Quick test: Futuremark 3DMark v2.3.3663 Vulkan API Overhead Benchmarks

by Hilbert Hagedoorn on: 03/23/2017 04:39 PM | source: | 50 comment(s)

Futuremark just released 3DMark v2.3.3663 with support for Vulkan in their API overhead test, we ran some quick tests to see what is happeing.

This means you can now compare the API performance of Vulkan, DirectX 12, and DirectX 11 with one easy-to-use test. Vulkan is a new graphics API that provides high-efficiency, low-level access to modern GPUs in a wide variety of devices from PCs to smartphones. APIs like Vulkan and DirectX 12 make better use of multi-core CPUs to streamline code execution and eliminate software bottlenecks, particularly for draw calls. Games typically make thousands of draw calls per frame, but each one creates performance-limiting overhead for the CPU. Vulkan and DirectX 12 reduce that overhead, which means more objects, textures and effects can be drawn to the screen. The 3DMark API Overhead feature test measures API performance by making a steadily increasing number of draw calls. The result of the test is the number of draw calls per second achieved by each API before the frame rate drops below 30 FPS. The Vulkan test replaces the Mantle test found in previous versions. Now I did chart up some results, see below:

The purpose of the test is to compare the relative performance of different APIs on a single system. The API Overhead feature test is not a general-purpose GPU benchmark, and it should not be used to compare graphics cards from different vendors. It more for you to see how much faster or slower DX12 is compared to Vulkan. You'll notice that the AMD Radeon card scores are way off. Again, this is not a graphics GPU test, but a test that shows your system API renderer performance in relation towards your setup. I have no clue about the AMD results as they are, I did run them three times and even re-installed drivers and re-seated the cards into another slot. I also have to mention that in the past we stepped away from the API test as the results back then also showed a lot of platform inconsistency.

Though overall Vulkan might be a notch slower compared to D3D12, it certainly is shaping up to become a very fast API. Talking about platforms, I figured why not quickly compare Ryzen 7 1700 versus Core i7 5960X (both 8-cores / 16-threads).

You will notice that Core i7 5960X is clocked at 4.3 GHz. The Ryzen 7 1700 is clocked at 4.1 GHz, it is the highest frequency we can obtain stable. Not bad really, the 1200 USD Intel processor is faster, especially combined with quad-channel and a 200 MHz x8 cores advantage. That 329 USD Ryzen 7 is holding up nicely.

Now I also ran the Ryzen Platform with the Fury X to be certain there isn't a bug on the X99 platform causing this behaviour:

So moving the Radeon Fury X towards a completely different system did not change the performance bracket at all. There might be a bug in the current AMD drivers, Futuremark still has some fixing to do or it just is what it is.

Related Stories

Quick Test: 3DMark TimeSpy DirectX 12 Benchmarks - 07/15/2016 11:29 AM
For those that missed the news. it seems that Time Spy, the new DirectX 12 benchmark test from Futuremark is available as DLC right now on Steam, coming soon to all Windows editions of 3DMark. With i...

Quick test: DirectX 12 API Overhead Benchmark results - 03/26/2015 08:09 PM
As you guys know, DirectX 12 is going to greatly free up processor utilization, and thus your games can make more draw-calls with your processor. More efficient usage of CPUs with multiple cores is tr...

NVIDIA PhysX Quick Test - 06/20/2008 01:47 PM
As you guys likely know NVIDIA issued a new PhysX supporting driver to the press, which (driver) actually already got leaked. This driver will enable and allow a significant set of Physics functions t...

10 pages « < 7 8 9 10

mtrai
Senior Member

Posts: 1183
Joined: 2013-09-21

#5412413 Posted on: 03/25/2017 08:43 PM

Good points, I noticed that in small sections of the benchmarks that my GPU hit 100% load - towards the end of the tests when there are shed loads of draw calls on the screen. It was at those points that my GPU was the bottleneck, so I can see why more powerful GPUs can increase the scores slightly, and it's probably why I have very similar scores to the guy I replied to earlier who has a GTX 1070 - we probably became GPU limited at the same point in the test because both of our CPUs were able to deliver enough draw calls - mine through high CPU clocks and 8 threads & the other guy by having a bazzillion cores on his server CPU (GeniusPr0 and his BWE Xeon 18C/36T 2.745Ghz).

While it answered my questions on the DX11 from my HD 7770 and RX 480 it did not explain very well reviewers much lower AMD scores across the board then real people are getting out in the real world. My question to futuremark was very specific on on the DX 11 results alone...as to reviewers results was to ask the reviewer,

Undying
Senior Member

Posts: 20791
Joined: 2008-08-28

#5412592 Posted on: 03/26/2017 03:41 PM

for me,

DX12 was consistently hitting just over 17,000,000 draw-calls per second. (most 17,220,483)
Vulcan was consistently hitting just under 16,000,000 draw-calls per second. (most 15,985,491)

Is that gtx780 and how far is it overclocked?

This is the best i could do with 280X. I dont think this translates to real world performance. Im positive my card outperformed yours in Doom using Vulkan according to the web benchmarks and my experience.

DX12 - 14 073 975
Vulkan - 13 710 626

-Tj-
Senior Member

Posts: 17811
Joined: 2012-05-18

#5412677 Posted on: 03/26/2017 08:59 PM

I got ~17mill with my old gtx780 @ 1150MHz boost,

http://www.3dmark.com/compare/aot/198728/aot/192915/aot/62970/aot/46129
last result was best off dx11 ST, now latest 378.92 driver went a bit further :nerd:

vase
Senior Member

Posts: 1652
Joined: 2015-03-20

#5412736 Posted on: 03/27/2017 02:33 AM

Is that gtx780 and how far is it overclocked?

This is the best i could do with 280X. I dont think this translates to real world performance. Im positive my card outperformed yours in Doom using Vulkan according to the web benchmarks and my experience.

DX12 - 14 073 975
Vulkan - 13 710 626

He gets higher drawcalls because his CPU has better multicore performance.
The GPU is just a small factor here.
If you replace your 280X with something older or newer you won't see much difference in the API benchmark as long as you use the same CPU.
But a small increase/decrease in CPU power (OC or underclock) and you will see a bigger impact on these numbers.

HeavyHemi
Senior Member

Posts: 6952
Joined: 2008-10-27

#5413121 Posted on: 03/28/2017 01:11 AM

May as well post my results too, interesting that your DX12 results are higher than Vulkan, for me it's the other way around, yet we both have Intel/NVidia.

DX11 mult:i 4,507,018
DX12: 27,231,630
Vulkan: 29,736,051

Anyone know why our results are swapped around & so different?

Yes. I explained why. For my result, Vulkan runs on my 980 Ti instead of the 1080 Ti. The Doom game does the same thing if you use the Vulkan API. Apparently, from my reading if you have two different AMD GPU's the Vulkan API will also select your secondary GPU. So this is a Vulkan API issue, not an AMD or Nvidia dirver issue. If they are the same model, you'd not notice the difference. In my case, since the 980 Ti was used, the score is lower.

He gets higher drawcalls because his CPU has better multicore performance.
The GPU is just a small factor here.
If you replace your 280X with something older or newer you won't see much difference in the API benchmark as long as you use the same CPU.
But a small increase/decrease in CPU power (OC or underclock) and you will see a bigger impact on these numbers.

You might be surprised at how increasing CPU and GPU affects the scores...

http://www.3dmark.com/compare/aot/173675/aot/179282/aot/202413

10 pages « < 7 8 9 10

Post New Comment

Click here to post a comment for this news story on the message forum.