Tag Archives: Nvidia

Review: Custom-built workstations from Mediaworkstations.net

By Brady Betzel

While workstations that optimize the tools you use in the media and entertainment industry are easily accessible, what if you wanted to drill down even deeper?

What if you wanted a workstation built specifically for Adobe Premiere Pro editing of 4K RAW Red R3D files? Or if you are working in Blackmagic’s Resolve, and want playback of 4K RAW Red R3D files? What if you also wanted to dabble in Adobe After Effects?

While the big workstation manufacturers allow customization, that can only go so far online. If you want more personal support and micro-customizations, you will need to find a smaller company that builds very niche computer systems. Mediaworkstations.net is one of those companies.

MediaWorkstations.net’s offerings are custom built computers focused on the media and entertainment professional working in high-end applications like Adobe Premiere Pro, Adobe After Effects, Maxon Cinema 4D, Foundry Nuke and many more including realtime renderers like Octane. When you call MediaWorkstations.net you will talk to a real person, like founder Christopher Johnson, who has extensive knowledge about the media and entertainment industry as it relates to hardware configurations.

Customization
To get started, I had a phone call with Christopher to go over my needs in a system, how much of a budget I had to work with and what I thought I would want to be doing in the future. We talked about how I am currently an online editor who does some color correction and grading. For editing, I primarily use Avid Media Composer but I am using Adobe Premiere and Blackmagic’s Resolve more often these days. I also like to jump into After Effects and Cinema 4D to do some basic stuff but without a slowdown. And, finally, I wanted to stay under $10,000 in price.

Following along on the website, Christopher directed me to the i-X series workstation they configure. He ran through some of my CPU options and explained why I would need one processor versus another processor. He suggested putting 128GB of DDR4 2800 SDRAM, which I went with but I considered changing that to 64GB — I would save a little less than $1,000, and it’s something I could always install more down the road.

Christopher had me throw in a 512GB Samsung 960 Pro SSD for the OS drive, a RAID-0 asset drive and the cherry on top was the 800GB Kingston DCP1000 PCIe for another asset drive. The Kingston DCP 1000 is a beast of a drive that I was super excited to test. You can check the specs out here, but essentially Kingston says it can read up to 6,800MB/s and write up to 6,000MB/s (that is megabytes not bits!). Without giving away too much, this drive is the fastest drive I have ever tested. Unfortunately, at the moment you can’t include it in the online configuration to get a price, but it seems like it retails for anywhere from $1,100 to $1,900.

For GPU power Christopher suggested three Nvidia GeForce 1080 Ti 11GB cards, which seem to retail for $979.99 each, according to Amazon and NewEgg. On MediaWorkstations.net that upgrade will run you an extra $4,149 over the standard Nvidia GeForce GTX 1050 Ti 4GB. Quite the difference in price, but you are paying for full configuration and support.

Beyond the configuration, Christopher took time to explain why I could benefit from three 1080 Ti’s as opposed to only two — apps like After Effects would take more advantage of three or more GPUs where typically Premiere and Resolve don’t see an exponential increase in power. In the end, the configured system totalled $12,095, which is a full 20% higher than the $10,000 budget I had mentioned. To knock that down I would probably cut the memory from 128GB to 64GB, get rid of a GTX 1080 Ti and bump up the CPU to the Intel i9 7920X, which adds a few cores and cache. This gets me to a total of about $8,828 before adding the cost of the Kingston DCP1000, which I assume would get me around $10,000. However, for this review I was sent the original configuration.

After talking with Christopher for over 30 minutes, I got the feeling that he knew what each part of a computer does and how I could use each component to its full potential. We focused on trying to build as much of a future-proof system as we could for around $10,000. Christopher mentioned that these systems will play RAW 4K Red R3D files no problem, and possibly 6K and 8K. That immediately caught my attention, especially when conforming and coloring the R3D files with all of their added benefits.

You should try to build a system for yourself on the Mediaworkstations.net and check out their other offerings like their Enterprise offering called the i-XL which allows for components like dual Xeons or increased memory like 1TB of ECC RAM. They also offer an i-X2 model, which is more like the i-X but with added Xeon processors as well as the a-X which offers AMD Threadripper processors. You can even call them and dial in exactly what components you will need for your specific needs.

Testing the System
So the i-X from MediaWorkstations.net arrived and boy is it loaded with high-end components. Right off the bat, I opened the side panel and started fiddling with the internal components. One of the more impressive parts of the build is the Fractal Design Define XL R2 case. It is easy to open and even has a layer of dense audio dampening material on the inside, which seemed to significantly reduce noise on the outside. You can check out specs on the case at their website.

Also, the power supply is a beast — I immediately noticed the power cable supplied with the system. The power cable is so thick I thought they sent me the wrong one. It definitely makes you feel like you are plugging in a high-end system. You can check out the EVGA SuperNova 1600 T2 power supply here.

Here is a list of the rest of the components that make up the MediaWorkstation.net’s i-X system:

1. Fractal Design’s Define XL R2 Black Silent EATX Full Tower
2. Intel Core i9-7900X Skylake X 13.75M Cache 10-Core CPU
3. Cooler Master Hyper 212 Evo CPU Cooler
4. MSI X299 XPower Gaming
5. Corsair Dominator Platinum 128GB (8x16GB) DDR4 2800
6. 512GB Samsung 960 Pro Internal SSD (OS drive)
7. Two 512GB Samsung 960 Pro Internal SSD (Asset drives – RAID 0)
8. 800GB Kingston DCP1000 PCIe asset drive
9. 3 – ASUS GeForce GTX 1080 ti 11GB Turbo Edition GPU
10. EVGA SuperNova 1600 T2, 80+ titanium 1600W power supply
11. Windows 10 Professional, 64-bit
12. Fractal Design FD-FAN-SSR2-140 2nd front internal fan
13. Two Prolimatech 140mm slim fans

That is a hefty build for any system.

MediaWorkstations.net sells this build at a retail price of $12,095. Besides opening the case right away and messing around with the internal components, I also needed to find out what the premium MediaWorkstations.net is charging on their systems.

I consider myself a pretty advanced user and can build my own computer systems, so what if I wanted to build this myself? Would it be worth it or should I just pay someone else to do it? A great place to build a custom PC from various websites and to find out the cheapest prices is www.pcpartpicker.com. If you want to follow along at home you can find the build I created with some prices here.

There are some caveats when using PC Part Picker: The prices can change, so it may not be 100% accurate, although it is typically pretty close and includes rebates. Also, you must add shipping and tax yourself. And, finally, I had to manually add some parts that I couldn’t find through PCPartPicker. I got my build to $9,387.68 without tax or shipping, this also included $60 in rebates. I would say that we could safely add about $250 in taxes and shipping.

So, if I assume the cost to be about $9,600 MediaWorkstations.net is adding about $2,500. In my opinion that isn’t a bad markup considering someone else is putting in the hours to build and test the system with the applications you use like Resolve, Premiere, Media Composer and many others. In addition to the standard one-, two- or three-year warranties, they offer a 24/7/365, as well next business day warranties available. You may have to add another $500 to $600 for a 24-hour-a-day-warranty, but it’s worth it.

Alright enough tech specs and pricing nerdiness and on to the testing. To be clear, I had a short amount of time to test the performance of the i-X, so I only dove into the basics. This system is very fast. However, I wasn’t able to playback Red RAW R3D files higher than 4K in realtime at full debayer quality in either Premiere or Resolve 14.3.

When I was first asked to review the i-X, I was told it should be able to playback Red RAW R3D files up to 8K in part because of the new Kingston DCP1000 SSD. While this drive was extremely fast, I wasn’t able to playback anything in realtime above 4K resolution. There are a few factors in this that could affect performance, such as whether they were 9:1 or 1:1, but I was told it would work and it didn’t.

On the bright side I was able to color and playback RAW Red R3D files in true 4K resolution. It was pretty amazing that I could add a few nodes and do live grading on 4K resolution Red R3D files without a dropped frame. I also tried exporting the same 18-second 4K Red R3D file in a few different scenarios.

In the first scenario I placed the RAW R3D file on the blazingly fast Kingston SSD and exported it back to itself. Initially, I exported the file with no effects on it other than a simple one-node color correction. I exported it as a 4K DPX sequence, and it took 30 seconds. When I added Temporal Noise Reduction it took 39 seconds. On top of that I then added a serial node with Gaussian Blur that took 40 seconds to export.

I quickly thought that these speeds were a little slow considering the power I had under the hood of this beast. I then exported the same file to the RAID-0 made up of the two Samsung 960 Pro 512GB SSDs, which confirmed my suspicion. With just a simple color correction, the 18-second  Red RAW R3D file took just 10.5 seconds, around 45-48fps to export. With Temporal Noise Reduction it took 38.5 seconds, but with Temporal Noise Reduction and a Gaussian Blur it took 39 seconds. In my testing, I turned off all caching and performance mode improvements.

While I didn’t have the system long enough to test as I would have liked, I was able to get a good taste at how fast the new Intel i9 processors run and how multiple 1080 ti GPUs can help with rendering resizes, noise reduction effects, or even blurring. In the first part of my review, I mentioned that I would likely have swapped out half the RAM and one of the GeForce GTX 1080 Ti’s for a higher-end processor, which would have kept the price down. I didn’t see any improvement in performance because of the the third GPU, but I also didn’t do any testing in After Effects or Cinema 4D, which may have harnessed that extra GPU energy.

Summing Up
Check out the Mediaworkstation site for yourself and maybe even compare those prices with a duplicate build on PCPartsPicker. If you are within a $1,000 or so, then going through MediaWorkstations.net is a great deal. If nothing else, having one single warranty through one company is worth hundreds of dollars in time, money and shipping costs instead of having to manage multiple warranties from dozens of companies.

For peace of mind, I would heavily consider the next–business-day or 24/7/365 warranty instead of the standard one-day warranty simply because waiting for your system to be fixed could leave you without a machine for days or even weeks.

GTC embraces machine learning and AI

By Mike McCarthy

I had the opportunity to attend GTC 2018, Nvidia‘s 9th annual technology conference in San Jose this week. GTC stands for GPU Technology Conference, and GPU stands for graphics processing unit, but graphics makes up a relatively small portion of the show at this point. The majority of the sessions and exhibitors are focused on machine learning and artificial intelligence.

And the majority of the graphics developments are centered around analyzing imagery, not generating it. Whether that is classifying photos on Pinterest or giving autonomous vehicles machine vision, it is based on the capability of computers to understand the content of an image. Now DriveSim, Nvidia’s new simulator for virtually testing autonomous drive software, dynamically creates imagery for the other system in the Constellation pair of servers to analyze and respond to, but that is entirely machine-to-machine imagery communication.

The main exception to this non-visual usage trend is Nvidia RTX, which allows raytracing to be rendered in realtime on GPUs. RTX can be used through Nvidia’s OptiX API, as well as Microsoft’s DirectX RayTracing API, and eventually through the open source Vulkan cross-platform graphics solution. It integrates with Nvidia’s AI Denoiser to use predictive rendering to further accelerate performance, and can be used in VR applications as well.

Nvidia RTX was first announced at the Game Developers Conference last week, but the first hardware to run it was just announced here at GTC, in the form of the new Quadro GV100. This $9,000 card replaces the existing Pascal-based GP100 with a Volta-based solution. It retains the same PCIe form factor, the quad DisplayPort 1.4 outputs and the NV-Link bridge to pair two cards at 200GB/s, but it jumps the GPU RAM per card from 16GB to 32GB of HBM2 memory. The GP100 was the first Quadro offering since the K6000 to support double-precision compute processing at full speed, and the increase from 3,584 to 5,120 CUDA cores should provide a 40% increase in performance, before you even look at the benefits of the 640 Tensor Cores.

Hopefully, we will see simpler versions of the Volta chip making their way into a broader array of more budget-conscious GPU options in the near future. The fact that the new Nvidia RTX technology is stated to require Volta architecture CPUs leads me to believe that they must be right on the horizon.

Nvidia also announced a new all-in-one GPU supercomputer — the DGX-2 supports twice as many Tesla V100 GPUs (16) with twice as much RAM each (32GB) compared to the existing DGX-1. This provides 81920 CUDA cores addressing 512GB of HBM2 memory, over a fabric of new NV-Link switches, as well as dual Xeon CPUs, Infiniband or 100GbE connectivity, and 32TB of SSD storage. This $400K supercomputer is marketed as the world’s largest GPU.

Nvidia and their partners had a number of cars and trucks on display throughout the show, showcasing various pieces of technology that are being developed to aid in the pursuit of autonomous vehicles.

Also on display in the category of “actually graphics related” was the new Max-Q version of the mobile Quadro P4000, which is integrated into PNY’s first mobile workstation, the Prevail Pro. Besides supporting professional VR applications, the HDMI and dual DisplayPort outputs allow a total of three external displays up to 4K each. It isn’t the smallest or lightest 15-inch laptop, but it is the only system under 17 inches I am aware of that supports the P4000, which is considered the minimum spec for professional VR implementation.

There are, of course, lots of other vendors exhibiting their products at GTC. I had the opportunity to watch 8K stereo 360 video playing off of a laptop with an external GPU. I also tried out the VRHero 5K Plus enterprise-level HMD, which brings the VR experience to whole other level. Much more affordable is TP-Cast’s $300 wireless upgrade Vive and Rift HMDs, the first of many untethered VR solutions. HTC has also recently announced the Vive Pro, which will be available in April for $800. It increases the resolution by 1/3 in both dimensions to 2880×1600 total, and moves from HDMI to DisplayPort 1.2 and USB-C. Besides VR products, they also had all sorts of robots in various forms on display.

Clearly the world of GPUs has extended far beyond the scope of accelerating computer graphics generation, and Nvidia is leading the way in bringing massive information processing to a variety of new and innovative applications. And if that leads us to hardware that can someday raytrace in realtime at 8K in VR, then I suppose everyone wins.


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.

V-Ray GPU is Chaos Group’s new GPU rendering architecture

Chaos Group has redesigned its V-Ray RT product. The new V-Ray GPU rendering architecture, according to the company, effectively doubles the speed of production rendering for film, broadcast and design artists. This represents a redesign of V-Ray’s kernel structure, ensuring a dual-blend of high-performance speed and accuracy.

Chaos Group has renamed V-Ray RT to V-Ray GPU, wanting to establish the latter as a professional production renderer capable of supporting volumetrics, advanced shading and other smart tech coming down the road.

Current internal tests have V-Ray GPU running 80 percent faster on the Nvidia’s Titan V, a big gain from previous benchmarks on the Titan Xp, and up to 10-15x faster than an Intel Core i7-7700K, with the same high level of accuracy across interactive and production renders. (For its testing, Chaos Group uses a battery of production scenes to benchmark each release.)

“V-Ray GPU might be the biggest speed leap we’ve ever made,” says Blagovest Taskov, V-Ray GPU lead developer at Chaos Group. “Redesigning V-Ray GPU to be modular makes it much easier for us to exploit the latest GPU architectures and to add functionality without impacting performance. With our expanded feature set, V-Ray GPU can be used in many more production scenarios, from big-budget films to data-heavy architecture projects, while providing more speed than ever before.”

Representing over two years of dedicated R&D, V-Ray GPU builds on nine years of GPU-driven development in V-Ray. New gains for production artists include:

• Volume Rendering – Fog, smoke and fire can be rendered with the speed of V-Ray GPU. It’s compatible with V-Ray Volume Grid, which supports OpenVDB, Field3D and Phoenix FD volume caches.
• Adaptive Dome Light – Cleaner image-based lighting is now faster and even more accurate.
• V-Ray Denoising – Offering GPU-accelerated denoising across render elements and animations.
• Nvidia AI Denoiser – Fast, real-time denoising based on Nvidia OptiX AI-accelerated denoising technology.
• Interface Support – Instant filtering of GPU-supported features lets artists know what’s available in V-Ray GPU (starting within 3ds Max).

V-Ray GPU will be made available as part of the next update of V-Ray Next for 3ds Max beta.

Epic Games, Nvidia team on enterprise solutions for VR app developers

Epic Games and Nvidia have teamed up to offer enterprise-grade solutions to help app developers create more immersive VR experiences.

To help ease enterprise VR adoption, Epic has integrated Nvidia Quadro professional GPUs into the test suite for Unreal Engine 4, the company’s realtime toolset for creating applications across PC, console, mobile, VR and AR platforms. This ensures Nvidia technologies integrate seamlessly into developers’ workflows, delivering results for everything from CAVEs and multi-projection systems through to enterprise VR and AR solutions.

“With our expanding focus on industries outside of games, we’ve aligned ourselves ever more closely with Nvidia to offer an enterprise-grade experience,” explains Marc Petit, GM of the Unreal Engine Enterprise business. “Nvidia Quadro professional GPUs empower artists, designers and content creators who need to work unencumbered with the largest 3D models and datasets, tackle complex visualization challenges and deliver highly immersive VR experiences.”

The Human Race

One project that has driven this effort is Epic’s collaboration with GM and The Mill on The Human Race, a realtime short film and mixed reality experience featuring a configurable Chevrolet Camaro ZL1, which was built using Nvidia Quadro pro graphics.

Says Bob Pette, VP of professional visualization at Nvidia, “Unreal, from version 4.16, is the first realtime toolset to meet Nvidia Quadro partner standards. Our combined solution provides leaders in these markets the reliability and performance they require for the optimum VR experience.”

PNY’s PrevailPro mobile workstations feature 4K displays, are VR-capable

PNY has launched the PNY PrevailPro P4000 and P3000, thin and light mobile workstations. With their Nvidia Max-Q design, these innovative systems are designed from the Quadro GPU out.

“Our PrevailPro [has] the ability to drive up to four 4K UHD displays at once, or render vividly interactive VR experiences, without breaking backs or budgets,” says Steven Kaner, VP of commercial and OEM sales at PNY Technologies. “The increasing power efficiency of Nvidia Quadro graphics and our P4000-based P955 Nvidia Max-Q technology platform, allows PNY to deliver professional performance and features in thin, light, cool and quiet form factors.”

P3000

PrevailPro features the Pascal architecture within the P4000 and P3000 mobile GPUs, with Intel Core i7-7700HQ CPUs and the HM175 Express chipset.

“Despite ever increasing mobility, creative professionals require workstation class performance and features from their mobile laptops to accomplish their best work, from any location,” says Bob Pette, VP, Nvidia Professional Visualization. “With our new Max-Q design and powered by Quadro P4000 and P3000 mobile GPUs, PNY’s new PrevailPro lineup offers incredibly light and thin, no-compromise, powerful and versatile mobile workstations.”

The PrevailPro systems feature either a 15.6-inch 4K UHD or FHD display – and the ability to drive three external displays (2x mDP 1.4 and HDMI 2.0 with HDCP), for a total of four simultaneously active displays. The P4000 version supports fully immersive VR, the Nvidia VRWorks software development kit and innovative immersive VR environments based on the Unreal or Unity engines.

With 8GB (P4000) or 6GB (P3000) of GDDR5 GPU memory, up to 32GB of DDR4 2400MHz DRAM, 512GB SSD availability, HDD options up to 2TB, a comprehensive array of I/O ports, and the latest Wi-Fi and Bluetooth implementations, PrevailPro is compatible with all commonly used peripherals and network environments — and provides pros with the interfaces and storage capacity needed to complete business-critical tasks. Depending on the use case, Mobile Mark 2014 projects the embedded Li polymer battery can reach five hours over a lifetime of 1,000 charge/discharge cycles.

PrevailPro’s thin and light form factor measures 14.96×9.8×0.73 inches (379mm x 248mm x 18mm) and weighs 4.8 lbs.

 

Choosing the right workstation set-up for the job

By Lance Holte

Like virtually everything in the world of filmmaking, the number of available options for a perfect editorial workstation are almost infinite. The vast majority of systems can be greatly customized and expanded, whether by custom order, upgraded internal hardware or with expansion chassis and I/O boxes. In a time when many workstations are purchased, leased or upgraded for a specific project, the workstation buying process is largely determined by the project’s workflow and budget.

One of Harbor Picture Company’s online rooms.

In my experience, no two projects have identical workflows. Even if two projects are very similar, there are usually some slight differences — a different editor, a new camera, a shorter schedule, bigger storage requirements… the list goes on and on. The first step for choosing the optimal workstation(s) for a project is to ask a handful of broad questions that are good starters for workflow design. I generally start by requesting the delivery requirements, since they are a good indicator of the size and scope of the project.

Then I move on to questions like:

What are the camera/footage formats?
How long is the post production schedule?
Who is the editorial staff?

Often there aren’t concrete answers to these questions at the beginning of a project, but even rough answers point the way to follow-up questions. For instance, Q: What are the video delivery requirements? A: It’s a commercial campaign — HD and SD ProRes 4444 QTs.

Simple enough. Next question.

Christopher Lam from SF’s Double Fine Productions/ Courtesy of Wacom.

Q: What is the camera format? A: Red Weapon 6K, because the director wants to be able to do optical effects and stabilize most of the shots. This answer makes it very clear that we’re going to be editing offline, since the commercial budget doesn’t allow for the purchase of a blazing system with a huge, fast storage array.

Q: What is the post schedule? A: Eight weeks. Great. This should allow enough time to transcode ProRes proxies for all the media, followed by offline and online editorial.

At this point, it’s looking like there’s no need for an insanely powerful workstation, and the schedule looks like we’ll only need one editor and an assistant. Q: Who is the editorial staff? A: The editor is an Adobe Premiere guy, and the ad agency wants to spend a ton of time in the bay with him. Now, we know that agency folks really hate technical slowdowns that can sometimes occur with equipment that is pushing the envelope, so this workstation just needs to be something that’s simple and reliable. Macs make agency guys comfortable, so let’s go with a Mac Pro for the editor. If possible, I prefer to connect the client monitor directly via HDMI, since there are no delay issues that can sometimes be caused by HDMI to SDI converters. Of course, since that will use up the Mac Pro’s single HDMI port, the desktop monitors and the audio I/O box will use up two or three Thunderbolt ports. If the assistant editor doesn’t need such a powerful system, a high-end iMac could suffice.

(And for those who don’t mind waiting until the new iMac Pro ships in December, Apple’s latest release of the all-in-one workstation seems to signal a committed return for the company to the professional creative world – and is an encouraging sign for the Mac Pro overhaul in 2018. The iMac Pro addresses its non-upgradability by futureproofing itself as the most powerful all-in-one machine ever released. The base model starts at a hefty $4,999, but boasts options for up to a 5K display, 18-core Xeon processor, 128GB of RAM, and AMD Radeon Vega GPU. As more and more applications add OpenCL acceleration (AMD GPUs), the iMac Pro should stay relevant for a number of years.)

Now, our workflow would be very different if the answer to the first question had instead been A: It’s a feature film. Technicolor will handle the final delivery, but we still want to be able to make in-house 4K DCPs for screenings, EXR and DPX sequences for the VFX vendors, Blu-ray screeners, as well as review files and create all the high-res deliverables for mastering.

Since this project is a feature film, likely with a much larger editorial staff, the workflow might be better suited to editorial in Avid (to use project sharing/bin locking/collaborative editing). And since it turns out that Technicolor is grading the film in Blackmagic Resolve, it makes sense to online the film in Resolve and then pass the project over to Technicolor. Resolve will also cover any in-house temp grading and DCP creation and can handle virtually any video file.

PCs
For the sake of comparison, let’s build out some workstations on the PC side that will cover our editors, assistants, online editors, VFX editors and artists, and temp colorist. PC vs. Mac will likely be a hotly debated topic in this industry for some time, but there is no denying that a PC will return more cost-effective power at the expense of increased complexity (and potential for increased technical issues) than a Mac with similar specs. I also appreciate the longer lifespan of machines with easy upgradability and expandability without requiring expansion chassis or external GPU enclosures.

I’ve had excellent success with the HP Z line — using z840s for serious finishing machines and z440s and z640s for offline editorial workstations. There are almost unlimited options for desktop PCs, but only certain workstations and components are certified for various post applications, so it pays to do certification research when building a workstation from the ground up.

The Molecule‘s artist row in NYC.

It’s also important to keep the workstation components balanced. A system is only as strong as its weakest link, so a workstation with an insanely powerful GPU, but only a handful of CPU cores will be outperformed by a workstation with 16-20 cores and a moderately high-end GPU. Make sure the CPU, GPU, and RAM are similarly matched to get the best bang for your buck and a more stable workstation.

Relationships!
Finally, in terms of getting the best bang for your buck, there’s one trick that reigns supreme: build great relationships with hardware companies and vendors. Hardware companies are always looking for quality input, advice and real-world testing. They are often willing to lend (or give) new equipment in exchange for case studies, reviews, workflow demonstrations and press. Creating relationships is not only a great way to stay up to date with cutting edge equipment, it expands support options, your technical network and is the best opportunity to be directly involved with development. So go to trade shows, be active on forums, teach, write and generally be as involved as possible and your equipment will thank you.

Our Main Image Courtesy of editor/compositor Fred Ruckel.

 


Lance Holte is an LA-based post production supervisor and producer. He has spoken and taught at such events as NAB, SMPTE, SIGGRAPH and Createasphere. You can email him at lance@lanceholte.com.

What was new at GTC 2017

By Mike McCarthy

I, once again, had the opportunity to attend Nvidia’s GPU Technology Conference (GTC) in San Jose last week. The event has become much more focused on AI supercomputing and deep learning as those industries mature, but there was also a concentration on VR for those of us from the visual world.

The big news was that Nvidia released the details of its next-generation GPU architecture, code named Volta. The flagship chip will be the Tesla V100 with 5,120 CUDA cores and 15 Teraflops of computing power. It is a huge 815mm chip, created with a 12nm manufacturing process for better energy efficiency. Most of its unique architectural improvements are focused on AI and deep learning with specialized execution units for Tensor calculations, which are foundational to those processes.

Tesla V100

Similar to last year’s GP100, the new Volta chip will initially be available in Nvidia’s SXM2 form factor for dedicated GPU servers like their DGX1, which uses the NVLink bus, now running at 300GB/s. The new GPUs will be a direct swap-in replacement for the current Pascal based GP100 chips. There will also be a 150W version of the chip on a PCIe card similar to their existing Tesla lineup, but only requiring a single half-length slot.

Assuming that Nvidia puts similar processing cores into their next generation of graphics cards, we should be looking at a 33% increase in maximum performance at the top end. The intermediate stages are more difficult to predict, since that depends on how they choose to tier their cards. But the increased efficiency should allow more significant increases in performance for laptops, within existing thermal limitations.

Nvidia is continuing its pursuit of GPU-enabled autonomous cars with its DrivePX2 and Xavier systems for vehicles. The newest version will have a 512 Core Volta GPU and a dedicated deep learning accelerator chip that they are going to open source for other devices. They are targeting larger vehicles now, specifically in the trucking industry this year, with an AI-enabled semi-truck in their booth.

They also had a tractor showing off Blue River’s AI-enabled spraying rig, targeting individual plants for fertilizer or herbicide. It seems like farm equipment would be an optimal place to implement autonomous driving, allowing perfectly straight rows and smooth grades, all in a flat controlled environment with few pedestrians or other dynamic obstructions to be concerned about (think Interstellar). But I didn’t see any reference to them looking in that direction, even with a giant tractor in their AI booth.

On the software and application front, software company SAP showed an interesting implementation of deep learning that analyzes broadcast footage and other content looking to identify logos and branding, in order to provide quantifiable measurements of the effectiveness of various forms of brand advertising. I expect we will continue to see more machine learning implementations of video analysis, for things like automated captioning and descriptive video tracks, as AI becomes more mature.

Nvidia also released an “AI-enabled” version of I-Ray to use image prediction to increase the speed of interactive ray tracing renders. I am hopeful that similar technology could be used to effectively increase the resolution of video footage as well. Basically, a computer sees a low-res image of a car and says, “I know what that car should look like,” and fills in the rest of the visual data. The possibilities are pretty incredible, especially in regard to VFX.

Iray AI

On the VR front, Nvidia announced a new SDK that allows live GPU-accelerated image stitching for stereoscopic VR processing and streaming. It scales from HD to 5K output, splitting the workload across one to four GPUs. The stereoscopic version is doing much more than basic stitching, processing for depth information and using that to filter the output to remove visual anomalies and improve the perception of depth. The output was much cleaner than any other live solution I have seen.

I also got to try my first VR experience recorded with a Light Field camera. This not only gives the user a 360 stereo look around capability, but also the ability to move their head around to shift their perspective within a limited range (based on the size the recording array). The project they were using to demo the technology didn’t highlight the amazing results until the very end of the piece, but when it did that was the most impressive VR implementation I have had the opportunity to experience yet.
———-
Mike McCarthy is an online editor/workflow consultant with 10 years of experience on feature films and commercials. He has been working on new solutions for tapeless workflows, DSLR filmmaking and multi-screen and surround video experiences. Check out his site.

Review: Nvidia’s new Pascal-based Quadro cards

By Mike McCarthy

Nvidia has announced a number of new professional graphic cards, filling out their entire Quadro line-up with models based on their newest Pascal architecture. At the absolute top end, there is the new Quadro GP100, which is a PCIe card implementation of their supercomputer chip. It has similar 32-bit (graphics) processing power to the existing Quadro P6000, but adds 16-bit (AI) and 64-bit (simulation). It is intended to combine compute and visualization capabilities into a single solution. It has 16GB of new HBM2 (High Bandwidth Memory) and two cards can be paired together with NVLink at 80GB/sec to share a total of 32GB between them.

This powerhouse is followed by the existing P6000 and P5000 announced last July. The next addition to the line-up is the single-slot VR-ready Quadro P4000. With 1,792 CUDA cores running at 1200MHz, it should outperform a previous-generation M5000 for less than half the price. It is similar to its predecessor the M4000 in having 8GB RAM, four DisplayPort connectors, and running on a single six-pin power connector. The new P2000 follows next with 1024 cores at 1076MHz and 5GB of RAM, giving it similar performance to the K5000, which is nothing to scoff at. The P1000, P600 and P400 are all low-profile cards with Mini-DisplayPort connectors.

All of these cards run on PCIe Gen3 x16, and use DisplayPort 1.4, which adds support for HDR and DSC. They all support 4Kp60 output, with the higher end cards allowing 5K and 4Kp120 displays. In regards to high-resolution displays, Nvidia continues to push forward with that, allowing up to 32 synchronized displays to be connected to a single system, provided you have enough slots for eight Quadro P4000 cards and two Quadro Sync II boards.

Nvidia also announced a number of Pascal-based mobile Quadro GPUs last month, with the mobile P4000 having roughly comparable specifications to the desktop version. But you can read the paper specs for the new cards elsewhere on the Internet. More importantly, I have had the opportunity to test out some of these new cards over the last few weeks, to get a feel for how they operate in the real world.

DisplayPorts

Testing
I was able to run tests and benchmarks with the P6000, P4000 and P2000 against my current M6000 for comparison. All of these test were done on a top-end Dell 7910 workstation, with a variety of display outputs, primarily using Adobe Premiere Pro, since I am a video editor after all.

I ran a full battery of benchmark tests on each of the cards using Premiere Pro 2017. I measured both playback performance and encoding speed, monitoring CPU and GPU use, as well as power usage throughout the tests. I had HD, 4K, and 6K source assets to pull from, and tested monitoring with an HD projector, a 4K LCD and a 6K array of TVs. I had assets that were RAW R3D files, compressed MOVs and DPX sequences. I wanted to see how each of the cards would perform at various levels of production quality and measure the differences between them to help editors and visual artists determine which option would best meet the needs of their individual workflow.

I started with the intuitive expectation that the P2000 would be sufficient for most HD work, but that a P4000 would be required to effectively handle 4K. I also assumed that a top-end card would be required to playback 6K files and split the image between my three Barco Escape formatted displays. And I was totally wrong.

Besides when using the higher-end options within Premiere’s Lumetri-based color corrector, all of the cards were fully capable of every editing task I threw at them. To be fair, the P6000 usually renders out files about 30 percent faster than the P2000, but that is a minimal difference compared to the costs. Even the P2000 was able to playback my uncompressed 6K assets onto my array of Barco Escape displays without issue. It was only when I started making heavy color changes in Lumetri that I began to observe any performance differences at all.

Lumetri

Color correction is an inherently parallel, graphics-related computing task, so this is where GPU processing really shines. Premiere’s Lumetri color tools are based on SpeedGrade’s original CUDA processing engine, and it can really harness the power of the higher-end cards. The P2000 can make basic corrections to 6K footage, but it is possible to max out the P6000 with HD footage if I adjust enough different parameters. Fortunately, most people aren’t looking for more stylized footage than the 300 had, so in this case, my original assumptions seem to be accurate. The P2000 can handle reasonable corrections to HD footage, the P4000 is probably a good choice for VR and 4K footage, while the P6000 is the right tool for the job if you plan to do a lot of heavy color tweaking or are working on massive frame sizes.

The other way I expected to be able to measure a difference between the cards would be in playback while rendering in Adobe Media Encoder. By default, Media Encoder pauses exports during timeline playback, but this behavior can be disabled by reopening Premiere after queuing your encode. Even with careful planning to avoid reading from the same disks as the encoder was accessing from, I was unable to get significantly better playback performance from the P6000 compared to the P2000. This says more about the software than it says about the cards.

P6000

The largest difference I was able to consistently measure across the board was power usage, with each card averaging about 30 watts more as I stepped up from the P2000 to the P4000 to the P6000. But they all are far more efficient than the previous M6000, which frequently sucked up an extra 100 watts in the same tests. While “watts” may not be a benchmark most editors worry too much about, among other things it does equate to money for electricity. Lower wattage also means less cooling is needed, which results in quieter systems that can be kept closer to the editor without being distracting from the creative process or interfering with audio editing. It also allows these new cards to be installed in smaller systems with smaller power supplies, using up fewer power connectors. My HP Z420 workstation only has one 6-pin PCIe power plug, so the P4000 is the ideal GPU solution for that system.

Summing Up
It appears that we have once again reached a point where hardware processing capabilities have surpassed the software capacity to use them, at least within Premiere Pro. This leads to the cards performing relatively similar to one another in most of my tests, but true 3D applications might reveal much greater differences in their performance. Further optimization of CUDA implementation in Premiere Pro might also lead to better use of these higher-end GPUs in the future.


Mike McCarthy is an online editor and workflow consultant with 10 years of experience on feature films and commercials. He has been on the forefront of pioneering new solutions for tapeless workflows, DSLR filmmaking and now multiscreen and surround video experiences. If you want to see more specific details about performance numbers and benchmark tests for these Nvidia cards, check out techwithmikefirst.com.

Netflix’s ‘Unbreakable Kimmy Schmidt’ gets crisper look via UHD

NYC’s Technicolor Postworks created a dedicated post workflow for the upgrade.

Having compiled seven Emmy Award nominations in its debut season, Netflix’s Unbreakable Kimmy Schmidt returned in mid-April with 13 new episodes in a form that is, quite literally, bigger and better.

The sitcom, from co-creators Tina Fey and Robert Carlock, features the ever-cheerful and ever-hopeful Kimmy Schmidt, whose spirit refuses to be broken, even after being held captive during her formative years. This season the series has boosted its delivery format from standard HD to the crisper, clearer, more detailed look of Ultra High Definition (UHD).

L-R: Pat Kelleher and Roger Doran

As with the show’s first season, post finishing was done at Technicolor PostWorks New York. Online editor Pat Kelleher and colorist Roger Doran once again served as the finishing team, working under the direction of series producer Dara Schnapper, post supervisor Valerie Landesberg and director of photography John Inwood. Almost everything else, however, was different.

The first season had been shot by Inwood with Arri Alexa, capturing in 1080p, and finished in ProRes 4444. The new episodes were shot with Red Dragon, capturing in 5K, and needed to be finished in UHD. That meant that the hardware and workflow used by Kelleher and Doran had to be retooled to efficiently manage UHD files four times larger than ProRes.

“It was an eye opener,” recalls Kelleher of the change. “Obviously, the amount of drive space needed for storage is huge. Everyone from our data manager through to the people who did the digital deliveries had to contend with the higher volume of data. The actual hands-on work is not that different from an HD show, but you need the horses to do it.”

Before post work began, engineers from Technicolor PostWorks’ in-house research unit, The Test Lab, analyzed the workflow requirements of UHD and began making changes. They built an entirely new hardware Unbreakable Kimmy Schmidtsystem for Kelleher to use, running Autodesk’s Flame Premium. It consisted of an HP Z820 workstation with Nvidia Quadro K6000 graphics, 64GB of RAM and dual Intel Xeon Processor E5-2687Ws (20M Cache, 3.10 GHz, 8.00 GT/s Intel QPI). Kelleher described its performance in handling UHD media as “flawless.”

Doran’s color grading suite got a similar overhaul. For him, engineers built a Linux-based workstation to run Blackmagic’s DaVinci Resolve, V11, and set up a dual monitoring system. That included a Panasonic 300 series display to view media in 1080p and a Samsung 9500 series curved LED to view UHD. Doran could then review color decisions in both formats (while maintaining a UHD signal throughout) and spot details or noise issues in UHD that might not be apparent at lower resolution.

While the extra firepower enabled Kelleher and Doran to work with UHD as efficiently as HD, they faced new challenges. “We do a lot of visual effects for this show,” notes Kelleher. “And now that we’re working in UHD, everything has to be much more precise. My mattes have to be tight because you can see so much more.”

Doran’s work in the color suite similarly required greater finesse. “You have to be very, very aware,” he says. “Cosmetically, it’s different. The lighting is different. You have to pay close attention to how the stars look.”

Doran is quick to add that, while grading UHD might require closer scrutiny, it’s justified by the results. “I like the increased range and greater detail,” he says. “I enjoy the extra control. Once you move up, you never want to go back.”

Both Doran and Kelleher credited the Technicolor PostWorks engineering team of Eric Horwitz, Corey Stewart and Randy Main for their ability to “move up” with a minimum of strain. “The engineers were amazing,” Kelleher insists. “They got the workflow to where all I had to think about was editing and compositing. The transition was so smooth, you almost forgot you were working in UHD, except for the image quality. That was amazing.”

Pixspan at NAB with 4K storage workflow solutions powered by Nvidia

During the NAB Show, Pixspan was demonstrating new storage workflows for full-quality 4K images powered by the Nvidia Quadro M6000. Addressing the challenges that higher resolutions and increasing amounts of data present for storage and network infrastructures, Pixspan is offering a solution that reduces storage requirements by 50-80 percent, in turn supporting 4K workflows on equipment designed for 2K while enabling data access times that are two to four times faster.

Pixspan software and the Nvidia Quadro M6000 GPU together deliver bit-accurate video decoding at up to 1.3GBs per second — enough to handle 4K digital intermediates or 4K/6K camera RAW files in realtime. Pixspan’s solution is based on its bit-exact compression technology, where each image is compressed into a smaller data file while retaining all the information from the original image, demonstrating how the processing power of the Quadro M6000 can be put to new uses in imaging storage and networking to save time and help users  meet tight deadlines.