SPECapc for 3ds Max™ 2015 GPU Scores

The SPECapc for 3ds Max™ 2015 is a benchmark developed by The Standard Performance Evaluation Corporation (SPEC). It is the updated version of the synonymous 2011 benchmark, a much needed update as the graphic engine in 3DS saw notable updates between v2013 and the latest v2015 .

SPECapc for 3ds Max 2015 contains 48 tests for comprehensive measurement of modeling, interactive graphics, visual effects, CPU and GPU performance.

Features in the latest SPECapc benchmark are keyed to upgrades in 3ds Max 2015, including new DirectX 11 shaders and vector maps, Nitrous viewport enhancements, and new dynamics and visual effects. The benchmark also improves run-to-run consistency and results reporting.

Benchmark results are derived by taking the total number of seconds to run each test and nomalizing it based on a reference machine, in this case a Dell Precision 690 workstation with 2.0-GHz Intel Xeon 5130 processor, 4 x 4GB FB-DIMM DDR2 SDRAM (ECC) memory, NVIDIA Quadro Q600 graphics card, and Western Digital 500GB 7200 rpm hard drive. The normalization process ensures a scoring system where a bigger score is better. Composite scores are reported for CPU, GPU and large-model (city scene) performance.

Since the CPU used for all of these tests is the same, the scores reported will the the composite GPU score, and separately the “large model” scene score for each GPU. Throughout our tests, CPU performance was verified to be consistent regardless of GPU used, i.e. your rendering performance (also determined by SPECapc for 3ds Max 2015) is not really affected by the GPU used. For the record, the CPU score our i7-4770K @ 4.5GHz  test bed rig achieves 5.43 CPU points and this is consistent within the second decimal for all discreet GPUs. CPU score drops slightly when the HD 4600 IGP is enabled..

The benchmark can be run at 4K and 1080p resolutions, and at different anti-aliasing levels: 0xAA (default), 2x, 4x, and 8x AA.  To keep the performance scores relative, all tests posted in this page are run @ 1080p, as 4K resolutions would be simply unbearable for any but the post powerful of GPUs. 3K & 4K resolutions are already picking up in the CG world, but 1080p resolutions are still native for the vast majority of monitors used.

In the first chart, the results from all the tests are weighted and normalized against the base system, as described above, that would score 1.0 points.

SpecAPC_3DS_2015 GPU Comp

And below, are isolated the GPU scores specifically for the high polygon model scenes included in the benchmark.

20150220_SpecAPC_3DS_2015-02

 

Finally, the bar chart below contains the results for the GPU rendering portion of the test. Note that the benchmark is unable to utilize both cards in case of a SLI pairing.

20150220_SpecAPC_3DS_2015-03


Disclaimer: Unless otherwise stated, all the cards in these tables were tested on PCFoo’s testbed with the listed specifications. Due to limited resources, the range of cards available is small. Should you wish to provide a card for a limited time in order to get it tested and included, feel free to contact us at: info@pcfoo.com. You can make it into our Sponsor list 😉

13 thoughts on “SPECapc for 3ds Max™ 2015 GPU Scores

  1. do you ever do this tests with gtx 960? or do you know about it? if i need to choose between 750ti and gtx 960, then which one should i take?

    • Sorry, I did not have a 960 available for testing. It is safe to assume it score somewhere in the upper middle between the 750Ti and the 970. How critical the difference will be for you, is hard to say, as for simple models the 750Ti does great – but what each one considers “complicated” is pretty subjective.

  2. Superb article. Very helpful indeed. I’m planning on building an intermediate-pro pc for VFX hobbying. I have a wide range of softwares and plugins available. To name the important ones , 3ds max (plugins= mental ray/v ray / fume fx /krakatoa), After effects, real flow, Vue. I am finding it tricky to get my cpu+gpu+mobo combo just right. I did read your articles on budget workstations. However I am not sure if it ticks all the boxes for me?

  3. I’m surprised with GTX750ti, I never included that in my shortlist among GTX 580, GTX660 ti and GTX680- does the test above include viewport performance and cuda render? Most of my models have high polygons. Thanks for this review!

    • The scores are taken out of a composite score that includes compute, although this should not be CUDA based as Radeon/Firepro cards can complete the test with an actual GPU rendering score. Unfortunately I don’t have any 6xx or 7xx series GTX anymore to provide direct comparisons with a 750Ti or 9xx GPU that are all Maxwell based.

  4. Reading all your articles, very interesting stuff, learning a lot! I’m trying to put together a budget 3ds Max CPU rendering workstation for a colleague, and I thought the 750ti was the card for the job… But this “Large Model Score” you present to us here makes me think otherwise. However, I’ve seen you recommending this low budget Maxwell GPU quite a lot on the comments section here. How many polys do you think the 750 can handle fluidly? I’m only concerned about viewport performance with fairly complex scenes (25 milllion polys, perhaps more), no GPU rendering at all of course.

    • I don’t know how reliably I can answer you question, as 25M Poly scenes can be very differently organized. Using many proxies / point cloud or even groups to organize your groups, makes it very easy for the adaptive degradation engine 3DS Max’s viewport to ensure workable FPS when you navigate your model. Having the same 25M polygons dispersed on a scene as raw, unorganized geometry can be painful. Someone working on automotive or ArchViz models for example, will probably find it easier to organize in discreet groups & layers, vs someone who is working on more organic models – something like character creation for games etc. Afaik the “large scene” city model used in SPECapc for 3DS Max 2015, is a carryover from the 2011 version, with a bit over 32M polys.

      • Thanks Dimitri, you’re absolutely right about the importance of organizing and optimizing a scene to achieve decent viewport fluidity, even though this would be one more thing on the artist’s mind to worry about. We’re still torn between the 750ti and the 960 for 3ds Max viewport though, and I can’t seem to find any benchmarks specifically tailored to judge viewport performance and polycount tolerance with these relatively new cards. Thanks again, always a pleasure to ask hardware questions to people who know what they’re talking about, specially considering that in quite a few parts of the world some purchase decisions can’t be taken lightly.

  5. Nul, did You finally make up Your mind and bought a card? What was the choice? Was it 750ti ? If Yes, how it performs in 3ds max vieport on heavy scenes?

  6. Simply outstanding. It is so difficult to find information on consumer level hardware. I am building a system and this was utterly helpful. Thanks a ton and happy holidays

  7. I’m really curious how my R9 290 would score. I’ve been using it for more than a year and I’ve been pretty happy with it but to this day I’m really not sure if I shouldn’t have gone with its main competition from the time – GTX970. FirePro W5000 is actually R7 260, pretty low-end card, and it does quite nicely with the large composite model. I’m really interested how R290 compares to GTX970, thus GCN 1.2 vs Maxwell. If you could add some Radeons to the results, I’d be REALLY grateful!

    • Marcel, from my experience powerful Radeons like the R9 280/290/380 etc do fine with 3DS (Direct3D) and usually better than GTX’s in OpenGL apps. I just didn’t happen to own or get my hands on anything faster than a R9 270, and even that did not stay in my hands long enough to complete all tests with it (have SPECviewperf® 12 scores but not 3DS 2015).

Leave a Reply to Dimitris Cancel reply

Your email address will not be published. Required fields are marked *