NVIDIA GeForce RTX 4090 vs NVIDIA GeForce RTX 5090
In the PassMark G3D benchmark the NVIDIA GeForce RTX 5090 scores 38,961 against 38,050 for the NVIDIA GeForce RTX 4090, making it about 2 percent faster than the NVIDIA GeForce RTX 4090.
The NVIDIA GeForce RTX 5090 carries more video memory (32 GB against 24 GB), which matters at higher resolutions and for larger workloads.
The NVIDIA GeForce RTX 4090 draws less board power (450 W against 575 W).
Measured as benchmark score per watt, the NVIDIA GeForce RTX 4090 is the more efficient of the two (84.6 against 67.8 points per watt).
Comparing 16 specifications that differ between these two cards. Coverage: 10/10 recorded for the NVIDIA GeForce RTX 4090, 10/10 for the NVIDIA GeForce RTX 5090. Fields neither card records are omitted; fields only one records are shown as "Not recorded".
Specifications compared
| NVIDIA GeForce RTX 4090 | NVIDIA GeForce RTX 5090 | |
|---|---|---|
| Shading units | 16,384 | 21,760 |
| Ray tracing cores | 128 | 170 |
| Tensor cores | 512 | 680 |
| Base clock | 2.24 GHz | 2.02 GHz |
| Boost clock | 2.52 GHz | 2.41 GHz |
| Memory size | 24 GB | 32 GB |
| Memory type | GDDR6X | GDDR7 |
| Memory bus width | 384-bit | 512-bit |
| Memory bandwidth | 1,010.0 GB/s | 1,792.0 GB/s |
| Board power | 450 W | 575 W |
| Process node | 5 nm | 4 nm |
| Die size | 609 mm² | 750 mm² |
| Transistors | 76,300 M | 92,200 M |
| Architecture | Ada Lovelace | Blackwell |
| Launch date | 2022-10-12 | 2025-01-30 |
| PassMark G3D | 38,050 | 38,961 |