GeForce GTX 580 vs GeForce GTX 660
Performance score
GeForce GTX 580 is 11% faster than GeForce GTX 660 by our estimate.
The score is our own estimate from the specs (core count, clock, architecture), not FPS from benchmarks. 100 = the fastest card in our database.
Real tests
What the test measures
Blender is a popular free 3D graphics program. The test makes the device render three ready-made 3D scenes (light, shadows, reflections) and counts how many render passes it finishes per minute. The higher the number, the faster the device handles heavy work: 3D, video editing, photo processing. On a CPU the test loads all cores at once, so it shows the chip's full power; on a graphics card it shows its compute strength.
GPU rendering (Blender)
Median of real test runs on users' computers. It does not show FPS in games. Data updated: 31.10.2026.
Overview
| GeForce GTX 580 | GeForce GTX 660 | |
|---|---|---|
| Score (0–100) | 4.3 | 3.88 |
| Ranking position | 235 | 239 |
| Power efficiency (0–100) | 8.45 | 13.29 |
| Manufacturer | NVIDIA | NVIDIA |
| Type | Desktop | Desktop |
| Architecture | Fermi | Kepler |
Specifications
| GeForce GTX 580 | GeForce GTX 660 | |
|---|---|---|
| Shader cores | 512 | 960 |
| Base clock | 772 MHz | 980 MHz |
| Boost clock | 1544 MHz | 1033 MHz |
| FP32 TFLOPS (calculated) | 1.58 | 1.98 |
| Hardware ray tracing | no | no |
Video memory
| GeForce GTX 580 | GeForce GTX 660 | |
|---|---|---|
| VRAM | 1 GB | 2 GB |
| Memory type | GDDR5 | GDDR5 |
| Memory bus | 384 bit | 192 bit |
| Memory bandwidth | 192 GB/s | 144 GB/s |
Power and connection
| GeForce GTX 580 | GeForce GTX 660 | |
|---|---|---|
| Board power | 244 W | 140 W |
Pros and cons
GeForce GTX 580
- 11% higher overall score
- higher memory bandwidth: 192 vs 144 GB/s
GeForce GTX 660
- more VRAM: 2 GB vs 1 GB
- lower power draw: 140 W vs 244 W
Other comparisons
Visitor rating
Questions and comments
Share your experience, ask a question or tell us about a mistake on this page: GeForce GTX 580 / GeForce GTX 660.
No comments yet — be the first.