Tag Archives: GPU

Autodesk Arnold 5.3 with Arnold GPU in public beta

Autodesk has made its Arnold 5.3 with Arnold GPU available as a public beta. The release provides artists with GPU rendering for a set number of features, and the flexibility to choose between rendering on the CPU or GPU without changing renderers.

From look development to lighting, support for GPU acceleration brings greater interactivity and speed to artist workflows, helping reduce iteration and review cycles. Arnold 5.3 also adds new functionality to help maximize performance and give artists more control over their rendering processes, including updates to adaptive sampling, a new version of the Randomwalk SSS mode and improved Operator UX.

Arnold GPU rendering makes it easier for artists and small studios to iterate quickly in a fast working environment and scale rendering capacity to accommodate project demands. From within the standard Arnold interface, users can switch between rendering on the CPU and GPU with a single click. Arnold GPU currently supports features such as arbitrary shading networks, SSS, hair, atmospherics, instancing, and procedurals. Arnold GPU is based on the Nvidia OptiX framework and is optimized to leverage Nvidia RTX technology.

New feature summary:
— Major improvements to quality and performance for adaptive sampling, helping to reduce render times without jeopardizing final image quality
— Improved version of Randomwalk SSS mode for more realistic shading
— Enhanced usability for Standard Surface, giving users more control
— Improvements to the Operator framework
— Better sampling of Skydome lights, reducing direct illumination noise
— Updates to support for MaterialX, allowing users to save a shading network as a MaterialX look

Arnold 5.3 with Arnold GPU in public beta will be available March 20 as a standalone subscription or with a collection of end-to-end creative tools within the Autodesk Media & Entertainment Collection. You can also try Arnold GPU with a free 30-day trial of Arnold. Arnold GPU is available in all supported plug-ins for Autodesk Maya, Autodesk 3ds Max, Houdini, Cinema 4D and Katana.

Review: AMD’s Radeon Pro WX8200

By Mike McCarthy

AMD has released the WX8200 high-end professional GPU as part of their Radeon Pro line. It’s based on the Vega architecture, with 3,584 compute cores accessing 8GB of HBM2 memory at up to 512GB/sec. Its hardware specs are roughly equivalent to their $400 Vega 56 gaming card but with professional drivers tuned for optimized performance in a variety of high-end 3D applications. AMD is marketing the WX8200, which is priced at $999, as an option between Nvidia’s Quadro P4000 and P5000.

Some background: I really haven’t used an AMD GPU before, at least not since they bought ATI over 10 years ago. My first Vaio laptop (2002) had an ATI Radeon 7500 in it, and we used ATI Radeon cards in our Matrox AXIO LE systems at Bandito Brothers in 2006. That was right around the time ATI got acquired by AMD. My last AMD-based CPUs were Opterons inside HP XW9300 workstations around the same time period, but we were already headed towards Nvidia GPUs when Adobe released Premiere Pro CS5 in early 2010.

CS5’s CUDA-based, GPU-accelerated Mercury Playback Engine locked us in to Nvidia GPUs for years to come. Adobe eventually included support for OpenCL as an alternative acceleration for the Mercury Playback Engine, primarily due to the Mac hardware options available in 2014, but it was never as mature or reliable on Windows as the CUDA-based option. By that point we were already used to using it, so we continued on that trajectory.

I have a good relationship with Nvidia, and have reviewed many of their cards over the years. Starting back in 2008, their Quadro CX card was the first piece of hardware I was ever provided with for the explicit purpose of reviewing, instead of just writing about the products I was already using at work.

When I was approached about doing this AMD review, I had to pause for a moment. I wanted to make sure I could really do an honest and unbiased review of an AMD card. I asked myself, “What if they worked just as well as the Nvidia cards I was used to?” That would really open up my options when selecting a new laptop, as most of the lighter weight options have had AMD GPUs for the last few years. Plus, it would be useful information and experience to have since I was about to outfit a new edit facility and more options are always good when finding ways to cut costs without sacrificing performance or stability.

So I agreed to review this new card and run it through the same tests I use for my Quadro reviews. Ideally, I would have a standard set of assets and timelines that I could use every time I needed to evaluate the performance of new hardware. Then I could just compare it to my existing records from previous tests. But the tests run in software that is changing as well, and Premiere Pro was on Version 11 when I tested the Pascal Quadros; it’s now on Version 13. Plus, I was testing 6K files then and have lots of 8K assets now, as well as a Dell UP3218K monitor to view them on. Just driving images to an 8K monitor smoothly is a decent test for a graphics card, so I ended up benchmarking not just the new AMD card, but all of the other (Nvidia) cards I have on hand for comparison, leading to quite a project.

The Hardware
The first step was to install the card in my Dell Precision 7910 workstation. Slot-wise, it just dropped into the location usually occupied by my Quadro P6000. It takes up two slots, with a single PCIe 3.0 x16 connector. It also requires both a six-pin and eight-pin PCIe power connector, which I was able to provide, with a bit of reconfiguration. Externally, it has four MiniDisplayPort connectors and nothing else. Dell has an ingenious system of shipping DP to mDP cables with their monitors that have both ports, allowing either source port to be used by reversing the cable. But that didn’t apply to my dual full-sized DisplayPort UP3218K monitor. I didn’t realize this until after ordering mDP-to-DP cables, which I already had from my PNY PrevailPro review for the same reason.

I prefer the full-sized connectors to ensure I don’t try to plug them in backwards, especially since AMD didn’t use the space savings to include any other ports on the card. (HDMI, USB-C, etc.) I also tried the card in an HP Z4 workstation a few days later to see if the Windows 10 drivers were any different. Those notes are included throughout.

The Drivers
Once I had my monitors hooked up, I booted the system to see what would happen. I was able to install the drivers and reboot for full functionality without issue. The driver install is a two-part process. You first you install AMD’s display software, and then that software allows you to install a driver. I like this approach because it allows you to change your driver version without reinstalling all of the other supporting software. The fact that driver packages these days are over 500MB is a bit ridiculous, especially for those of us not fortunate enough to live in areas where Fiber Internet connections are available. Hopefully this approach can alleviate that issue a bit.

AMD advertises that this functionality also allows you to switch driver versions without rebooting, and their RadeonPro line fully supports their gaming drivers as well. This can be an advantage for a game developer who uses the professional feature set for their work but then wants to test the consumer experience without having a separate, dedicated system. Or maybe it’s just for people who want better gaming performance on their dual-use systems.

The other feature I liked in their software package is screen-capture functionality called RadeonPro ReLive. It records all onscreen images or specific window selections, as well as application audio and optionally microphone audio. It saves the screen recordings to AVC or HEVC files generated by the VCE 4.0 hardware video compression engine on the card. When I tested, it worked as expected, and the captured files looked good, including a separate audio file for my microphone voice while the system audio was embedded in the video recording.

This is a great tool for making software tutorials, or similar type tasks, and I intend to use it in the near future for posting videos of my project workflows. Nvidia offers similar functionality in the form of ShadowPlay, but doesn’t market it to professionals since it’s part of the GeForce Experience software. I tested for comparison, and it does work on Quadro cards but has fewer options and controls. Nvidia should take the cue from AMD and develop a more professional solution for their users who need this functionality.

I used the card with both my main curved 34-inch monitor at 3440×1440, and my 8K monitor at 7680×4320. The main display worked perfectly the entire time, but I had issues with the 8K one. I went through lots of tests on both operating systems, with various cables and drivers before discovering that a firmware update for the monitor solved the issues. So if you have a UP3218K, take the time to update the firmware for maximum GPU compatibility. My HDMI-based home theater system on the other hand worked perfectly and even allowed me access to the 5.1 speaker in Premiere through the AMD HDMI audio drivers.

10-Bit Display Support
One of the main reasons to get a “professional” GPU over a much cheaper gaming card is that they support 10-bit color in professional applications instead of just in full-screen outputs like games and video playback that are supported at 10-bit on consumer GPUs. But when I enabled 10-bit mode in the RadeonPro Advanced panel, I ran into some serious issues. On Windows 7, it disabled the view ports on most of my professional apps, like Adobe’s Premiere, After Effects and Character Animator. When I enabled it on a Windows 10 system to see if it worked any better in a newer OS, the Adobe application interfaces looked even crazier and still no video playback.

10-bit in Premiere

I was curious to see if my Print Screen captures would still look this way once I disabled the 10-bit setting because, in theory, even after seeing them that way when I pasted them into a Photoshop doc, that could still be a display distortion of proper-looking screen capture. But no, the screen captures look exactly how the interface looked on my display.

AMD is aware of the problem and they are working on it. It is currently listed as a known issue in their newest driver release.

Render Performance
The time then came to analyze the card’s performance and see what it could do. While GPUs are actually designed to accelerate the calculations required to display 3D graphics, that processing capacity can be used in other ways. I don’t do much true 3D processing besides the occasional FPS game, so my GPU use is all for image processing in video editing and visual effects. This can be accelerated by AMD’s GPUs through the OpenCL (Compute Language) framework (as well as through Metal on the Mac side).

My main application is Adobe Premiere Pro 12, and it explicitly supports OpenCL acceleration, as does Adobe After Effects and Media Encoder. So I opened them up and started working. I didn’t see a huge difference in interface performance, even when pushing high-quality files around, but that is a fairly subjective test and fairly intermittent. I might drop frames during playback one time, but not the next time. Render time is a much more easily quantifiable measure of computational performance, so I created a set of sequences to render in the different hardware configuration for repeatable tests.

I am pretty familiar with which operations are CPU-based and which run on the GPU, so I made a point of creating test projects that work the GPUs as much as possible. This is based on the clip resolution, codec and selection of accelerated effects to highlight the performance differences in that area. I rendered those sequences with OpenCL acceleration enabled on the WX8200 and with all GPU acceleration disabled in the Mercury software playback mode, then with a number of different Nvidia GPUs for comparison.

Trying to push the cards as hard as possible, I used 6K Venice files and 8K Red files with Lumetri grades and other GPU effects applied. I then exported them to H.265 files at 10-bit Rec.2020 in UHD and 8K. (I literally named the 8K sequence “Torture Test.”)

My initial tests favored Nvidia GPUs by a factor of at least three to one, which was startling, and I ran them repeatedly to verify with the same result. Further tests and research revealed that usually AMD (and OpenCL) is about 25% slower than Adobe’s CUDA mode on similarly priced hardware, verified by a variety of other sources. But my results were made worse for two reasons: Red decoding is currently more optimized for acceleration on Nvidia cards and rendering at 10-bit ground the AMD-accelerated OpenCL renders to a halt.[

When exporting 10-bit HDR files at “Maximum Bit Depth,” it took up to eight times as long to finish rendering. Clearly this was a bug, but it took a lot of experimentation to narrow it down. And the Intel-based OpenCL acceleration doesn’t suffer from the same issue. Once I was able to test the newest Media Encoder 13 release on the Windows 10 system, the 10-bit performance hit while using the AMD disappeared. When I removed the Red source footage and exported 8-bit HEVC files, the WX8200 was just as fast as any of my Nvidia cards (P4000 and P6000). When I was sourcing from Red footage, the AMD took twice as long but GPU-based effects seemed to have no effect on render time, so those are accelerated properly by the card.

So, basically, as long as you aren’t using Red source and you use Premiere Pro and Media Encoder 13 or newer, this card is comparable to the alternatives for most AME renders.

[Statement from Adobe: “Premiere Pro has a long history of supporting multiple GPU technologies. Adobe works closely with all hardware partners to ensure maximum performance across the wide range of systems that may exist in creative workflows. For example, Adobe has been at the forefront of GPU development across CUDA, OpenCL, Metal and now Vulkan. In the case of OpenCL we partnered with AMD and provided deep access to our code base to ensure that the maximum performance levels were being achieved. It has been, and will remain to be our policy to deeply collaborate with all vendors who create high performance hardware/ software layers for video and audio creatives.”]

Is the RadeonPro WX8200 Right for Your Workflow?
That depends on what type of work you do. Basically, I am a Windows-based Adobe editor, and Adobe has spent a lot of time optimizing their CUDA-accelerated Mercury Playback Engine for Premiere Pro in Windows. That is reflected in how well the Nvidia cards perform for my renders, especially with Version 12, which is the final release on Windows 7. Avid or Resolve may have different results, and even Premiere Pro on OS X may perform much better with AMD GPUs due to the Metal framework optimizations in that version of the program. It is not because the card is necessarily “slower,” it just isn’t being used as well by my software.

Nvidia has invested a lot of effort into making CUDA a framework that applies to tasks beyond 3D calculations. AMD has focused their efforts directly on 3D rendering with things like ProRender, and GPU accelerates true 3D renders. If you are doing traditional 3D work, either animation or CAD projects, this card will probably be much more suitable for you than it is for me.

Mike McCarthy is an online editor/workflow consultant with 10 years of experience on feature films and commercials. He has been involved in pioneering new solutions for tapeless workflows, DSLR filmmaking and multi-screen and surround video experiences. Check out his site.

AMD offering FireRender plug-in for 3ds Max

AMD, makers of the line of FirePro graphics cards and engines, has released a free software–based rendering plug-in, the FireRender for Autodesk 3ds Max, which is designed for content creators with 4K workflows and who are looking for photorealistic rendering. FireRender for Max offers physically accurate raytracing and comes with an extensive material library.

AMD FireRender is built on OpenCL 1.2, which means it can run on any hardware. It also provides a CPU backend, which means that FireRender can run on GPU, CPU, CPU+GPU, or a variety of combinations of multiple CPUs and GPUs. Within the FireRender, integrated materials are editable in the 3ds Max Material Slate Editor as nodes. There is also Active Shade Viewport Integration, which means you can work with FireRender in realtime and see your changes as you make them. Physically Correct materials and lighting help with true design decisions via global illumination — including caustics. Emissive and Photometric Lighting, as well as lights from HDRI environments, enable artists to blend a scene in with its surroundings.

AMD says to keep an eye out for other upcoming free software plug-ins for other animation software, including Autodesk Maya and Rhino.


In other AMD news, at the NAB show last month, the company introduced the AMD FirePro W9100 32GB workstation graphics card designed for large asset workflows with creative applications. It will be available in Q2 of this year. The FirePro W9100 16GB is currently available.

Raytracing today and in the future

By Jon Peddie

More papers, patents and PhDs have been written and awarded on ray tracing than any other computer graphic technique.

Ray tracing is a subset of the rendering market. The rendering market is a subset of software for larger markets, including media and entertainment (M&E), architecture, engineering and construction (AEC), computer-aided design (CAD), scientific, entertainment content creation and simulation-visualization. Not all users who have rendering capabilities in their products use it. At the same time there are products that have been developed solely as rendering tools and there are products that include 3D modeling, animation and rendering capabilities, and they may be used primarily for rendering, primarily for modeling or primarily for animation.

Because ray tracing is so important, and at the same time computationally burdensome, individuals and organizations have spent years and millions of dollars trying to speed things up. A typical ray traced scene on an old-fashioned HD screen can tax a CPU so heavily the image can only be upgraded maybe every second or two — certainly not the 33ms needed for realtime rendering.

GPUs can’t help much because one of the characteristics of ray tracing is it has no memory and every frame is a new frame, so the computational load is immutable. Also, the branching that occurs in raytracing defeats the power of a GPU’s SIMD architecture.

Material Libraries Critical
Prior to 2015, all ray tracer engines came with their own materials libraries. Cataloging the characteristics of all the types of materials in the world is beyond the resources of any company’s ability to develop and support. And the lack of standards has held back any cooperative development in the industry. However, a few companies have agreed to work together and share their libraries.

I believe we will see an opening up of libraries and the ability of various ray tracing engines to be able to avail themselves of a much larger library of materials. Nvidia is developing a standard-like capability they are calling the Material Definition Language — (MDL) and using it to allow various libraries to work with a wide range of ray tracing engines.

Rendering Becomes a Function of Price
In the near future, I expect to see 3D rendering become a capability offered as an online service. While it’s not altogether clear how this will affect the market, I think it will boost the use of ray tracing and lower the cost to an as-needed basis. It also offers the promise of being able to apply huge quantities of processing power limited only by the amount of money the user is willing to pay. Ray tracing will resolve to time (to render a scene) divided by cost.

That will continue to bring down the time to generate a ray traced frame for an animation for example, but it probably won’t get us to realtime ray tracing at 4K or beyond.

Shortcuts and Semiconductors
Work continues on finding clever ways to short circuit the computational load by using intelligent algorithms to look at the scene and deterministically allocate what objects will be seen, and which surfaces need to be considered.

Hybrid techniques are being improved and evolved where only certain portions of a scene are ray traced. Objects in the distance for example don’t need to be ray traced and flat, dull colored objects don’t need it.

Chaos Group says the use of variance-based adaptive sampling on this model of Christmas cookies from Autodesk 3ds Max provided a better final image in record time. (Source: Chaos Group)

Semiconductors are being developed to specifically accelerate ray tracing. Imagination Technologies, the company that designs Apple’s iPhone and iPad GPU, has a specific ray tracing engine that, when combined with the advance techniques just described can render an HD scene with partial ray traced elements several times a second. Siliconarts is a startup in Korea that has developed a ray tracing accelerator and I have seen demonstrations of it generating images at 30fps. And Nvidia is working ways to make a standard GPU more ray-tracing friendly.

All these ideas and developments will come together in the very near future and we will begin to realize realtime ray tracing.

Market Size
It is impossible to know how many users there are of ray tracing programs because the major 3D modeling and CAD programs, both commercial and free (e.g., Autodesk, Blender, etc.) have built-in ray tracing engines, as well as the ability to use pluggable add-on software programs for ray tracing.

The potentially available market vs. the totally available market (TAM).

Also, not all users make use of ray tracing on a regular basis— some use it every day, others maybe occasionally or once a project. Furthermore, some users will use multiple ray tracing programs in a project, depending upon their materials library, user interface, specific functional requirements or pipeline functionality.

Free vs. Commercial
A great deal of the raytracing software available on the market is the result of university projects. Some of the developers of such programs have formed companies, others have chosen to stay in academia or work as independent programmers.

The number of new suppliers has not slowed down indicating a continued demand for ray tracing

The non-commercial developers continue to offer their ray tracing rendering software as an open source and for free — and continue to support it, either individually or as part of a group.

Raytracing Engine Suppliers
The market for ray tracing is entering into a new phase. This is partially due to improved and readily available low-cost processors (thank you, Moore’s law), but more importantly it is because of the demand and need for accurate virtual prototyping and improved workflows.

Rendering in the cloud using GPUs (Source OneRender).

As with any market, there is a 20/80 rule, where 20 percent of the suppliers represent 80 percent of the market. The ray tracing market may be even more unbalanced. There would appear to be too many suppliers in the market despite failures and merger and acquisition activities. At the same time many competing suppliers have been able to successfully coexist by offering features customized for their most important customers.

Ray tracing is to manufacturing what a storyboard is to film — the ability to visualize the product before it’s built. Movies couldn’t be made today with the quality they have without ray tracing. Think of how good the characters in Cars looked — that imagery made it possible for you to suspend disbelief and get into the story. It used to be: “Ray tracing — Who needs it?” Today it’s: “Ray tracing? Who doesn’t use it?”

Our Main Image: An example of different materials being applied to the same object (Source Nvidia)

Dr. Jon Peddie is president of Jon Peddie Research, which just completed an in-depth market study on the ray tracing market. He is the former president of Siggraph Pioneers and  serves on advisory boards of several companies. In 2015, he was given the Life Time Achievement award from the CAAD society. His most recent book is “The History of Visual Magic in Computers.”

Nvidia’s GPU Technology Conference: Part III

Entrepreneurs, self-driving cars and more

By Fred Ruckel

Welcome to the final installment of my Nvidia GPU Technology Conference experience. If you have read Part I and Part II, I’m confident you will enjoy this wrap-up — from a one-on-one meeting with one of Nvidia’s top dogs to a “shark tank” full of entrepreneurs to my take on the status of self-driving cars. Thanks for following along and feel free to email if you have any questions about my story.

Going One on One
I had the pleasure sitting down with Nvidia marketing manager Greg Estes, along with Gail Laguna, their PR expert in media and entertainment. They allowed me to pick their brains about Continue reading

Nvidia’s GPU Technology Conference: Part II

By Fred Ruckel

A couple of weeks ago I had the pleasure of attending Nvidia’s GPU Technology Conference in San Jose. I spent five days sitting in on conferences, demos, and in a handful of one-on-one meetings. If the Part I of my story had you interested in the new world of GPU technology, take a dive into this installment and learn what other cool things Nvidia has created to enhance your workflow.

Advanced Rendering Solutions
We consider rendering to be the final output of an animation. While that’s true, there’s a lot
more to rendering than just the final animated result. We could jump straight to the previz Continue reading

Nvidia GPU Technology Conference 2015: Part I

By Fred Ruckel

Recently, I had the pleasure of attending the Nvidia GPU Technology Conference 2015 in San Jose, California, a.k.a. Silicon Valley. This was not a conference for the faint of heart; it was an in-depth look at where the development of GPU technology is heading and what strides it had made over the last year. In short, it was the biggest geek fest I have ever known, and I mean that as a compliment. The cast of The Big Bang Theory would have fit right in.

While some look at “geek” as having a negative connotation, in the world of technology geeks Continue reading

Quick Chat: AMD’s Raja Koduri

By Randi Altman

In a new semi-regular column here at postPerspective, we are asking tool makers how they ensure they are providing gear targeting your needs.

Today’s deadlines call for fast turnarounds due to tighter deadlines, making fast graphics cards even more important tools of the trade. In this installment, we chatted up AMD’s corporate VP of visual computing, Raja Koduri, as he was gearing up for NAB 2014.

It seems like the industry is really embracing OpenCL. Can you talk about the benefits?
Continue reading

Ted Schilowitz tackles supercomputing that targets M&E

By Randi Altman

As history shows, when Ted Schilowitz finds a product he believes in, he’s all in.

Shortly after his retirement last year from Red Digital Cinema, where he had been the public face of the company since its inception, Schilowitz began talks with Silverdraft, a company known for its on-set services truck housing a supercomputer for post production needs, including rendering.

He was impressed with the technology and what it offered, but knew the strategy of the company needed to focus more on what he calls “the flattening out of the entertainment Continue reading