RTX PRO 5000 72 GB Blackwell vs Radeon Pro Duo Polaris

Primary details

GPU architecture, market segment, value for money and other general parameters compared.

Place in the rankingnot ratednot rated
Place by popularitynot in top-100not in top-100
ArchitectureGCN 4.0 (2016−2020)Blackwell 2.0 (2025−2026)
GPU code nameEllesmereGB202
Market segmentWorkstationWorkstation
Release date24 April 2017 (8 years ago)21 October 2025 (recently)
Launch price (MSRP)$999 no data

Detailed specifications

General parameters such as number of shaders, GPU core base clock and boost clock speeds, manufacturing process, texturing and calculation speed. Note that power consumption of some graphics cards can well exceed their nominal TDP, especially when overclocked.

Pipelines / CUDA cores2304 ×214080
Core clock speed1243 MHz1740 MHz
Boost clock speedno data2377 MHz
Number of transistors5,700 million92,200 million
Manufacturing process technology14 nm5 nm
Power consumption (TDP)250 Watt300 Watt
Texture fill rate179.0 ×21,045.9
Floating-point processing power5.728 TFLOPS ×266.94 TFLOPS
ROPs32 ×2176
TMUs144 ×2440
Tensor Coresno data440
Ray Tracing Coresno data110
L1 Cache576 KB13.8 MB
L2 Cache2 MB96 MB

Form factor & compatibility

Information on compatibility with other computer components. Useful when choosing a future computer configuration or upgrading an existing one. For desktop graphics cards it's interface and bus (motherboard compatibility), additional power connectors (power supply compatibility).

InterfacePCIe 3.0 x16PCIe 5.0 x16
Length305 mm267 mm
Width2-slot2-slot
Supplementary power connectors1x 6-pin + 1x 8-pin1x 16-pin

VRAM capacity and type

Parameters of VRAM installed: its type, size, bus, clock and resulting bandwidth. Integrated GPUs have no dedicated video RAM and use a shared part of system RAM.

Memory typeGDDR5GDDR7
Maximum RAM amount16 GB ×272 GB
Memory bus width256 Bit ×2384 Bit
Memory clock speed1750 MHz1750 MHz
Memory bandwidth224.0 GB/s ×21.34 TB/s

Connectivity and outputs

This section shows the types and number of video connectors on each GPU. The data applies specifically to desktop reference models (for example, NVIDIA’s Founders Edition). OEM partners often modify both the number and types of ports. On notebook GPUs, video‐output options are determined by the laptop’s design rather than the graphics chip itself.

Display Connectors1x HDMI 2.0b, 3x DisplayPort 1.4a4x DisplayPort 2.1b
HDMI+-

API and SDK support

List of supported 3D and general-purpose computing APIs, including their specific versions.

DirectX12 (12_0)12 Ultimate (12_2)
Shader Model6.76.8
OpenGL4.64.6
OpenCL2.13.0
Vulkan1.31.4
CUDA-12.0
DLSS-+

Pros & cons summary


Recency 24 April 2017 21 October 2025
Maximum RAM amount 16 GB 72 GB
Chip lithography 14 nm 5 nm
Power consumption (TDP) 250 Watt 300 Watt

Pro Duo Polaris has 20% lower power consumption.

RTX PRO 5000 72 GB Blackwell, on the other hand, has an age advantage of 8 years, a 350% higher maximum VRAM amount, and a 180% more advanced lithography process.

We couldn't decide between Radeon Pro Duo Polaris and RTX PRO 5000 72 GB Blackwell. We've got no test results to judge.

Vote for your favorite

Do you think we are right or mistaken in our choice? Vote by clicking "Like" button near your favorite graphics card.


AMD Radeon Pro Duo Polaris
Radeon Pro Duo Polaris
NVIDIA RTX PRO 5000 72 GB Blackwell
RTX PRO 5000 72 GB Blackwell

Other comparisons

We selected several comparisons of graphics cards with performance close to those reviewed, providing you with more options to consider.

Community ratings

Here you can see the user ratings of the compared graphics cards, as well as rate them yourself.


3.2 35 votes

Rate Radeon Pro Duo Polaris on a scale of 1 to 5:

  • 1
  • 2
  • 3
  • 4
  • 5

No user ratings yet.

Rate RTX PRO 5000 72 GB Blackwell on a scale of 1 to 5:

  • 1
  • 2
  • 3
  • 4
  • 5

Comments

Here you can give us your opinion about Radeon Pro Duo Polaris or RTX PRO 5000 72 GB Blackwell, agree or disagree with our ratings, or report errors or inaccuracies on the site.