NVIDIA’s Maxwell architecture has already torn up the gaming world, thanks to cards like the GeForce GTX Titan X and the GeForce GTX 980. NVIDIA recently took the time to bring that same Maxwell goodness over the workstation market as well, and the result is the new Quadro M6000, NVIDIA’s new highest-end workstation platform. Like the Titan X, the M6000 is based on the full-fat version of the Maxwell GPU, the G200. Also, like the GeForce GTX Titan X, the Quadro M6000 has 12GB of GDDR5, 3072 GPU cores, 192 texture units (TMUs), and 96 render outputs (ROPs).
At first glance, the M6000 may seem like a relatively small upgrade compared to NVIDIA’s previous top-end GPU, the K6000. Both the older GPU and the newer one have 12GB of RAM, 384-bit memory buses, and roughly similar numbers of cores (the K6000 had 2880 GPU cores compared to 3072 on the newer M6000). The memory clock is slightly higher, at 6.6Gbps, up from 6Gbps, but the on-paper advantages of the M6000 over the K6000 are relatively small.
Sometimes, however, paper specs don’t really tell the whole story. Just as Maxwell delivered substantially better gaming performance than Kepler, NVIDIA has said that the Quadro M6000 will beat out the previous generation Quadro K6000 in a significant way in pro workstation applications as well as GPGPU or rendering and encoding applications that can be GPU-accelerated. Will those claims prove true? That’s what we’re going to look at in-depth here.
As far as board-level comparisons are concerned, the M6000 uses 1x 8-pin PCIe connector, compared to 2x 6-pin connectors for the K6000. It offers four DisplayPort 1.2 ports and a single DVI-I port (the K6000 offered 2x DVI and two DP ports)
NVIDIA’s documentation indicates that the M6000 supports DX12 Hardware Feature Level 12_1, whereas previous GPUs from the Kepler family only support DX12 Feature Level 11_0. Both GPUs will work under DX12, in other words, but certain advanced features of the API will only be available on the M6000.
AMD’s competitive offering
The W9100 still tops the stack, followed by the W8100, W9000, and various other lower-end GPUs. As we covered in July, the W9100 is a full Tahiti-class product, with a 16GB frame buffer, 2816 GPU cores, 320GB/s of memory bandwidth, and six mini-DisplayPorts.
AMD retains the same display output advantage over NVIDIA, therefore, that it had last year. If you need to drive six 4K displays off a single GPU, or you have a GPU-specific workload that requires up to 16GB of frame buffer memory, then the W9100 is your only option on the market today. The M6000 does chip away at one of these advantages — it offers 4x DisplayPort 1.2 ports, compared to the K6000’s two.
The W9100’s ace in the hole may be its overall price/performance ratio. While it wasn’t as fast as the K6000 in all tests, AMD’s highest-end workstation GPU came in a hair over $3000, significantly cheaper than the $5000 price tag of the Quadro M6000, though 12GB K6000s can be had for £4265 currently – which speaks to possibly lower street pricing likewise down the road for the M6000 as well.