Superseded. NVIDIA replaced this card with the 24 GB RTX 5090 D V2 in August 2025 and the nvidia.cn product page now redirects there, so the specifications here come from the archived original. Only FP32, FP64 and FP4 are recorded: NVIDIA publishes a single capped AI TOPS figure for this SKU and the usual family ratios provably do not hold, so the intermediate precisions are refused rather than guessed.

NVIDIA GeForce RTX 5090 D 32GB

Blackwell PCIe 2025 4nm Spec confidence: Official
FP32
105
TFLOPS
VRAM
32 GB
GDDR7
TDP
575 W
182.3 TFLOPS/kW
Download datasheet

Overview

The GeForce RTX 5090 D was NVIDIA's China export version of the RTX 5090, announced alongside it at CES 2025. It keeps the full 21,760 CUDA cores, the same 2.41 GHz boost clock, the same 32 GB of GDDR7 on a 512-bit bus and the same 575 W board power, so its FP32 shader throughput is identical at 104.8 TFLOPS. What is cut is Tensor Core throughput: NVIDIA rates it at 2,375 AI TOPS against the RTX 5090's 3,352, which works out to 1,187.5 TFLOPS of dense FP4 rather than 1,676. That is 11.3 times the FP32 rate where every other Blackwell GeForce card sits at 16 times, so the cap is applied somewhere inside the Tensor Core rather than by simply reducing clocks. NVIDIA has never said where, which is why this row carries FP32, FP64 and FP4 and nothing between them. The card was withdrawn in 2025 and replaced by the RTX 5090 D V2.

Performance

Peak theoretical throughput by precision type

PrecisionDense2:4 Sparse
FP64
64-bit floating point
1.6TFLOPS Structured sparsity is a tensor-core feature; this vector precision has no sparse form
FP32
32-bit floating point
105TFLOPS Structured sparsity is a tensor-core feature; this vector precision has no sparse form
TF32
No verified data available The part supports 2:4 sparsity, but the vendor publishes no sparse figure for this precision
BF16
No verified data available The part supports 2:4 sparsity, but the vendor publishes no sparse figure for this precision
FP16
No verified data available The part supports 2:4 sparsity, but the vendor publishes no sparse figure for this precision
FP8
No verified data available The part supports 2:4 sparsity, but the vendor publishes no sparse figure for this precision
FP6
No verified data available The part supports 2:4 sparsity, but the vendor publishes no sparse figure for this precision
FP4
4-bit floating point
1,188TFLOPS 2,375TFLOPS
INT8
No verified data available The part supports 2:4 sparsity, but the vendor publishes no sparse figure for this precision

Specifications

Architecture

Blackwell

Form Factor

PCIe

Launch Year

2025

Process Node

4nm

Memory

32 GB GDDR7

Bandwidth

No verified data available

TDP

575 W

Max Power

575 W

Transistors

92.2 billion

CUDA Cores

21,760

Spec Confidence

Official

Full Specifications

Compute Engine
CUDA Cores 21,760
Tensor Cores 680 (5th Gen)
Streaming Multiprocessors 170
Base Clock 2.01 GHz
Boost Clock 2.41 GHz
Chip Design
Transistors 92.2 billion
Die Size 750 mm²
Process Node 4nm
Memory
VRAM 32 GB
Memory Type GDDR7
Bandwidth No verified data available
Interface Width 512-bit
Interconnect & I/O
PCIe 5.0
Cache
L2 Cache 96 MB
Power & Thermal
TDP 575 W
Max Board Power 575 W
Power Connector 4x PCIe 8-pin or 1x 600 W PCIe Gen 5
Cooling Active
Enterprise Features
Sparsity Yes
Physical & Media
Card Length 304 mm
Width 2-slot
NVENC Engines 3
NVDEC Engines 2
General
Form Factor PCIe
Architecture Blackwell
Process Node 4nm
Launch Year 2025

Datasheet & Resources

Data Provenance

Every figure traced to a source

Primary Source

Publisher
NVIDIA
Published
2025-01-08

Data Quality

Spec confidence
Official
Clock basis
Boost
Core precisions with figures
3 of 9
Normalization
All values in TFLOPS

Compare cloud providers offering on-demand GPU instances for AI training, inference, and HPC workloads.

Browse GPU Cloud Providers

Similar GPUs

Frequently Asked Questions

How many TFLOPS does the NVIDIA GeForce RTX 5090 D have?

The NVIDIA GeForce RTX 5090 D delivers 105 TFLOPS FP32 at peak. Flopper does not currently have verified FP16 and FP8 throughput figures for it.

What is the power consumption of the NVIDIA GeForce RTX 5090 D?

The NVIDIA GeForce RTX 5090 D has a TDP (Thermal Design Power) rating of 575 watts.

How much memory does the NVIDIA GeForce RTX 5090 D have?

The NVIDIA GeForce RTX 5090 D is equipped with 32 GB of VRAM. Flopper does not currently have a verified memory bandwidth figure for it.

What architecture is the NVIDIA GeForce RTX 5090 D based on?

The NVIDIA GeForce RTX 5090 D is based on the Blackwell architecture, launched in 2025.

Stay Updated on GPU Releases

Get notified when new GPUs are added or specifications are updated.

Loading verification...

No spam, unsubscribe anytime.

Back to GPUs