NVIDIA GeForce RTX 5090 32GB
Blackwell PCIe 2025 4nm
FP32
104.8
TFLOPS
VRAM
32 GB
GDDR7
TDP
575 W
182.3 TFLOPS/kW
Bandwidth
1.8 TB/s
memory
CUDA Cores
21,760
680 tensor cores
Performance Metrics
Peak theoretical throughput by precision type
| Precision | Description | Bits | Peak TFLOPS | Efficiency |
|---|---|---|---|---|
| INT8 | 8-bit integer | 8 | 838.2 | 1.458 TFLOPS/W |
| FP8 | 8-bit floating point | 8 | 838.2 | 1.458 TFLOPS/W |
| FP4 | 4-bit floating point | 4 | 1676.0 | 2.915 TFLOPS/W |
| FP16 | 16-bit floating point | 16 | 419.1 | 0.729 TFLOPS/W |
| BF16 | Brain Float 16 | 16 | 419.1 | 0.729 TFLOPS/W |
| TF32 | TensorFloat-32 | 32 | 209.5 | 0.364 TFLOPS/W |
| FP32 | 32-bit floating point | 32 | 104.8 | 0.182 TFLOPS/W |
| FP64 | 64-bit floating point | 64 | 1.6 | 0.003 TFLOPS/W |
FP8 Efficiency
1.458 TFLOPS/W
838.2 TFLOPS / 575W
FP16 Efficiency
0.729 TFLOPS/W
419.1 TFLOPS / 575W
FP32 Efficiency
0.182 TFLOPS/W
104.8 TFLOPS / 575W
Power Specifications
TDP
575 W
Max Power
661 W
Power Connector
PCIe 16-pin
Cooling
Air
Memory Specifications
Capacity
32 GB
Type
GDDR7
Bandwidth
1792 GB/s
Interface
512-bit
Hardware & Design
Form Factor
PCIe
Architecture
Blackwell
Process Node
4nm
Launch Year
2025
Variant
32GB
Market Segment
Data Center
Transistors
92.0 billion
Die Size
750 mm²
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 | 92.0 billion |
| Die Size | 750 mm² |
| Process Node | 4nm |
| Memory | |
| VRAM | 32 GB |
| Memory Type | GDDR7 |
| Bandwidth | 1.8 TB/s |
| Interface Width | 512-bit |
| Power & Thermal | |
| TDP | 575 W |
| General | |
| Form Factor | PCIe |
| Architecture | Blackwell |
| Launch Year | 2025 |
Common Use Cases
General Compute AI/ML Workloads Data Processing
The NVIDIA GeForce RTX 5090 is optimized for high-performance computing tasks with Blackwell architecture delivering 104.8 TFLOPS of compute power.
Where to Rent
Compare cloud providers offering on-demand GPU instances for AI training, inference, and HPC workloads.
Browse GPU Cloud ProvidersSimilar GPUs
NVIDIA RTX PRO 6000 Blackwell Workstation Edition
PCIe · 2025
FP32: 125.0 TFLOPS
Compare vs RTX PRO 6000 BlackwellStay Updated on GPU Releases
Get notified when new GPUs are added or specifications are updated.
Loading verification...
No spam, unsubscribe anytime.