NVIDIA RTX PRO 6000D Blackwell Max-Q

The NVIDIA RTX PRO 6000D Blackwell Max-Q is an actively produced workstation GPU released on 2025-03-17. It is built around the GB202 chip on Blackwell 2.0 architecture, manufactured by TSMC on a 5 nm process with 92,200 million transistors on a 750 mm² die. The core configuration includes 24,064 shading units, 752 TMUs, 192 ROPs, 188 RT cores, and 752 tensor cores. Base clock is 1590 MHz, boost clock is 2288 MHz. The database entry carries no measured scores: the benchmarks array is empty, the average benchmark score is 0, and nearestRivals is empty, so this profile is a specification-level analysis rather than a benchmark report.

Power and Cooling

The card is rated at 300 W TDP. The data sheet suggests a 700 W PSU for systems using this GPU, and the power input is a single 16-pin connector. Builders should confirm that the PSU has that connector rather than relying only on total wattage. The interface is PCIe 5.0 x16, which is the listed host connection.

The cooling solution is dual-slot. Physical dimensions are 267 mm (10.5 inches) long, 111 mm (4.4 inches) high, and 40 mm (1.6 inches) wide. That means the card needs two expansion slots of clearance and enough case length for the 267 mm board. The Max-Q designation appears in the product name alongside the 300 W TDP, so this listing describes a power-constrained workstation card rather than an unlimited-boost model.

The 5 nm process and 122.9M/mm² transistor density help place the 92,200 million transistor GB202 die in a 300 W envelope. The dual-slot cooler is the only cooling detail listed, and the 700 W suggested PSU is the reference point for system power planning. Since the data records exactly one 16-pin connector, there is no second connector requirement to plan for.

Ray Tracing and Feature Set

Hardware ray tracing is handled by 188 RT cores. Tensor compute is handled by 752 tensor cores. These sit on the Blackwell 2.0 architecture alongside 24,064 shading units, 752 TMUs, and 192 ROPs.

The recorded API support is DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. That DirectX feature level is the highest listed in the profile, while Vulkan 1.4 and OpenGL 4.6 provide additional cross-platform graphics coverage. The display output is four DisplayPort 2.1b connectors, so external display connectivity is entirely through DisplayPort in this data.

Ray tracing and tensor workloads are not separately benchmarked in this entry. The 188 RT cores and 752 tensor cores indicate where the hardware resources are allocated, but no frame-rate, render-time, or inference-time scores exist in the profile. Pixel rate is 439.3 GPixel/s and texture rate is 1,720.6 GTexel/s, which are the listed fill-rate figures associated with the ROP and TMU arrays.

Benchmark Performance

The benchmark data for this GPU is empty. There are no entries in the benchmarks array, the average benchmark score is 0, and nearestRivals is empty. Because nearestRivals has no names, scores, or deltaPct values, no percentage comparison against rival GPUs is possible from this profile.

The only comparative field is percentileVsAllGpus, which is 50. In a fully populated database, that would place a GPU at the midpoint. Here, with a zero average benchmark score and no benchmark records, that percentile is not supported by measured runs. The correct interpretation is that this GPU has no benchmark ranking yet.

The specification sheet does provide performance-relevant throughput figures. FP32 compute is 110.1 TFLOPS, and FP16 is also 110.1 TFLOPS with a 1:1 ratio. Texture rate is 1,720.6 GTexel/s, and pixel rate is 439.3 GPixel/s. The base clock is 1590 MHz and boost clock is 2288 MHz, so the listed clocks are the governing frequency figures. These numbers describe raw throughput ceilings, not application-level benchmark results.

The predecessor field is Workstation Ada, but without benchmark scores for either product, no performance delta can be computed. Since no exact rival deltas are available, any percentage comparison would be unsupported by the data. The database profile currently lacks the measurements needed to rank this GPU against its peers.

FAQ

Q: How much memory does this GPU have, and what type is it?

A: The GPU has 96 GB of GDDR7 memory on a 512-bit bus, with 1.79 TB/s bandwidth and 28 Gbps effective memory speed (1750 MHz memory clock).

Q: What are the power requirements?

A: The TDP is 300 W, the suggested PSU is 700 W, and the card uses one 16-pin power connector.

Q: What display outputs are listed?

A: Four DisplayPort 2.1b outputs.

Q: Does the database include benchmark comparisons against rivals?

A: No. The benchmarks array is empty, the average benchmark score is 0, and nearestRivals is empty, so no rival names or deltaPct values exist in the entry.

Q: Which graphics APIs are supported?

A: DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4 are the API entries in the profile.

Q: What is the launch MSRP?

A: The launch MSRP is 8,565 USD.

Who Should Consider It

This card is for users whose workloads depend on memory capacity and high raw throughput. The 96 GB GDDR7 frame buffer is the standout specification in the profile. High-resolution rendering, large texture sets, and datasets that need to stay resident on the GPU are the natural cases for that capacity.

The 512-bit bus and 1.79 TB/s bandwidth feed the 24,064 shading units, and the 1,720.6 GTexel/s texture rate supports detail-heavy shading work. Pixel rate of 439.3 GPixel/s gives the ROP backend enough output capability for high-resolution displays. Four DisplayPort 2.1b connectors make multi-monitor workstation setups the expected deployment, since no other display outputs are listed.

The generation field identifies this as a Blackwell PRO W (x000) product, and the production status is Active. That means it is a current workstation catalog item rather than a legacy part. The 188 RT cores and 752 tensor cores add ray tracing and matrix-compute resources, but without benchmark scores, no definitive application-level performance claim can be made. The 300 W TDP and 700 W suggested PSU should be part of any system design with this GPU. Users with workloads that fit in smaller memory pools may not need this capacity, but the database does not list an alternative memory configuration to compare.

Memory Subsystem

The memory subsystem is defined by 96 GB of GDDR7, a 512-bit bus, and 1.79 TB/s bandwidth. Memory clock is 1750 MHz, with 28 Gbps effective signaling. These are the complete memory specifications in the profile.

For high-resolution workloads, capacity and bandwidth do different jobs. The 96 GB capacity determines how much geometry, texture data, and render data can be held on the GPU. The 1.79 TB/s bandwidth determines how quickly that data can be moved to the cores. Texture operations depend on this bandwidth as well, since the 752 TMUs feed the 1,720.6 GTexel/s texture rate. Pixel output also consumes memory bandwidth, and the 192 ROPs sustain the 439.3 GPixel/s pixel rate.

The 512-bit bus gives the memory controller a wide path to the frame buffer, while the GDDR7 type is listed as the explicit memory generation. No other memory type or bus width appears in the entry. The memory clock and effective speed are both stated, so the data supports planning around a 28 Gbps effective memory design. In combination with the 300 W TDP, this memory subsystem is the part of the specification most directly linked to high-resolution behavior.