General Compute Performance – SPECviewperf12
SPECviewperf 12 is a benchmarking application centered around workstation graphics performance. Both OpenGL and DirectX performance is measured by the various workloads or "viewsets." It is important to note that the AMD Radeon Pro Duo seemed to be running in single GPU mode for these tests. SPECviewperf does not provide any controls to target additional GPUs.
CATIA Viewset (Catia-04)
The catia-04 viewset was created from traces of the graphics workload generated by the CATIA V6 R2012 application from Dassault Systemes. Model sizes range from 5.1 to 21 million vertices.
The viewset includes numerous rendering modes supported by the application, including wireframe, anti-aliasing, shaded, shaded with edges, depth of field, and ambient occlusion
With the CATA viewset of SPECviewperf 12, we see a clear deliniation betwen the different products in the Quadro lineup. The Radeon Pro Duo running in Single GPU mode struggles to compete with the low end Quadro P2000.
Energy Viewset (Energy-01)
The energy-01 viewset is representative of a typical volume rendering application in the seismic and oil and gas fields. Similar to medical imaging such as MRI or CT, geophysical surveys generate image slices through the subsurface that are built into a 3D grid. Volume rendering provides a 2D projection of this 3D volumetric grid for further analysis and interpretation.
At every frame, depending on the viewer position, a series of coplanar slices aligned with the viewing angle are computed on the CPU and then sent to the graphics hardware for texturing and further calculations such as transfer function lookup, lighting and clipping to reveal internal structures. Finally, the slices are blended together before the image is displayed.
In the Energy workload we see a similar story. Performance between the Quadro options scale as you would expect, while the Radeon Pro Duo comes in last place despite its powerful Fiji XT GPU.
Medical Viewset (Medical-01)
The medical-01 viewset is representative of a typical volume rendering application that renders a 2D projection of a 3D volumetric grid. A typical 3D grid in this viewset is a group of 3D slices acquired by a scanner (such as CT or MRI).
At every frame, depending on the viewer position, a series of coplanar slices aligned with the viewing angle are computed on the CPU and then sent to the graphics hardware for texturing and further calculations, such as transfer function lookup, lighting and clipping to reveal internal structures. Finally, the slices are blended together before the image is displayed.
For the rendering application of the Medical viewset, performance again scales nicely with the Quadro GPUs while the AMD card finishes last.
Showcase Viewset (showcase-01)
The showcase-01 viewset was created from traces of Autodesk’s Showcase 2013 application. The model used in the viewset consists of 8 million vertices.
The viewset is the first viewset in SPECviewperf to feature DX rendering. Rendering modes included in the viewset include shading, projected shadows, and self-shadows.
Utilizing standard DirectX for rendering, we see that the Showcase viewset is the first test so far to break the Quadro superiority. The Radeon Pro Duo in single GPU mode manages to best the P2000 and is very competitive with the P4000 in this test.
Siemens NX (snx-02)
The snx-02 viewset was created from traces of the graphics workload generated by the NX 8.0 application from Siemens PLM. Model sizes range from 7.15 to 8.45 million vertices.
In testing with the Siemens NX viewset, the Radeon Pro Duo scores a very low score compared to its Quadro competitors. This likely points to some issue with this workload and AMD GPUs as opposed to the actual relative performance levels of the Radeon Pro Duo.
Maya viewset (maya-04)
The maya-04 viewset was created from traces of the graphics workload generated by the Maya 2013 application from Autodesk. Model size is 727,500 vertices.
The viewset includes numerous rendering modes supported by the application, including shaded mode, ambient occlusion, multi-sample anti aliasing, and transparency.
In the Maya viewset, we see that the Quadro P4000 and P5000 score similar results despite their almost 4 TFLOPs compute performance gap. This test uses Maya 2013, so it may not be as optimized for today's high power GPUs.
3DS Max Viewset (3dsmax-05)
The 3dsmax-05 viewset was created from traces of the graphics workload generated by 3ds Max 2016 using the default Nitrous DX11 driver.
The models for this viewset came from the SPECapc for 3ds Max 2015 benchmark and other sources. In order to best approximate real-world use cases, several tests incorporate multiple viewsets on screen, each using a different rendering method. The styles of rendering in the viewset reflect those most commonly used in major markets, including realistic, shaded and wireframe. Some lesser-used but interesting rendering modes such as facets, graphite and clay are also incorporated. The animations in the viewset are a combination of model spin and camera fly-through, depending on the model.
In 3ds Max using the Nitrous DX11 renderer, the Radeon Pro Duo in single GPU mode only manages to score half of the score we see from the Quadro P5000 despite the similarities in raw GPU compute perforamance of these two products.
Solidworks viewset (sw-03)
The sw-03 viewset was created from traces of Dassault Systemes’ SolidWorks 2013 SP1 application. Models used in the viewset range in size from 2.1 to 21 million vertices.
The viewset includes numerous rendering modes supported by the application, including shaded mode, shaded with edges, ambient occlusion, shaders, and environment maps.
Once again, the Radeon Pro Duo is completely outclasses by even the low end Quadro P2000 in the Solidworks component of SPECviewperf 12.
I wouldn’t worry about these
I wouldn’t worry about these benchmarks, Since the Nvidia Pascal cards just came out,and AMD got the VEGA Pro cards coming out around June or so this year to replace there current WX line up.
Really disingenuous to not
Really disingenuous to not include Radeon pro WX lineup but I guess you might not have had access to one.
We really would have liked to
We really would have liked to include them, but we haven't had our hands on them!
You should mention this to
You should mention this to avoid further bashing.
Great job, Ken.
Yeah well…
Yeah well… http://hothardware.com/reviews/amd-radeon-pro-wx-4100-and-wx-5100-workstation-gpu-reviews?page=3
http://www.tomshardware.com/reviews/amd-radeon-pro-wx-7100,4896-3.html They have been reviewed by others as early as December after a mid year launch. Meanwhile the quadros were announced 2 mo. ago. and are getting reviewed first here.
To avoid bashing get your hands on them and make a review. The quadros will probably still win on many things, no reason to use an old gen card.
WX7100 is ~$600 and performs
WX7100 is ~$600 and performs like it.
http://hothardware.com/reviews/nvidia-quadro-p4000-and-p2000-pro-workstation-gpu-review?page=5
It falls behind the P2000 in many tasks and then is close to the p5000 for openCL.
They simply aren’t competitive this gen in anything other than price. and OpenCL performance.
Hi Ken, at 26:58 in the PC
Hi Ken, at 26:58 in the PC Perspective Podcast #444 you mentioned that the Radeon Pro Duo uses the “Radeon Graphics Drivers,” but AMD also supplies professional drivers for the Radeon Pro: http://support.amd.com/en-us/download/workstation?os=Windows%2010%20-%2064#pro-driver
Some professional applications require certified drivers to work properly, which may be why you got a few strange results in SPECviewperf12
Yes AMD does provide the
Yes AMD does provide the professional drivers so software developers can save money by getting the “Radeon Pro Duo” and using it to develop for production systems that use the more expensive Radeon Pro WX(Formally called FirePro) SKUs. But when benchmarking professional cards the review is not scientific and proper if the real professional GPUs are not tested against other real professional GPU SKUs.
Any real certification testing for prodction software will be done on the Radeon Pro WX professional SKUs(production certified) even if the software was developed using the Radeon Pro Duo(Just as a Low Cost development platform). So and final testing/certification and reviewing(Proper Reviewing) needs to be done on the Radeon Pro WX SKUs.
So if you can’t test the latest Quadro SKUs against the latest Radeon Pro WX competition, why would a reviewer want to tarnish their reputation!
Interesting that AMD doesn’t
Interesting that AMD doesn't seem to surface the Pro driver link unless you dig for it. If you go to the AMD default driver site, the Radeon Pro Duo option brings you to download Radeon Software Crimson Edition ReLive
That’s not going to help
That’s not going to help because the Radeon Pro WX SKUs may have more memory and other tweaks. If Nvidia has Quadros to offer for testing then Nvidia must have some Radeon Pro SKU samples also for research that it can offer to make for a more scientific/balanced testing run. There has to be more comparisons Apples to Apples or the testing should be done completely without any Radeon SKUs, unless those Radeon Pro WX samples can be had from either AMD or Nvidia.
Yes, this kind of testing
Yes, this kind of testing with professional GPUs against non professional GPU it’s not going to go over well. Where is any Radeon Pro WX SKUs. The Radeon Pro duo is not more like the Titan X is for Nvidia with Quadro being Nvidia’s professional line. So where are the Radeon Pro WX(Formally called Fire Pro) SKUs.
Very Disingenuous this “review” is.
I haven’t messed with 3dsMax
I haven’t messed with 3dsMax since ~2004. I can only imagine how much nicer it runs one something faster than a GeForce4 ti 4200.
Such a damn good, awesome,
Such a damn good, awesome, useful AF review. Finally, no peasanty AMD crap. Please, more stuff like this.
I know really… Why even
I know really… Why even bother with this pro duo toy… At least those blue cards weren’t there to sully the graphs.
Damn, the pro duo is a beast
Damn, the pro duo is a beast in OpenCL and it’s only $800. If I didn’t have to upgrade my PSU for it, I’d probably buy it.
Anyone use those cards for
Anyone use those cards for professional real-time VR ? (stereo 3d)
It seem this would be easy now that headset are so easily available.
(Specially in term of CAD / modeling / Architecture)
On another note, VR gaming I wonder how the duo perform VS a GTX 1080 ti. It seem trivial for driver to render the same frame in parallel. Its near 100% scaling.
But I heard very little on sli/crossfire and VR…
I’m curious why you didn’t
I’m curious why you didn’t include results for the 1060/1070/1080/1080ti.
You reference them on the first page, and I’m sure many readers would love to know what sort of performance the much cheaper retail cards 🙂
Nvidia only issues the
Nvidia only issues the certified professional drivers for their professional line of cards. Wow it could work on the consumer cards, nvidia does not allow it.
Quote:”And for these
Quote:”And for these consumers, the price increase is worth it”.
I think that you are thinking about years ago when you put this line in. Upwards from the performance of the 1060 maybe but from the performance of the 1070 upwards to the titan series, if a $800 card could compete with the more expensive consumer cards, a lot of us would consider the price. Yeah, there are some seriously more expensive pro-cards, but if this set could better consumer cards at these prices, it would be a win.
I remember some time ago when it was possible to flash a certain consumer card and make it a pro-card, but performance was not quite the same. So the best (todays) consumer card will be better than a pro card even if it would run everyday software, games especially.
Seems like a broken tenet where a pro isnt as good as the peoples ….unless you are are of a few that has the few games that run entirely on opencl