RTX 3060 12GB
NVIDIAconsumer| Overview | |
|---|---|
| Architecture | Ampere (2021) |
| Compute capability | 8.6 |
| Memory | |
| VRAM | 12 GB GDDR6 |
| Memory bandwidth | 360 GB/s |
| Compute — vector | |
| FP32 | 12.8 TFLOPS |
| Cores & clocks | |
| Shader cores | 3,584 |
| Peak clock | 1,780 MHz |
| Board & system | |
| TDP | 170 W |
| PCIe | PCIe Gen4 |
| MSRP | $329 |
Compatible CUDA Toolkit versions
PyTorch on the RTX 3060 12GB
Check against your driver and Python version →
The newest PyTorch, 2.14.0, runs on this card with a plain pip install torch.
| PyTorch | CUDA builds that run on this card | Plain pip install torch |
|---|---|---|
| 2.14.0 | 12.6, 13.0, 13.2 | works |
| 2.13.0 | 12.6, 12.9, 13.0, 13.2 | works |
| 2.12.0 | 12.6, 13.0, 13.2 | works |
| 2.11.0 | 12.6, 12.8, 12.9, 13.0 | works |
| 2.10.0 | 12.6, 12.8, 12.9, 13.0 | works |
| 2.9.0 | 12.6, 12.8, 12.9, 13.0 | works |
| 2.8.0 | 12.6, 12.8, 12.9 | works |
| 2.7.0 | 11.8, 12.6, 12.8 | works |
| 2.6.0 | 11.8, 12.4, 12.6 | works |
| 2.5.0 | 11.8, 12.1, 12.4 | works |
| 2.4.0 | 11.8, 12.1, 12.4 | works |
| 2.3.0 | 11.8, 12.1 | works |
| 2.2.0 | 11.8, 12.1 | works |
| 2.1.0 | 11.8, 12.1 | works |
| 2.0.0 | 11.7, 11.8 | works |
| 1.13.0 | 11.6, 11.7 | works |
| 1.12.0 | 11.3, 11.6 | — |
| 1.11.0 | 11.3 | — |
From each build's compiled architecture list (Linux x86_64 wheels). Your driver must also support the CUDA version — see compatible CUDA versions.
Libraries on the RTX 3060 12GB
Whether each library's own requirements cover this card. Each row links to the version and source it was checked against on library support.
| Library | Supported | Details |
|---|---|---|
| vLLM 0.30.0 | Yes | |
| FlashAttention-2 2.8.3.post1 | Yes | |
| FlashAttention-3 main (2026-09-27) | No | no build for compute capability 8.6 |
| bitsandbytes 0.50.2 | Yes | |
| TensorRT 11.3.0 | Yes | |
| llama.cpp master (2026-09-27) | Yes |
Native low-precision formats
Whether the card's matrix hardware runs these formats natively. Without native support a format can still work, but through slower emulation or conversion, not at the card's rated speed.
| BF16 | FP8 | FP4 |
|---|---|---|
| Native | No | No |
Source: NVIDIA TensorRT 11.3.0 support matrix (hardware precision table).
Added 2026-09-27 as one of the mid-range cards local-LLM users actually buy (12 GB for $329 at launch). Sources, all NVIDIA: 3,584 CUDA cores, 1.78 GHz boost, 12 GB GDDR6 and 170 W Graphics Card Power from the RTX 3060 family product page (nvidia.com/en-us/geforce/graphics-cards/30-series/ rtx-3060-3060ti/); 360 GB/s from NVIDIA's own launch Q&A ("GeForce RTX 3060 is 192-bit and GDDR6 (at 15 Gbps), delivering 360GB/s of bandwidth", nvidia.com/en-us/geforce/news/game-on-you-asked-we-answered-qa/); $329 and the late-February 2021 availability from NVIDIA's announcement article (nvidia.com/en-us/geforce/news/geforce-rtx-3060/). compute capability 8.6 from NVIDIA's GeForce compare page (CUDA Capability row, RTX 30 Series). fp32TFLOPS is derived, not published: 3,584 cores × 2 FLOPs × 1.78 GHz = 12.76. NVIDIA publishes no Tensor-core throughput for this card on its product page, so those fields are omitted rather than filled from secondary sources. The 8 GB RTX 3060 is a different, narrower-bus card and is not this entry.