
GeForce GTX 660

GeForce RTX 4070
GeForce GTX 660 vs GeForce RTX 4070 Performance Spectrum
About G3D Mark
G3D Mark is a standard benchmark that measures graphics performance in real-world gaming scenarios. It simplifies comparing cards from different brands, where higher scores directly correlate with better fps and smoother gaming experiences.
GeForce GTX 660 vs GeForce RTX 4070 FPS Benchmarks
Predicted gaming performance across popular games. Tested paired with Ryzen 7 9800X3D to isolate GPU performance.
Search any supported game below to compare 1080p FPS for both components.

Path of Exile 2

Counter-Strike 2

League of Legends

Valorant

Among Us

Apex Legends

ARC Raiders

Baldur's Gate 3

Call of Duty: Black Ops 6
GeForce GTX 660 vs GeForce RTX 4070: Pros, Cons & Final Verdict
See where each GPU makes more sense in practice: raw FPS, VRAM, features, power draw, pricing, and long-term headroom.
GeForce GTX 660
2012Why buy it
- ✅Costs $370 less on MSRP ($229 MSRP vs $599 MSRP).
- ✅Draws 140W instead of 200W, a 60W reduction.
- ✅Measures 241mm instead of 304mm, a 63mm shorter card that is more SFF-friendly.
Trade-offs
- ❌Lower average FPS than GeForce RTX 4070 across 50 tracked games in our benchmark data.
- ❌Less VRAM, with 3 GB vs 12 GB for high-resolution textures and newer games.
- ❌No equivalent frame-generation stack like DLSS 3.5 + Frame Generation (2023).
- ❌2012 hardware with 3 GB of VRAM is already well past its comfortable zone for modern gaming, so it is hard to recommend now.
- ❌Lower G3D Mark per dollar, at 17.6 vs 44.9 G3D/$ ($229 MSRP vs $599 MSRP).
GeForce RTX 4070
2023Why buy it
- ✅546.0% more average FPS across 50 tracked games in our benchmark data.
- ✅Delivers 154.7% more G3D Mark for each dollar spent, at 44.9 vs 17.6 G3D/$ ($599 MSRP vs $229 MSRP).
- ✅Access to a newer frame-generation stack with DLSS 3.5 + Frame Generation (2023).
- ✅300% more VRAM for high-resolution textures and newer games (12 GB vs 3 GB).
- ✅Better long-term bet: Ada Lovelace (2022−2024) on 5nm gives it a newer hardware base for upcoming games.
Trade-offs
- ❌42.9% higher power demand at 200W vs 140W.
- ❌26.1% longer card at 304mm vs 241mm.
Quick Answers
Which GPU is faster for gaming right now?
Which GPU is the safer long-term pick for 2026 and beyond?
Which GPU is the better buy today?
GeForce GTX 660 vs GeForce RTX 4070 Technical Specifications
Side-by-side specs, architecture details, clocks, memory, power, and platform differences.

GeForce GTX 660
The GeForce GTX 660 is manufactured by NVIDIA. It was released in September 6 2012. It features the Kepler architecture. The core clock ranges from 980 MHz to 1033 MHz. It has 960 shading units. The thermal design power (TDP) is 140W. Manufactured using 28 nm process technology. G3D Mark benchmark score: 4,040 points. Launch price was $229.

GeForce RTX 4070
The GeForce RTX 4070 is manufactured by NVIDIA. It was released in April 12 2023. It features the Ada Lovelace architecture. The core clock ranges from 1920 MHz to 2475 MHz. It has 5888 shading units. The thermal design power (TDP) is 200W. Manufactured using 5 nm process technology. It features 46 dedicated ray tracing cores for enhanced lighting effects. G3D Mark benchmark score: 26,919 points. Launch price was $599.
Graphics Performance
In G3D Mark, the GeForce GTX 660 scores 4,040 versus the GeForce RTX 4070's 26,919 — the GeForce RTX 4070 leads by 566.3%. The GeForce GTX 660 is built on Kepler while the GeForce RTX 4070 uses Ada Lovelace, both on 28 nm vs 5 nm. Shader units: 960 (GeForce GTX 660) vs 5,888 (GeForce RTX 4070). Raw compute: 1.981 TFLOPS (GeForce GTX 660) vs 29.15 TFLOPS (GeForce RTX 4070). Boost clocks: 1033 MHz vs 2475 MHz.
| Feature | GeForce GTX 660 | GeForce RTX 4070 |
|---|---|---|
| G3D Mark Score | 4,040 | 26,919+566% |
| Architecture | Kepler | Ada Lovelace |
| Process Node | 28 nm | 5 nm |
| Shading Units | 960 | 5888+513% |
| Compute (TFLOPS) | 1.981 TFLOPS | 29.15 TFLOPS+1371% |
| Boost Clock | 1033 MHz | 2475 MHz+140% |
| ROPs | 24 | 64+167% |
| TMUs | 80 | 184+130% |
| L1 Cache | 0.08 MB | 5.8 MB+7150% |
| L2 Cache | 0.38 MB | 36 MB+9374% |
Advanced Features (DLSS/FSR)
The clearest feature edge for the GeForce RTX 4070 is support for DLSS 3.5 + Frame Generation. In games that support it, that can smooth out motion and lift perceived FPS. The GeForce GTX 660 does not have comparable native support in the same tier.The GeForce RTX 4070 supports the newer DLSS 3.5 Super Resolution, whereas the GeForce GTX 660 is capped at Upscaling support.
| Feature | GeForce GTX 660 | GeForce RTX 4070 |
|---|---|---|
| Upscaling Tech | Upscaling support | DLSS 3.5 Super Resolution |
| Frame Generation | Not Supported | DLSS 3.5 + Frame Generation |
| Ray Reconstruction | No | Yes (DLSS 3.5) |
| Low Latency | NVIDIA Reflex | NVIDIA Reflex |
Video Memory (VRAM)
The GeForce GTX 660 has 3 GB of VRAM, while the GeForce RTX 4070 carries 12 GB. GeForce RTX 4070 gives you 300% more memory capacity, which matters more once you move into heavier textures, mods, or higher resolutions. Memory bandwidth: 144.2 GB/s (GeForce GTX 660) vs 504 GB/s (GeForce RTX 4070) — a 249.5% advantage for the GeForce RTX 4070. Memory bus width is 192-bit on the GeForce GTX 660 and 192-bit on the GeForce RTX 4070. L2 Cache: 0.38 MB (GeForce GTX 660) vs 36 MB (GeForce RTX 4070) — the GeForce RTX 4070 has significantly larger on-die cache to reduce VRAM reliance.
| Feature | GeForce GTX 660 | GeForce RTX 4070 |
|---|---|---|
| VRAM Capacity | 3 GB | 12 GB+300% |
| Memory Type | GDDR5 | GDDR6X |
| Memory Bandwidth | 144.2 GB/s | 504 GB/s+250% |
| Bus Width | 192-bit | 192-bit |
| L2 Cache | 0.38 MB | 36 MB+9374% |
Display & API Support
DirectX support: 12 (GeForce GTX 660) vs 12.2 (GeForce RTX 4070). Maximum simultaneous displays: 4 vs 4.
| Feature | GeForce GTX 660 | GeForce RTX 4070 |
|---|---|---|
| DirectX | 12 | 12.2+2% |
| Max Displays | 4 | 4 |
Media & Encoding
Hardware encoder: NVENC 1st Gen (GeForce GTX 660) vs 8th Gen NVENC (2x) (GeForce RTX 4070). Decoder: NVDEC 1st Gen vs 5th Gen NVDEC. Supported codecs: H.264 (GeForce GTX 660) vs MPEG-2,H.264,HEVC,VP9,AV1 (GeForce RTX 4070).
| Feature | GeForce GTX 660 | GeForce RTX 4070 |
|---|---|---|
| Encoder | NVENC 1st Gen | 8th Gen NVENC (2x) |
| Decoder | NVDEC 1st Gen | 5th Gen NVDEC |
| Codecs | H.264 | MPEG-2,H.264,HEVC,VP9,AV1 |
Power & Dimensions
The GeForce GTX 660 draws 140W versus the GeForce RTX 4070's 200W — a 35.3% difference. The GeForce GTX 660 is more power-efficient. Recommended PSU: 450W (GeForce GTX 660) vs 650W (GeForce RTX 4070). Power connectors: 1x 6-pin vs 8-pin. Card length: 241mm vs 304mm, occupying 2 vs 3 slots.
| Feature | GeForce GTX 660 | GeForce RTX 4070 |
|---|---|---|
| TDP | 140W-30% | 200W |
| Recommended PSU | 450W-31% | 650W |
| Power Connector | 1x 6-pin | 8-pin |
| Length | 241mm | 304mm |
| Height | — | 137mm |
| Slots | 2-33% | 3 |
| Temp (Load) | — | 80°C |
| Perf/Watt | 28.9 | 134.6+366% |
Value Analysis
At launch, the GeForce GTX 660 came in at $229, while the GeForce RTX 4070 launched at $599. On MSRP, GeForce GTX 660 was 61.8% cheaper ($370 less). Performance per dollar on MSRP (G3D Mark / MSRP): 17.6 (GeForce GTX 660) vs 44.9 (GeForce RTX 4070) — the GeForce RTX 4070 offers 155.1% better value. The newer card here is GeForce RTX 4070 (2023 vs 2012).
| Feature | GeForce GTX 660 | GeForce RTX 4070 |
|---|---|---|
| MSRP | $229-62% | $599 |
| Performance per Dollar | 17.6 | 44.9+155% |
| Codename | GK106 | AD104 |
| Release | September 6 2012 | April 12 2023 |
| Ranking | #490 | #32 |
Affiliate Disclosure
ChipVERSUS is a participant in the Amazon Services LLC Associates Program, an affiliate advertising program designed to provide a means for sites to earn advertising fees by advertising and linking to Amazon.com. We may earn a commission on qualifying purchases made through our links. This comes at no additional cost to you and helps support our work in providing comprehensive PC building guides and tools.
Amazon and the Amazon logo are trademarks of Amazon.com, Inc. or its affiliates.
Top Performing GPUs
The most powerful gpus ranked by G3D Mark benchmark scores.














