NVIDIA Quadro RTX 8000

The NVIDIA Quadro RTX 8000 is a Turing-generation workstation GPU built on TSMC's 12 nm process, using the TU102 chip with 18,600 million transistors on a 754 mm² die. It carries 48 GB of GDDR6 memory on a 384-bit bus, and its core configuration pairs 4,608 shading units with 72 RT cores and 576 tensor cores. The card's average benchmark score is 31,401, placing it at the 75th percentile of all GPUs in the database. Against its nearest rivals, it sits 0.6% above the NVIDIA GRID M60-1Q, 0.8% above the NVIDIA Quadro M5000, and 0.9% below both the NVIDIA TITAN RTX and AMD Radeon Pro 570X.

Memory Subsystem

The Quadro RTX 8000's defining specification is its memory subsystem. It has 48 GB of GDDR6, a 384-bit bus, and 672.0 GB/s of bandwidth. The memory clock is 1750 MHz, with an effective data rate of 14 Gbps. For high-resolution workloads, this combination means large textures, frame buffers, or geometry data can stay resident on the GPU. The wide bus and high bandwidth allow that data to move quickly between memory and the 4,608 shading units, 288 TMUs, and 96 ROPs. In rendering or compute tasks where memory capacity is the limiting factor, the 48 GB frame buffer is a major asset. The 384-bit interface also balances the 672.0 GB/s bandwidth against the GPU's 169.9 GPixel/s pixel rate and 509.8 GTexel/s texture rate.

Ray Tracing and Feature Set

Hardware ray tracing is present through 72 dedicated RT cores. Tensor processing is covered by 576 tensor cores, making the GPU ready for workloads that offload matrix math to dedicated hardware. API support includes DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4, so modern graphics and compute interfaces are available. The card also provides four DisplayPort 1.4a outputs and one USB Type-C port, connected through a PCIe 3.0 x16 host interface. For a workstation GPU, this feature set combines rendering, compute, and display connectivity in one dual-slot package.

Benchmark Performance

The average benchmark score is 31,401. That result places the card at the 75th percentile, meaning it outperforms 75% of all GPUs in the database. In Geekbench, scores are 125,554 for OpenCL and 125,781 for Vulkan. Passmark results are more varied: G3D 19,799, GPU compute 9,992, G2D 866, DirectX 9 211, DirectX 10 137, DirectX 11 188, and DirectX 12 79. The high Geekbench figures indicate strong compute throughput. The GPU is rated at 16.31 TFLOPS FP32 and 32.62 TFLOPS FP16 with a 2:1 ratio, running at 1395 MHz base and 1770 MHz boost. Pixel and texture rates are 169.9 GPixel/s and 509.8 GTexel/s respectively.

Against nearest rivals, the differences are small: 0.6% above the GRID M60-1Q, 0.8% above the Quadro M5000, and 0.9% below both the TITAN RTX and Radeon Pro 570X. The data shows a tightly clustered group rather than a dominant leader. The average benchmark scores of the four cards fall within a narrow band, so aggregate performance parity is the main conclusion.

How It Compares

NVIDIA GRID M60-1Q: The Quadro RTX 8000's average score of 31,401 is 0.6% higher than the GRID M60-1Q's 31,220. The margin is small enough that the two are effectively at parity in aggregated benchmarks.

NVIDIA Quadro M5000: Against the Quadro M5000's 31,142, the RTX 8000 leads by 0.8%. This is a narrow lead that does not represent a meaningful day-to-day performance gap.

NVIDIA TITAN RTX: The TITAN RTX averages 31,676, which is 0.9% higher than the RTX 8000. The RTX 8000 trails, but by less than a percentage point.

AMD Radeon Pro 570X: The Radeon Pro 570X averages 31,682, also 0.9% above the RTX 8000. The positions are reversed, but the margin is identical to the TITAN RTX comparison, reinforcing the parity conclusion.

Power and Cooling

The card has a TDP of 260 W. The suggested power supply is 600 W. Power is delivered through one 6-pin and one 8-pin PCIe power connector. The cooler occupies a dual-slot width, and the board measures 267 mm / 10.5 inches in length and 111 mm / 4.4 inches in height. Production status is end-of-life, with the Quadro Volta listed as predecessor and Workstation Ampere as successor. Released on 2018-08-12, the card uses a PCIe 3.0 x16 interface.

FAQ

Q: How much VRAM does the Quadro RTX 8000 have?

A: It has 48 GB of GDDR6 memory on a 384-bit bus, with 672.0 GB/s bandwidth and an effective data rate of 14 Gbps.

Q: Does it support hardware ray tracing?

A: Yes. It has 72 RT cores and 576 tensor cores. API support includes DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4.

Q: How does its average benchmark score compare with its nearest rivals?

A: Its average is 31,401. That is 0.6% above the NVIDIA GRID M60-1Q, 0.8% above the NVIDIA Quadro M5000, and 0.9% below both the NVIDIA TITAN RTX and AMD Radeon Pro 570X. It also ranks at the 75th percentile of all GPUs.

Q: What power supply and connector setup is recommended?

A: The TDP is 260 W, the suggested PSU is 600 W, and the card requires one 6-pin plus one 8-pin power connector.

Q: What is the launch MSRP?

A: The launch MSRP is 9,999 USD.

Q: What display outputs are available?

A: The card provides 4x DisplayPort 1.4a and 1x USB Type-C outputs.

Who Should Consider It

The 48 GB frame buffer and 672.0 GB/s bandwidth make the Quadro RTX 8000 a candidate for high-resolution rendering and large dataset workloads. Its 75th percentile average score means it sits above the majority of GPUs in the database, while the nearest rivals are within 0.9% in either direction, so raw aggregate performance is not the deciding factor among this group. Workloads that use hardware ray tracing can utilize the 72 RT cores, and tensor-based processing can use the 576 tensor cores. The FP16 rate of 32.62 TFLOPS at 2:1 and FP32 rate of 16.31 TFLOPS describe a compute-oriented GPU. However, because production status is end-of-life and the successor is Workstation Ampere, consideration should focus on the memory capacity and feature set rather than expecting a commanding benchmark lead over the nearest competitors.