05-Favorite GPU Poll" title="false" description="false"]   What are your thoughts of our survey? Was your favorite GPU not in the list? Let us know why you think it should've been included in the comments below.