"The NVIDIA RTX™ A4000 is the most powerful single-slot GPU for professionals, delivering real-time ray tracing, AI-accelerated compute, and high-performance graphics to your desktop"
NVIDIA begin the eagerly-awaited upgrade of their professional Quadro RTX graphics cards with the October 2020 release of the , a replacement for their high-end Quadro RTX6000/8000 graphics cards. The A in A6000 stands for “ampere” which is NVIDIA’S newest GPU architecture. The RTX A6000 proved to be a major advance in both power and speed over the old Quadro cards, which were already highly regarded., a major workstation packager called the RTX A6000, “easily the fastest workstation-class GPU we have ever tested.”
While the A6000 found its niche in the high-end of the workstation graphics card market, the recent (April 2021) release of the NVIDIA and the cards focused on the mid-range markets. These cards retain NVIDIA’S advanced graphics technology but at a lower price point.
Today, we will be reviewing the which enters the professional graphics card market with 16GB’s of EEC DDR6 memory (twice as much as the Turing-based it replaces) housed in a single-slot GPU with a maximum power consumption of 140W delivered through a single 6-pin PCIe connector. In addition to being a beautifully designed graphics card, the RTX A4000 has next-gen RT, Tensor, and CUDA cores to help creators finish complex creative tasks faster and more reliably. And second-generation RTX GPUs like the RT A4000 now supports hardware-accelerated motion blur, allowing these GPUs to significantly speed up motion blur during the rendering process
Built on the 8 nm process0r, and based on the GA104 graphics processor, the RTX A4000 graphics card supports DirectX 12 Ultimate. The GA104 graphics processor is a large chip with a die area of 392 mm² and 17,400 million transistors. It features 6144 shading units, 192 texture mapping units, and 96 ROPs. Also included are 192 third-generation tensor cores which help improve the speed of machine learning applications. The card also has 48 second-generation raytracing acceleration cores. NVIDIA has paired 16 GB GDDR6 ECC memory with the RTX A4000, which are connected using a 256-bit memory interface. The GPU is operating at a frequency of 735 MHz, which can be boosted up to 1560 MHz, memory is running at 1750 MHz (14 Gbps effective).
The RTX A4000 display outputs include four DisplayPort 1.4a ports and can drive up to four displays at 5K resolution. The card is connected to the rest of the system using a PCI-Express 4.0 x16 interface. The card measures 241 mm in length, 112 mm in width, and features a single fan housed in the body of the card. It looks an awful lot like the obelisk in 2001: A Space Odyssey )
The RTX A4000 graphics card has three new features which are unique to the Ampere series of cards.
- PCI Express Gen 4 — The A4000 joins the existing A5000 and A6000 as NVIDIA’s first professional graphics board family to support PCI Express Gen 4, which provides up to 2x the bandwidth of PCI-E Gen 3.
- 8K HDR Capture — Using NVIDIA RTX Experience software, users can record their desktops at resolutions up to 8K allowing them to capture the highest levels of detail.
- AV1 Decode — Traditionally, 8K streaming using H.264 required up to 140 Mbps of Internet bandwidth for smooth real-time playback. With new AV1 decoding, bandwidth is reduced by more than 50%. However, even high-end CPUs struggle to keep up with AV1 decode. With AV1 GPU hardware acceleration on RTX 30-Series GPUs, users can consume up to 8K60 HDR content in AV1 seamlessly from YouTube
You can see from the chart below how the RTX A4000 fits into the tech specs of both of the new A series of cards, but also the older Quadro series and two GeForce gaming cards as well.
Testing the A4000
We conducted many more tests than are listed here, but rather than bombard the reader with data we’ll focus on five major testing methodologies: , , , and the . Remember, raw speed is not the complete measure of a graphic card’s performance. The actual real-world use of a graphics card in a professional workflow is often a better indicator of how well a graphics card actually performs. We will comment on what it’s like to actually work with the RTX A4000 card after sharing the raw test results.
CPU: AMD Ryzen 5 5600X
SSD: Samsung 970 Evo NVMe PCIe M.2 1TB
SSD: Adata XPG SX8200 Pro NVMe PCIe M.2 1TB
HDD: Seagate Barracuda 2TB (2018)
RAM: Crucial Ballistix 3600 MHz DDR4 DRAM Desktop Gaming Memory Kit 16GB
MBD: Gigabyte GA-X570 AORUS ELITE WIFI
Operating System: Windows 10 64 bit
My focus on testing was primarily on content creation, but the A4000 (and the entire series of cards) are also built for the engineering, architectural viz and VR, deep learning and real-time ray-tracing markets. Although I didn’t primarily test in these areas, other reviewers who did have indicated that the A4000 graphics card performs extremely well in these areas. Look at review for an example.
Note that the A4000 runs hotter under load than the RTX 3060 card. Of course, the has two fans and a heat spreader and takes up two full pc lanes which explains why. The peak temperature during testing for the RTX 3060 card was 65c whereas the A4000 got as high as 86c. However, NVIDIA informs me that the card is well within the heat limits it is designed for. You also have to consider that the A series of graphics cards from NVIDIA use a single fan that pulls cool air from the top, then pushes it through a radiator and then blows the hot air out the back of the card instead of directly in the case like the RTX 3060. This enables you to use several A series cards next to each other (which is what I think a lot of workstation creators will do) without heating the areas near the cards. Again, NVIDIA’s careful testing and design are apparent in their considerations for cooling in the RTX A4000 card.
Working with the A4000
I regularly edit videos and create/render 3D content on my mid-range workstation (see specs above). There was a noticeable system-wide increase in the quickness of content creation using the A4000 card. I ran several rendering tests on Da Vinci Resolve and Premiere Pro (using 4k and 2k renders) and the A4000 card was much faster in rendering by about 20-30%. Even setting up the test render was easier because both programs were faster to respond to changes on the timeline.
Another noticeable improvement was in the number of programs I could have open to work on at the same time. This is due to the 2x increase in video RAM size (16GB ECC). At one point I had 6 separate creation programs open at the same time and there was no slowdown. I could easily move from Blender to Unity to Character Creator 3 to Omniverse to Premiere Pro to Audition and each was quite responsive. Omniverse Machinima creation had a high FPS rate as did the Omniverse Create application and both of these are ray-traced rendering live in the viewscreen.
I did a quick gaming benchmark and, of course, the RTX 3060 was the better card in terms of higher FPS, but that’s to be expected. The A4000 didn’t do too bad (10221 score in the ) though despite the card drivers being more focused on quality than speed. It’s possible to use the A4000 card in machinima production as it would still capture the game at a good frame rate and at a decent resolution. I plan on using the A4000 for Omniverse Machinima content creation which I believe is perfectly suited for. Advanced 4K gaming isn’t necessary for machinima creation as you’ll most likely be editing the footage at 1080p anyway. And the A4000 will render more quickly than the RTX 3060.
A professional graphics card is qualitatively different from a consumer graphics card. The focus is much more on power, quality and reliability. This is especially true for team-based content creation. You cannot afford to have a graphics card go down in the middle of work sessions. The NVIDIA RTX A4000 (and all of the A-series cards) comes with pro drivers that have. NVIDIA also puts their professional graphics cards through more rigorous testing than their consumer cards.
On paper, the NVIDIA RTX A4000 is obviously a much stronger performer than the Quadro 4000 it is replacing. You can see via the testing and real-world evaluation noted above that this card is one of the best and most affordable professional graphics cards on the market today. At $1,000 you cannot get a better card for a professional workstation either in a complete package or part of a workstation build. Bringing high-quality NVIDIA Ampere technology plus 16GB DDR6 memory at an affordable price is simply outstanding.
Due to Coronavirus Pandemic, more professionals are working remotely. That isn’t going to significantly change in the future as the work style is very appealing to both workers and employers. That means that those working at home might be prone to setting up their own workstations either by building it themselves or by buying a ready-made solution from a reliable packager like Puget Systems or Dell, HP or Lenovo. The read-made solution might actually be a good bet as the supply situation for graphics cards is insane.
However, as of June 22 when this review was written, I was able to find the RTX A4000 at or near its MSRP at Amazon and B&H Photo. It is also entirely possible that NVIDIA will prioritize its professional cards and will make more of the A series of cards available. Note that Jon Peddie Research (JPR) recently reported quarter-to-quarter graphics add-in board shipments increased by 7.1% for the period and increased significantly by 24.4% year-to-year. That’s good news for consumers.
I’d like to thank NVIDIA and in particular Sean Kilbride, Senior Technical Marketing Manager – Profession Visualization at NVIDIA, for providing this card for review and for helping out with technical questions.