NVIDIA GeForce GTX 460 v2 vs NVIDIA Tesla C2070 Comparison

FAQ

Q: Which GPU has the higher OpenCL benchmark score?

A: The NVIDIA Tesla C2070 scores 9716 in Geekbench OpenCL, which is 11.1% higher than the GeForce GTX 460 v2's 8743. The Tesla C2070 wins the head-to-head comparison by one benchmark.

Q: How do the two cards rank against their closest rivals?

A: The Tesla C2070 sits 0.1% behind the Tesla M10 and 0.5% ahead of the Quadro P4000 in average score, with a Geekbench OpenCL score of 9716 placing it at the 47th percentile of all GPUs. The GTX 460 v2 scores 8743, placing it at the 44th percentile, 0.7% ahead of the Quadro P2200 and 1.2% behind the Radeon Pro WX 5100.

Q: What are the memory capacities of each card?

A: The Tesla C2070 ships with 6 GB of GDDR5 memory on a 384-bit bus with 143.4 GB/s bandwidth. The GTX 460 v2 has 1024 MB of GDDR5 on a 192-bit bus with 96.96 GB/s bandwidth (the recorded memory clock is 1002 MHz, 4 Gbps effective on the GTX 460 v2, 3 Gbps effective on the Tesla C2070).

Q: What is the transistor count difference between the two chips?

A: The Tesla C2070 uses the GF100 chip with 3,100 million transistors on TSMC 40 nm process with 5.9M transistors per square mm, for a die size of 529 square mm. The GTX 460 v2 uses GF114 with 1,950 million transistors on TSMC 40 nm, also 5.9M per square mm, and a 332 square mm die.

Q: What is the power draw of each card?

A: The Tesla C2070 has a TDP of 238 W with a suggested power supply of 550 W and connectors of 1x 6-pin plus 1x 8-pin. The GTX 460 v2 has a TDP of 160 W with a suggested 450 W PSU and two 6-pin connectors.

Q: Which card was released later?

A: The GTX 460 v2 was released on September 23, 2011, while the Tesla C2070 was released on July 24, 2011. Both are now end-of-life products.

Architecture Differences

The Tesla C2070 is built on the GF100 chip, which is the Fermi architecture, while the GTX 460 v2 uses GF114, which is the Fermi 2.0 architecture. Both are fabricated on TSMC's 40 nm process with identical transistor density of 5.9M per square mm, but the die size differs substantially: the GF100 measures 529 square mm versus 332 square mm for GF100.

The Tesla C2070 packs 3,100 million transistors, the GTX 460 v2 has 1,950 million. That size difference fuels the Tesla's shading unit count: 448 shading units on the Tesla versus 336 on the GTX 460 v2. The Tesla also has more ROPs, 48 versus 24, while both have 56 texture mapping units. The GTX 460 v2's memory controller is narrower: 192 bit versus 384 bit, and the memory clock is higher on the GTX 460 v2 at 1002 MHz (4 Gbps effective) versus 747 MHz (3 Gbps effective) on the Tesla, though the Tesla's wider bus gives it more total bandwidth. Both share the same API support: DirectX 12 (11_0), OpenGL 4.6, and no Vulkan. The Tesla is from the Tesla Fermi (x20xx) generation while the GTX 460 v2 is from the GeForce 400 generation.

Head-to-Head Benchmarks

The single recorded head-to-head benchmark is Geekbench OpenCL. The Tesla C2070 scores 9716 against the GTX 460 v2's 8743, a delta of 11.1% in favor of the Tesla. This matches the Tesla's wider 384-bit memory interface and higher ROP count, which help with compute workloads that benefit from memory bandwidth. However, the GTX 460 v2 is not far behind in raw FP32 throughput: the Tesla delivers 1,027.7 GFLOPS while the GTX 460 v2 delivers 1,046.3 GFLOPS, slightly higher. The GTX 460 v2 also has a higher texture rate, 43.62 GTexel/s versus 32.14 GTexel/s on the Tesla, which suggests the GTX 460 v2 is more balanced per clock despite a smaller memory bus. In pixel rate, the Tesla leads at 16.07 GPixel/s versus 10.91 GPixel/s for the GTX 460 v2. The benchmark result confirms the Tesla's advantage in a compute-oriented test, but the GTX 460 v2's higher FP32 and texture rates indicate the gap is not universal.

Specification Differences

| Specification | Tesla C2070 | GTX 460 v2 |

|--------------|-------------|------------|

| Architecture | Fermi | Fermi 2.0 |

| Chip | GF100 | GF114 |

| Transistors | 3,100 million | 1,950 million |

| Die Size | 529 mm² | 332 mm² |

| Memory Clock | 747 MHz (3 Gbps) | 1002 MHz (4 Gbps) |

| Memory Size | 6 GB | 1024 MB |

| Memory Bus | 384 bit | 192 bit |

| Memory Bandwidth | 143.4 GB/s | 96.19 GB/s |

| Shading Units | 448 | 336 |

| ROPs | 48 | 24 |

| Pixel Rate | 16.07 GPixel/s | 10.91 GPixel/s |

| Texture Rate | 32.14 GTexel/s | 43.62 GTexel/s |

| FP32 | 1,027.7 GFLOPS | 1,046.3 GFLOPS |

| TDP | 238 W | 160 W |

| Power Connectors | 1x 6-pin + 1x 8-pin | 2x 6-pin |

| Suggested PSU | 550 W | 450 W |

| Display Outputs | 1x DVI | 2x DVI, 1x mini-HDMI 1.3a |

| Release Date | July 24, 2011 | September 23, 2011 |

| Length | 248 mm (9.8 inches) | 210 mm (8.3 inches) |

| Launch MSRP | None | 199 USD |

| Predecessor | Tesla | GeForce 200 |

| Successor | Tesla Kepler | GeForce 500 |

Where Each One Wins

The Tesla C2070 wins in compute throughput: its 48 ROPs, 384-bit memory bus, and 143.4 GB/s of bandwidth give it a clear edge in the Geekbench OpenCL test. Its 16.07 GPixel/s pixel rate is 47% higher than the GTX 460 v2's 10.91 GPixel/s. This makes it the stronger choice for workloads that stress memory bandwidth and pixel processing, such as compute-heavy tasks or rendering pipelines that rely on ROP throughput. Its 6 GB of memory is also far larger, which helps when the working set exceeds the 1 GB frame buffer of the GTX 460 v2.

The GTX 460 v2 counters with a higher FP32 rating, 1,046.3 GFLOPS versus 1,027.7 GFLOPS, and a higher texture rate, 43.62 GTexel/s versus 32.14 GTexel/s. It also draws 78 W less power: 160 W versus 238 W TDP, and has a lower suggested PSU of 450 W versus 550 W. Its higher memory clock, 1002 MHz (4 Gbps effective), partially compensates for the narrower 192-bit bus. The GTX 460 v2 also has dual DVI plus mini-HDMI outputs against the Tesla's single DVI. The GTX 460 v2's 210 mm length is shorter than the Tesla's 248 mm, making it easier to fit in compact cases. For users who prioritize texture throughput, FP32 density, or lower power consumption, the GTX 460 v2 has the advantage.

The Verdict

The data points to the Tesla C2070 as the overall compute winner in this pairing. It beats the GTX 460 v2 by 11.1% in Geekbench OpenCL, and its 6 GB memory capacity, 384-bit interface, and 143.4 GB/s bandwidth are substantial advantages for memory-bound workloads. The Tesla also has a higher pixel rate and more ROPs, so tasks that push pixel fill rate will favor it. Its 47th percentile ranking versus the GTX 460 v2's 44th percentile confirms the Tesla sits slightly higher in the overall GPU distribution.

The GTX 460 v2 is the better option where power efficiency and texture throughput matter most. It has a higher texture rate, higher FP32, and lower TDP, with a smaller PSU requirement. Its 160 W TDP and 450 W suggested PSU make it easier to integrate into systems with modest power budgets. The GTX 460 v2 also supports more display outputs, which is useful for multi-monitor setups. The GTX 460 v2's launch MSRP was 199 USD, while the Tesla C2070 has no recorded launch MSRP.

For compute-centric users who need large memory and high bandwidth, the Tesla C2070 is the clear pick from the benchmark data. For users who want better texture performance, lower power draw, and more display connectivity, the GTX 460 v2 is the more balanced alternative. The recorded head-to-head result gives the Tesla one win, but the GTX 460 v2's higher FP32 and texture rates show that the Tesla's victory is not absolute across all workloads.