GeekRadar / The comparison desk

COMPARE.

GPU

NVIDIA

NVIDIA RTX PRO 5000 72 GB Blackwell

Graphics processor

Video memory: 72 GB^*
Memory bus: 384 bit^*
Rated memory bandwidth: 1,340 GB/s^*

No matched benchmark results are available for this entry yet.

Processing & platform

Architecture: Blackwell 2.0^*
Process node: 5 nm^*
GPU cores: 14,080^*
Graphics interface: PCIe 5.0 x16^*
Reported chip die area: 750 mm²^*
Transistors: 92,200 million^*
GPU L2 cache: 96 MB^*
Graphics processor: GB202^*
Texture mapping units: 440^*
Raster output units: 160^*
Streaming multiprocessors: 110^*
Tensor cores: 440^*
Ray tracing cores: 110^*
Chip foundry: TSMC^*
Card or module format: Dual-slot^*
Declared GPU TDP: 300 W^*
Suggested system PSU: 700 W^*
External power connectors: 1x 16-pin^*
DirectX / feature level: 12 Ultimate (12_2)^*
OpenGL: 4.6^*
OpenCL: 3.0^*
Vulkan: 1.4^*
CUDA compute capability: 12.0^*
Shader model: 6.9^*

Memory & storage

Memory options: 72^*
Video memory: 72 GB^*
Video memory type: GDDR7^*
Memory bus: 384 bit^*
Rated memory bandwidth: 1,340 GB/s^*

Power specifications

Thermal design power: 300 W^*

System configuration

Detailed specifications

Specifications 1 · 12 fields

GPU architecture — NVIDIA RTX PRO 5000 72GB Blackwell — 72 GB GDDR7

Blackwell^*

Theoretical peak: FP32 non-tensor dense per GPU — NVIDIA RTX PRO 5000 72GB Blackwell — 72 GB GDDR7

65^* TFLOPS

Calculated theoretical peak: FP4 tensor dense per GPU — NVIDIA RTX PRO 5000 72GB Blackwell — 72 GB GDDR7

1,032^* TFLOPS

Scope and conditions

Calculation: 2064 / 2^*

Theoretical peak: FP4 tensor with sparsity per GPU — NVIDIA RTX PRO 5000 72GB Blackwell — 72 GB GDDR7

2,064^* TFLOPS

Physical configuration — NVIDIA RTX PRO 5000 72GB Blackwell — 72 GB GDDR7

PCIe, full height, dual slot (4.4 in x 10.5 in L)^*

PCIe interface declaration — NVIDIA RTX PRO 5000 72GB Blackwell — 72 GB GDDR7

PCIe Gen 5 x16^*

Maximum board power — NVIDIA RTX PRO 5000 72GB Blackwell — 72 GB GDDR7

300^* W

Peak memory bandwidth per GPU — NVIDIA RTX PRO 5000 72GB Blackwell — 72 GB GDDR7

1,344^* GB/s

Memory interface width per GPU — NVIDIA RTX PRO 5000 72GB Blackwell — 72 GB GDDR7

384^* bit

Memory capacity per GPU — NVIDIA RTX PRO 5000 72GB Blackwell — 72 GB GDDR7

72^* GB

Memory technology — NVIDIA RTX PRO 5000 72GB Blackwell — 72 GB GDDR7

GDDR7^*

GPU model — NVIDIA RTX PRO 5000 72GB Blackwell — 72 GB GDDR7

NVIDIA RTX PRO 5000 72GB Blackwell^*

Specifications 2 · 31 fields

proc transistors: 92,200 million^*
proc die size: 750^* mm²
cores: 14,080^*
tensor cores: 440^*
rt cores: 110^*
sm count: 110^*
base clock: 1,740^*
boost clock: 2,377^*
memory clock: 1,750^*
ram: 72^*
name: NVIDIA RTX PRO 5000 72 GB Blackwell^*
ram type: GDDR7^*
bus width: 384^*
bus interface: PCIe 5.0 x16^*
ram bandwidth: 1.34 TB/s^*
fp16 performance: 66.94^*
fp32 performance: 66.94^*
fp64 performance: 1.0459^*
tdp: 300^* W
suggested psu: 700^* W
slot width: Dual-slot^*
power connectors: 1x 16-pin^*
api cuda: 12^*
is retail board: No^*
gpi value: 35.64^*
vendor: nvd^*
manufacturer: NVIDIA^*
arch name: Blackwell 2.0^*
card release date: 2025-10-21^*
proc foundry: TSMC^*
proc process size: 5^* nm

Specifications 3 · 37 fields

CUDA compute capability

12.0^*

DirectX / feature level

12 Ultimate (12_2)^*

OpenCL

3.0^*

OpenGL

4.6^*

Shader model

6.9^*

Vulkan

1.4^*

Reported theoretical BF16 peak

133.9^* TFLOPS

Scope and conditions

Arithmetic density: Not stated^*

Graphics interface

PCIe 5.0 x16^*

Memory bus

384^* bit

Reported shader processors

14,080^*

Reported theoretical FP16 peak

66.94^* TFLOPS

Scope and conditions

Arithmetic density: Not stated^*

Reported theoretical FP32 peak

66.94^* TFLOPS

Scope and conditions

Arithmetic density: Not stated^*

Reported theoretical FP64 peak

1.046^* TFLOPS

Scope and conditions

Arithmetic density: Not stated^*

Declared L1 cache configuration

128 KB (per SM)^*

Declared GPU L2 cache

96 MB^*

Reported display outputs

4x DisplayPort 2.1b^*

Theoretical pixel fill rate

380.3 GPixel/s^*

External power connectors

1x 16-pin^*

Reported chip die area

750 mm²^*

Chip foundry

TSMC^*

Process node declaration

5 nm^*

Transistors

92.2 billion^*

Graphics processor

GB202^*

Reported video memory

72^* GB

Rated memory bandwidth

1.34 TB/s^*

Video memory type

GDDR7^*

Raster output units

160^*

Ray tracing cores

110^*

Reported shading units

14,080^*

Card or module format

Dual-slot^*

Streaming multiprocessors

110^*

Suggested system PSU

700 W^*

Declared GPU TDP

300 W^*

Tensor cores

440^*

Theoretical texture fill rate

1,045.9 GTexel/s^*

Reported theoretical TF32 peak

66.94^* TFLOPS

Scope and conditions

Arithmetic density: Not stated^*

Texture mapping units

440^*

* Provisional data