GPU Comparison

# NVIDIA Tesla P40 vs Intel Arc Pro A60

The NVIDIA Tesla P40 and Intel Arc Pro A60 are professional workstation cards from different eras, with the P40 built for compute density and the A60 designed for modern graphics workloads. The benchmark data shows a split decision: the P40 wins the Vulkan test by 19.3%, while the A60 edges out the P40 in OpenCL by 2.3%. Overall, the P40 holds a higher average benchmark score of 65,095 compared to the A60's 60,326, placing them at the 89th and 88th percentiles of all GPUs respectively.

Where Each One Wins

The NVIDIA Tesla P40 dominates in Vulkan performance. Its Geekbench Vulkan score of 68,172 stands 19.3% above the Intel Arc Pro A60's 57,166. This is the largest performance gap in either direction between the two cards. The P40's Vulkan advantage is substantial enough that any workload leveraging Vulkan compute or graphics will see a meaningful benefit from choosing the older NVIDIA card. Interestingly, the P40's Vulkan score is also higher than its own OpenCL score of 62,017, suggesting the Pascal architecture responds particularly well to Vulkan's low-level API.

The Intel Arc Pro A60 wins the OpenCL test, scoring 63,485 against the P40's 62,017, a 2.3% margin. This is a modest but consistent advantage. The A60's OpenCL performance also exceeds its own Vulkan score of 57,166, which is unusual and indicates that Intel's Xe-HPG architecture has optimized OpenCL paths that outperform its Vulkan implementation in this benchmark. For users running OpenCL-based compute tasks, the A60 provides slightly better raw throughput.

Beyond the raw benchmark scores, the A60 offers features the P40 lacks entirely. It has 16 ray tracing cores, whereas the P40 has none. The A60 also supports DirectX 12 Ultimate (12_2), while the P40 is limited to DirectX 12 (12_1). Display outputs are another differentiator, the A60 has 4x DisplayPort 2.0 connectors, while the P40 has no display outputs at all. For any workload involving ray tracing or direct display output, the A60 is the only viable option between the two.

Architecture Differences

The architectural gap between these two cards is generational. The Tesla P40 uses NVIDIA's GP102 chip built on the Pascal architecture, manufactured on TSMC's 16 nm process. It packs 11,800 million transistors into a 471 mm² die, yielding a transistor density of 25.1 million transistors per mm². The Intel Arc Pro A60 uses the DG2-256 chip with Xe-HPG architecture, built on a much more advanced 6 nm process from TSMC. It contains 11,500 million transistors in a smaller 269 mm² die, achieving 42.8 million transistors per mm², a density advantage of over 70%.

Despite having slightly fewer transistors, the A60's newer process allows for dramatically higher boost clocks. The A60 boosts to 2050 MHz compared to the P40's 1531 MHz, though its base clock is lower at 900 MHz versus 1303 MHz. The A60 also has a much higher memory clock at 2000 MHz (16 Gbps effective) compared to the P40's 1808 MHz (7.2 Gbps effective).

Memory configurations differ significantly. The P40 offers 24 GB of GDDR5 on a 384-bit bus, delivering 347.1 GB/s of bandwidth. The A60 has 12 GB of GDDR6 on a 192-bit bus, but achieves slightly higher bandwidth at 384.0 GB/s due to the faster memory technology. The P40's advantage is capacity, double the VRAM, while the A60 wins on raw bandwidth.

The compute units tell a story of specialization. The P40 has 3,840 shading units, 240 TMUs, and 96 ROPs. The A60 has 2,048 shading units, 128 TMUs, and 64 ROPs. The P40 also has no ray tracing cores, while the A60 includes 16. The P40's FP32 throughput is 11.76 TFLOPS versus the A60's 8.397 TFLOPS, but the A60 offers 16.79 TFLOPS of FP16 performance at a 2:1 ratio, while the P40's FP16 is a negligible 183.7 GFLOPS at a 1:64 ratio. The A60's FP16 capability is a massive 91x advantage in half-precision compute.

The Verdict

The data presents a clear choice based on workload priorities. For Vulkan-based applications and raw FP32 compute throughput, the NVIDIA Tesla P40 is the stronger card. Its 19.3% Vulkan lead and 11.76 TFLOPS FP32 performance make it suitable for compute-heavy tasks where double the VRAM (24 GB vs 12 GB) is also valuable. The P40's 89th percentile ranking versus the A60's 88th, combined with its higher average benchmark score of 65,095, confirms its overall performance edge.

However, the Intel Arc Pro A60 is the more modern and versatile card. It wins OpenCL by 2.3%, offers ray tracing cores, supports DirectX 12 Ultimate, and includes display outputs. Its 130 W TDP is nearly half the P40's 250 W, and it requires only a 300 W suggested PSU versus 600 W for the P40. The A60 is also single-slot versus the P40's dual-slot design. For users needing a workstation card that can handle modern graphics APIs, ray tracing, or direct display connectivity, the A60 is the obvious choice despite its lower overall benchmark scores.

If the workload is compute-centric and Vulkan-based, the Tesla P40's performance advantage and larger memory pool make it the pick. If the workload involves OpenCL, modern graphics features, or power efficiency, the Arc Pro A60 wins.

FAQ

Q: Which card has the higher average benchmark score?

A: The NVIDIA Tesla P40 has an average benchmark score of 65,095, compared to the Intel Arc Pro A60's 60,326. The P40 also ranks higher at the 89th percentile versus the A60's 88th.

Q: How much faster is the Tesla P40 in Vulkan?

A: The Tesla P40 scores 68,172 in Geekbench Vulkan, which is 19.3% higher than the Arc Pro A60's score of 57,166. This is the largest performance gap between the two cards in any test.

Q: Does the Arc Pro A60 support ray tracing?

A: Yes, the Intel Arc Pro A60 includes 16 ray tracing cores and supports DirectX 12 Ultimate (12_2). The NVIDIA Tesla P40 has no ray tracing cores and only supports DirectX 12 (12_1).

Q: What is the memory capacity difference?

A: The Tesla P40 has 24 GB of GDDR5 memory, while the Arc Pro A60 has 12 GB of GDDR6. Despite having half the capacity, the A60 achieves higher bandwidth at 384.0 GB/s versus the P40's 347.1 GB/s.

Q: Which card has display outputs?

A: The Intel Arc Pro A60 has 4x DisplayPort 2.0 outputs. The NVIDIA Tesla P40 has no display outputs at all, making it unsuitable for direct display connection.

Q: How do the power requirements compare?

A: The Tesla P40 has a TDP of 250 W and suggests a 600 W PSU, while the Arc Pro A60 has a TDP of 130 W and suggests a 300 W PSU. The A60 is also single-slot compared to the P40's dual-slot design.

Head-to-Head Benchmarks

The Geekbench OpenCL test shows the Intel Arc Pro A60 winning with a score of 63,485 against the Tesla P40's 62,017. This 2.3% margin is relatively narrow, suggesting that in OpenCL workloads, the two cards perform at a similar level. The A60's advantage here may stem from its higher memory bandwidth (384.0 GB/s vs 347.1 GB/s) and its faster effective memory clock of 16 Gbps versus 7.2 Gbps. Interestingly, the A60's OpenCL score is significantly higher than its Vulkan score, indicating that Intel's OpenCL drivers are more mature or better optimized than their Vulkan path.

The Geekbench Vulkan test tells a very different story. The Tesla P40 scores 68,172, which is 19.3% higher than the Arc Pro A60's 57,166. This is a substantial victory for NVIDIA. The P40's Vulkan score is also higher than its OpenCL score, showing that Pascal's Vulkan implementation is particularly strong. The 19.3% delta is the single largest performance difference between the two cards in any metric, and it dwarfs the A60's OpenCL win. For Vulkan-based applications, the P40 is categorically superior.

Looking at the nearest rival data adds context. The Tesla P40's closest rival is the AMD Radeon VII, which scores 66,004 and is 1.4% ahead of the P40. The AMD Radeon Pro WX 9100 trails the P40 by 1.4% with a score of 64,212. The NVIDIA CMP 30HX and AMD Radeon RX 9060 XT LP both sit 2% behind at 63,842 and 63,830 respectively. The P40's position among these rivals shows it is competitive with contemporary high-end cards.

The Arc Pro A60's nearest rival list is unusual. The NVIDIA GeForce RTX 4090 scores 60,347, essentially tied with the A60's 60,326 at a 0% delta. The AMD Radeon Pro Vega 48 is 0.3% behind, and the AMD Radeon Pro W6600M is 2.5% ahead. The AMD Radeon PRO V710 trails by 2.8%. The A60's average score places it in the same performance tier as these very different cards, showing that its overall compute capability is competitive despite its modest specifications on paper.

Specification Differences

| Specification | NVIDIA Tesla P40 | Intel Arc Pro A60 |

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

| Architecture | Pascal | Xe-HPG |

| Process Node | 16 nm | 6 nm |

| Transistors | 11,800 million | 11,500 million |

| Die Size | 471 mm² | 269 mm² |

| Transistor Density | 25.1M / mm² | 42.8M / mm² |

| Base Clock | 1303 MHz | 900 MHz |

| Boost Clock | 1531 MHz | 2050 MHz |

| Memory Clock | 1808 MHz (7.2 Gbps) | 2000 MHz (16 Gbps) |

| Memory Size | 24 GB | 12 GB |

| Memory Type | GDDR5 | GDDR6 |

| Memory Bus | 384 bit | 192 bit |

| Memory Bandwidth | 347.1 GB/s | 384.0 GB/s |

| Shading Units | 3840 | 2048 |

| TMUs | 240 | 128 |

| ROPs | 96 | 64 |

| Ray Tracing Cores | 0 | 16 |

| Pixel Rate | 147.0 GPixel/s | 131.2 GPixel/s |

| Texture Rate | 367.4 GTexel/s | 262.4 GTexel/s |

| FP32 | 11.76 TFLOPS | 8.397 TFLOPS |

| FP16 | 183.7 GFLOPS (1:64) | 16.79 TFLOPS (2:1) |

| TDP | 250 W | 130 W |

| Slot Width | Dual-slot | Single-slot |

| Suggested PSU | 600 W | 300 W |

| Bus Interface | PCIe 3.0 x16 | PCIe 4.0 x16 |

| Display Outputs | None | 4x DisplayPort 2.0 |

| DirectX | 12 (12_1) | 12 Ultimate (12_2) |

| Release Date | 2016-09-12 | 2023-06-05 |

| Production Status | End-of-life | Active |