AMD Radeon Instinct MI60 vs AMD Radeon PRO V710 Comparison

Head-to-Head Benchmarks

The only directly comparable benchmark recorded for both cards is Geekbench OpenCL, and the result is decisive. The AMD Radeon PRO V710 scores 116460, while the AMD Radeon Instinct MI60 scores 92488. That is a 20.6% advantage for the PRO V710 in this compute-oriented test, a substantial margin that places the newer card clearly ahead in raw OpenCL throughput.

Looking at the broader database context, the MI60's average benchmark score of 92466 puts it in the 93rd percentile of all GPUs. Its nearest rivals are the NVIDIA RTX A4500 at 91671 (0.9% behind the MI60) and the NVIDIA RTX A4500 Mobile at 91134 (1.5% behind). The MI60 also sits ahead of the AMD Radeon Pro VII, which scores 97131, meaning the MI60 trails that card by 4.8%. Against the AMD Radeon RX 7900M at 97487, the MI60 is 5.2% behind. So while the MI60 is a strong performer in its own right, it is not the top of its immediate peer group.

The PRO V710's average benchmark score is 58657, which lands it in the 88th percentile. That average is dragged down by the inclusion of a 3DMark Steel Nomad DX12 result of 853, a test that the MI60 does not have a recorded score for. When comparing the two cards on the shared Geekbench OpenCL test, the PRO V710's 116460 is the clear winner. Its nearest rivals in the database include the NVIDIA P102-100 at 58528 (0.2% behind), the AMD Radeon RX 6950 XT at 58392 (0.5% behind), the Intel Arc A570M at 58239 (0.7% behind), and the AMD Radeon RX 5600 OEM at 58085 (1% behind). These deltas are tiny, meaning the PRO V710's average score cluster is extremely tight around 58.6k, but its OpenCL score is far above that cluster.

The MI60 wins no head-to-head benchmarks in the recorded data. The PRO V710 takes the single shared test. That does not make the MI60 a poor card; it simply means that in the one workload where both have been measured, the newer architecture is faster by a significant margin. The 20.6% delta is not a marginal difference; it is a clear generational step in compute performance.

Architecture Differences

The two cards come from different architectural eras. The MI60 uses the Vega 20 chip built on GCN 5.1, fabricated on TSMC's 7 nm process. It packs 13,230 million transistors into a 331 mm² die, giving a transistor density of 40.0M per mm². The PRO V710 uses the Navi 32 chip on RDNA 3.0, also from TSMC but on a 5 nm node. It contains 28,100 million transistors across a 346 mm² die, for a density of 81.2M per mm². The PRO V710 more than doubles the transistor count on a slightly larger die, and its density is roughly double that of the MI60.

Memory configurations differ sharply. The MI60 has 32 GB of HBM2 on a 4096-bit bus, delivering 1.02 TB/s of bandwidth. The PRO V710 has 28 GB of GDDR6 on a 224-bit bus, delivering 504.0 GB/s. The MI60's memory bandwidth is more than double that of the PRO V710, which is typical for HBM versus GDDR6. However, the PRO V710's memory clock is listed as 2250 MHz with 18 Gbps effective, while the MI60's memory runs at 1000 MHz with 2 Gbps effective. The bus width difference explains the bandwidth gap.

Compute resources also differ. The MI60 has 4096 shading units, 256 TMUs, and 64 ROPs. The PRO V710 has 3456 shading units, 216 TMUs, and 96 ROPs. The MI60 has more shaders and TMUs, but the PRO V710 has more ROPs. The PRO V710 also includes 54 ray tracing cores, while the MI60 has none listed. Neither card has tensor cores recorded.

Clock speeds favor the PRO V710. Its base clock is 1900 MHz and boost is 2000 MHz, versus the MI60's 1200 MHz base and 1800 MHz boost. Despite the MI60's higher shader count, the PRO V710's FP32 throughput is 27.65 TFLOPS versus 14.75 TFLOPS for the MI60. That is nearly double the FP32 compute. FP16 tells a different story: the MI60 reaches 29.49 TFLOPS with a 2:1 ratio, while the PRO V710 does 27.65 TFLOPS at 1:1. The MI60's FP16 advantage is modest but real.

Pixel and texture rates also split. The PRO V710 has a pixel rate of 192.0 GPixel/s versus 115.2 GPixel/s for the MI60. Texture rate goes the other way: 460.8 GTexel/s for the MI60 versus 432.0 GTexel/s for the PRO V710.

Power and physical design are major differentiators. The MI60 is rated at 300 W TDP, requires a 1x 6-pin plus 1x 8-pin power connector, and a 700 W suggested PSU. It is dual-slot and 267 mm long. The PRO V710 is rated at 158 W TDP, needs only a single 8-pin connector, and a 450 W suggested PSU. It is single-slot, though its dimensions are not recorded. The PRO V710 consumes nearly half the power of the MI60 while delivering higher FP32 and OpenCL performance. That efficiency gap is the most striking architectural takeaway.

API support also differs. The MI60 supports DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.3. The PRO V710 supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The PRO V710 is ahead on both DirectX and Vulkan versions. The MI60 has a single mini-DisplayPort 1.4a output; the PRO V710 has no display outputs at all, making it a pure compute accelerator.

The Verdict

The data points to a clear choice for most buyers. The PRO V710 wins the only shared benchmark by 20.6%, delivers nearly double the FP32 throughput, uses less than half the power, and comes from a newer architecture with ray tracing support and newer API versions. If the workload is OpenCL compute, the PRO V710 is the better card, period.

The MI60 still has reasons to exist. Its 32 GB of HBM2 memory with 1.02 TB/s of bandwidth is unmatched by the PRO V710. Applications that are memory-bandwidth bound, not compute bound, could prefer the MI60. Its FP16 throughput of 29.49 TFLOPS also edges out the PRO V710's 27.65 TFLOPS. But those advantages come at a cost in power draw and physical size, and they do not translate into a win in the recorded OpenCL benchmark.

For a new deployment, the PRO V710 is the sensible pick. It is single-slot, low power, and faster in the measured compute test. For an existing system built around the MI60, the upgrade path is clear if compute workloads dominate. The MI60 is end-of-life, while the PRO V710's production status is not recorded, but its 2024 release date makes it the current-generation option.

FAQ

Q: Which card is faster in Geekbench OpenCL?

A: The AMD Radeon PRO V710 scores 116460 versus 92488 for the MI60, a 20.6% advantage.

Q: Does the MI60 have more memory bandwidth?

A: Yes. The MI60 has 32 GB of HBM2 on a 4096-bit bus with 1.02 TB/s bandwidth. The PRO V710 has 28 GB of GDDR6 on a 224-bit bus with 504.0 GB/s.

Q: Which card supports ray tracing?

A: The PRO V710 has 54 ray tracing cores. The MI60 has no ray tracing cores listed.

Q: What are the power requirements?

A: The MI60 is rated at 300 W with a 700 W suggested PSU and needs 1x 6-pin plus 1x 8-pin connectors. The PRO V710 is rated at 158 W with a 450 W suggested PSU and needs a single 8-pin connector.

Q: Which card has higher FP32 compute?

A: The PRO V710 delivers 27.65 TFLOPS FP32 versus 14.75 TFLOPS for the MI60.

Q: Does either card have display outputs?

A: The MI60 has one mini-DisplayPort 1.4a output. The PRO V710 has no display outputs.

Where Each One Wins

The PRO V710 wins in raw compute. It takes the Geekbench OpenCL test by 20.6%, has nearly double the FP32 throughput, and offers higher pixel rate at 192.0 GPixel/s versus 115.2 GPixel/s. It also supports DirectX 12 Ultimate and Vulkan 1.4, making it the better option for modern compute and graphics workloads that use those APIs. Its 158 W TDP and single-slot design make it far easier to integrate into dense servers.

The MI60 wins in memory-centric scenarios. Its 1.02 TB/s bandwidth is more than double the PRO V710's 504.0 GB/s, and its 32 GB capacity exceeds the PRO V710's 28 GB. For workloads that stream large datasets through memory, the MI60's HBM2 setup is the stronger asset. Its FP16 throughput of 29.49 TFLOPS also beats the PRO V710's 27.65 TFLOPS, which matters for mixed-precision workflows. The MI60's texture rate of 460.8 GTexel/s is also higher than 432.0 GTexel/s.

The PRO V710 wins on efficiency by a wide margin. It delivers higher OpenCL scores and FP32 compute while drawing 158 W versus 300 W. The MI60 is end-of-life, so availability and long-term support favor the PRO V710. The MI60's only clear advantages are memory bandwidth, FP16 throughput, and texture rate. For every other measured metric, the PRO V710 is ahead.

Specification Differences

| Component | AMD Radeon Instinct MI60 | AMD Radeon PRO V710 |

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

| Chip | Vega 20 | Navi 32 |

| Architecture | GCN 5.1 | RDNA 3.0 |

| Process node | 7 nm | 5 nm |

| Transistors | 13,230 million | 28,100 million |

| Die size | 331 mm² | 346 mm² |

| Transistor density | 40.0M / mm² | 81.2M / mm² |

| Base clock | 1200 MHz | 1900 MHz |

| Boost clock | 1800 MHz | 2000 MHz |

| Memory size | 32 GB | 28 GB |

| Memory type | HBM2 | GDDR6 |

| Memory bus | 4096 bit | 224 bit |

| Memory bandwidth | 1.02 TB/s | 504.0 GB/s |

| Shading units | 4096 | 3456 |

| TMUs | 256 | 216 |

| ROPs | 64 | 96 |

| Ray tracing cores | None listed | 54 |

| Pixel rate | 115.2 GPixel/s | 192.0 GPixel/s |

| Texture rate | 460.8 GTexel/s | 432.0 GTexel/s |

| FP32 | 14.75 TFLOPS | 27.65 TFLOPS |

| FP16 | 29.49 TFLOPS (2:1) | 27.65 TFLOPS (1:1) |

| TDP | 300 W | 158 W |

| Slot width | Dual-slot | Single-slot |

| Power connectors | 1x 6-pin + 1x 8-pin | 1x 8-pin |

| Suggested PSU | 700 W | 450 W |

| Display outputs | 1x mini-DisplayPort 1.4a | No outputs |

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

| OpenGL | 4.6 | 4.6 |

| Vulkan | 1.3 | 1.4 |

| Production status | End-of-life | Not recorded |

| Release date | 2018-11-17 | 2024-10-02 |