WHEN WE REVIEWED AMD’s new Radeon Pro W6600, the big sister to this new Radeon Pro W6400 GPU, we made note that the W6600 was partly aimed squarely at the architecture side of the AEC market. This latest GPU is aimed at a wider professional audience that includes both AEC and manufacturing design professionals as well as a wider business-class audience that also needs accelerated performance across a range of post-Covid “new normal” business functions like video chat communications.
Economics
AMD’s latest professional GPU is well-timed for a challenging economic context. After two years in pandemic survival mode, many businesses are seeing exploding demand yet struggling to find qualified talent. The net result means firms must be even smarter today about their information technology budgets as salaries and sign-on bonuses rise.
From this view, AMD’s Radeon Pro W6400 delivers a valuable option, offering the pro-market the most affordable GPU with capacity for real-time, hardware-accelerated raytracing graphics. Like its AMD RDNA 2 Architecture big siblings, the new GPU taps the excellent Radeon Pro Software for graphics management, performance tuning, recording, monitoring, and more. It also supports Viewport Boost technology to accelerate real-time raytracing in applications like Twinmotion and Revit, for example.
AMD launched its Radeon Pro W6400 with real-time raytracing at half the cost of its rival NVIDIA. The new GPU offers all the other benefits of its AMD RDNA 2 Architecture, plus a dead-silent fan and strong OpenGL performance per cost benchmarks.
With a 229 USD MSRP, the W6400 also delivers excellent performance per dollar metrics for a wide range of dominant CAD industry workflows. This is especially true for AEC users working with small to medium-sized models in SketchUp and leading BIM applications, as well as MCAD users working in average-sized models and assemblies predominantly in OpenGL viewports rendered with shaded with edges with and without anti-aliasing. (see our test results section below).
The Raytracing Value Proposition
Most AEC and MCAD users face tight deadlines and tend to optimize their software tools to allow for the fastest workflows. There is considerable pressure (and need) for knowledge about optimizing viewport responsiveness across nearly all CAD/BIM tools. That issue isn’t going away.
Now combined with that pressing demand is the reality that real-time raytracing is truly here and is transforming workflows in both AEC and MCAD. In AEC, it means that architects can loop in views of their building designs with much more realistic raytraced shadows and lighting as part of their iterative design workflow.
One of the things that impressed us with the W6400’s arrival was its packaging. It was quite enjoyable to open this up. Somebody has learned a thing or two from Apple! Click on the image for a larger view. The main specs are on the box.
The industry is also moving (though it will move slowly according to a Tech Soft 3D executive we recently spoke to) from older generation graphics engines like OpenGL to the utilization of more modern, low-level APIs like Vulkan and Metal, and DirectX 12. As CAD and BIM developers modernize their software’s graphics pipeline, real-time raytracing is beginning to factor into future workflow options.
Autodesk is working on its One Graphics system, Dassault Systèmes has Project Romulan, the Nemetschek Group in the near future will proliferate its Redshift Renderer throughout its daughter companies’ digital tools, and Tech Soft 3D has a new partnership with Chaos. From this perspective, AMD’s entry-level workstation GPU gets included (not excluded) in future innovations with rendering engines that will leverage real-time raytracing in the viewport. This should give buyers some assurance.
AMD Radeon Pro W6400
The newest AMD RDNA 2 Architecture workstation GPU is built on a 6nm process node, features 16 MB AMD Infinity Cache (L3), and 4 GB of high-performance GDDR6 memory.
Like its big sister GPUs (W6600, W6800, etc), it features hardware-accelerated raytracing, in this case via 12 Ray Accelerators. Importantly, this newest Radeon Pro GPU is also VR-ready (more on tests below) and supports AMD’s unique Viewport Boost technology useful for apps like Revit and Twinmotion. Other key details include:
- 768 Stream Processors
- PCI Express 4.0 Support for Advanced Data Transfer
- DirectX 12, OpenGL 4.6, OpenCL 2.2, Vulkan 1.2
- 4GB GDDR6 memory with 128-bit memory interface width
- 128 GB/s peak memory bandwidth
- 2 x DisplayPort 1.4
- 7680 x 4320 Max resolution
- 50 watts (Peak Power) / no connectors
- Single-Slot, Half Height, 6.6″ length (SFF capable)
The AMD Radeon Pro W6400 sustains 3.54 Peak Teraflops (FP32), compared to AMD’s Radeon Pro W6600 at 10.4 Peak Teraflops, the bigger sister GPU we do a lot of benchmark comparisons to in this article. The GPU with its 5.4 billion transistors has OS support for Windows 10, 11, and Linux.
Business Apps
Before we jump into our CAD/BIM/3D testing benchmarks, a quick note about the W6400 is in order. AMD sees this GPU as an excellent solution for SFF and entry-level workstations directed at mixed business loads, across various industries, not just those industry segments connected to the CAD markets.
AMD’s own research shows that most users who would buy this GPU prefer two 2K displays, a single 4K display, or a 4K and 2K display dual setup. Thus, AMD saved costs and provided two DisplayPorts. It can also power two 5K displays or a single 8K display.
AMD also sees the Radeon Pro W6400 as an excellent upgrade choice for existing workstation owners using similar-sized SFF based GPUs or as a better option than NVIDIA’s similarly priced T600, a GPU that does not support real-time raytracing at the hardware level. It is also half the price of NVIDA’s newest SFF pro-GPU the RTX A2000, a graphics card that performs similarly to AMD’s W6600 we reviewed recently.
In comparison to the NVIDIA T600, the W6400 performs anywhere from 4 percent to about 63 percent faster at a range of tasks. It is particularly faster at photo editing (63 percent) and rendering and visualization (42 percent), based on AMD’s own benchmarking.
next page: Benchmarking and Performance Results
Reader Comments
Comments for this story are closed