NVIDIA Quadro GV100 vs NVIDIA Tesla M60 Comparison

# FAQ

Q: What is the average benchmark score for each GPU in this comparison?

A: The NVIDIA Quadro GV100 records an average benchmark score of 35,520, while the NVIDIA Tesla M60 sits at 30,490. The Quadro GV100 also places in the 80th percentile among all GPUs, compared to the Tesla M60's 75th percentile.

Q: Which GPU has the higher memory capacity and bandwidth?

A: The Quadro GV100 features 32 GB of HBM2 memory with a 4096-bit bus and 868.4 GB/s bandwidth. The Tesla M60 offers 8 GB of GDDR5 on a 256-bit bus with 160.4 GB/s bandwidth.

Q: How do the two cards compare in Geekbench OpenCL performance?

A: The Quadro GV100 scores 150,004 in Geekbench OpenCL, which is 408.4% higher than the Tesla M60's 29,506.

Q: What are the respective transistor counts and die sizes?

A: The Quadro GV100 packs 21,100 million transistors on an 815 mm² die, while the Tesla M60 has 5,200 million transistors on a 398 mm² die.

Q: Does the Tesla M60 support any display outputs?

A: No. The Tesla M60 has no display outputs, while the Quadro GV100 provides 4x DisplayPort 1.4a connections.

Q: Which GPU has a higher FP32 (single-precision) throughput?

A: The Quadro GV100 delivers 16.66 TFLOPS FP32, versus the Tesla M60's 4.825 TFLOPS. The GV100 also supports FP16 at 33.32 TFLOPS (2:1), while the Tesla M60 has no listed FP16 capability.

# Where Each One Wins

The data clearly separates these two NVIDIA workstation and datacenter cards across every recorded benchmark. The Quadro GV100 wins both head-to-head tests, and the margin is substantial.

In Geekbench OpenCL, the Quadro GV100's score of 150,004 dwarfs the Tesla M60's 29,506. This is a compute-oriented workload, and the GV100's advantage reflects its newer architecture and far larger resource pool. The Tesla M60, by contrast, was designed for virtualization and cloud workloads where raw compute density per watt matters more than peak performance.

When looking at Geekbench Vulkan, the pattern repeats: the Quadro GV100 scores 139,526 against the Tesla M60's 31,473. Vulkan is a low-level graphics API, and the GV100's 5120 shading units and 320 texture mapping units give it a decisive edge over the M60's 2048 shading units and 128 TMUs.

The Tesla M60 does not win any of the recorded benchmarks. It does, however, have a lower 300 W TDP compared to the GV100's 250 W? No, that is incorrect: the GV100 has a 250 W TDP and the Tesla M60 has a 300 W TDP. The M60 also requires a 700 W suggested PSU versus the GV100's 600 W, so the GV100 is more power-efficient in terms of performance per watt based on the FP32 and benchmark data.

For use cases, the Quadro GV100 is the choice when single-GPU compute performance is paramount, such as deep learning training, scientific simulation, or high-end rendering. The Tesla M60, with its 8 GB capacity and no display outputs, is suited for server-side virtual desktop infrastructure or cloud gaming, where multiple users share the GPU and display output is handled elsewhere.

# Architecture Differences

The Quadro GV100 is built on the Volta architecture, manufactured by TSMC on a 12 nm process. It uses the GV100 chip, which contains 21,100 million transistors on an 815 mm² die, yielding a transistor density of 25.9 million transistors per mm². The GV100 includes 640 tensor cores, a feature the Tesla M60 completely lacks.

The Tesla M60 uses the Maxwell 2.0 architecture, also from TSMC but on a 28 nm process. Its GM204 chip has 5,200 million transistors on a 398 mm² die, giving a transistor density of 13.1 million per mm². Maxwell 2.0 does not include tensor cores, and the M60 has no FP16 support listed.

Memory technology differs as well. The GV100 uses HBM2 with a 4096-bit memory bus, while the M60 uses GDDR5 on a 256-bit bus. The GV100's memory clock runs at 848 MHz with an effective 1696 Mbps rate, whereas the M60's memory clocks at 1253 MHz with 5 Gbps effective. The bandwidth gap is enormous: 868.4 GB/s versus 160.4 GB/s.

In terms of rendering hardware, the GV100 has 5120 shading units, 320 TMUs, and 128 ROPs. The M60 has 2048 shading units, 128 TMUs, and 64 ROPs. Neither card has dedicated ray tracing cores. The GV100's pixel rate is 208.3 GPixel/s and its texture rate is 520.6 GTexel/s, while the M60 achieves 75.39 GPixel/s and 150.8 GTexel/s.

Both cards support DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.4, so API compatibility is identical. The GV100 offers display outputs, while the M60 has none.

# Specification Differences

| Specification | NVIDIA Quadro GV100 | NVIDIA Tesla M60 |

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

| Architecture | Volta | Maxwell 2.0 |

| Process node | 12 nm | 28 nm |

| Transistors | 21,100 million | 5,200 million |

| Die size | 815 mm² | 398 mm² |

| Transistor density | 25.9M / mm² | 13.1M / mm² |

| Base clock | 1132 MHz | 557 MHz |

| Boost clock | 1627 MHz | 1178 MHz |

| Memory size | 32 GB | 8 GB |

| Memory type | HBM2 | GDDR5 |

| Memory bus | 4096 bit | 256 bit |

| Memory bandwidth | 868.4 GB/s | 160.4 GB/s |

| Shading units | 5120 | 2048 |

| TMUs | 320 | 128 |

| ROPs | 128 | 64 |

| Tensor cores | 640 | None |

| FP32 | 16.66 TFLOPS | 4.825 TFLOPS |

| FP16 | 33.32 TFLOPS (2:1) | Not listed |

| Pixel rate | 208.3 GPixel/s | 75.39 GPixel/s |

| Texture rate | 520.6 GTexel/s | 150.8 GTexel/s |

| TDP | 250 W | 300 W |

| Suggested PSU | 600 W | 700 W |

| Display outputs | 4x DisplayPort 1.4a | No outputs |

| Release date | 2018-03-26 | 2015-08-29 |

The GV100 also has a launch MSRP of 8,999 USD. The Tesla M60 has no launch MSRP recorded in the database.

# Head-to-Head Benchmarks

The database records two head-to-head benchmark comparisons between these GPUs: Geekbench OpenCL and Geekbench Vulkan. The Quadro GV100 wins both, and the margins are decisive.

In Geekbench OpenCL, the GV100 scores 150,004 against the M60's 29,506. The delta is 408.4%. This means the GV100 delivers more than five times the OpenCL compute throughput of the M60. This is the largest performance gap in the entire comparison and reflects the GV100's 5120 shading units, 640 tensor cores, and 868.4 GB/s memory bandwidth working in concert.

In Geekbench Vulkan, the GV100 scores 139,526 while the M60 scores 31,473. The delta is 343.3%. Vulkan performance often scales with shading unit count and memory bandwidth, both of which favor the GV100 heavily. The M60's 2048 shading units and 160.4 GB/s bandwidth simply cannot keep pace.

The average benchmark scores tell a similar story: the GV100's 35,520 average is 16.5% higher than the M60's 30,490. The GV100's nearest rivals include the NVIDIA GeForce RTX 5070 Ti Mobile (35,435, 0.2% ahead), the AMD Radeon Pro Duo (35,860, 0.9% behind), and the NVIDIA T1000 (36,289, 2.1% behind). The Tesla M60's nearest rivals include the NVIDIA CMP 70HX (30,476, 0% delta), the AMD Radeon RX 6700 (30,433, 0.2% ahead), and the AMD Radeon RX 6800 (30,095, 1.3% ahead).

In terms of percentile ranking, the GV100 sits at the 80th percentile of all GPUs, while the M60 is at the 75th percentile. That five-point gap in percentile is modest compared to the raw benchmark deltas, indicating that the M60 is still competitive within its own generation, but the GV100 operates in a higher performance tier.

The GV100 also shows superiority in pixel rate (208.3 GPixel/s vs 75.39 GPixel/s) and texture rate (520.6 GTexel/s vs 150.8 GTexel/s). These metrics are not directly benchmarked in the head-to-head tests, but they feed into the OpenCL and Vulkan results.

# The Verdict

The data points to a clear hierarchy: the NVIDIA Quadro GV100 is the superior GPU across every recorded benchmark and specification category. Its 408.4% lead in Geekbench OpenCL and 343.3% lead in Geekbench Vulkan are not incremental improvements; they represent a generational leap in compute capability.

For users who need maximum single-GPU performance in OpenCL or Vulkan workloads, the Quadro GV100 is the only choice between these two. Its 32 GB of HBM2 memory, 640 tensor cores, and 16.66 TFLOPS FP32 throughput make it suitable for compute-intensive tasks like AI training, scientific simulation, and high-resolution rendering. The GV100 also supports display outputs, so it can function as a workstation card, not just a compute accelerator.

The Tesla M60, on the other hand, is a datacenter-oriented card with no display outputs and 8 GB of GDDR5 memory. It was released in 2015, three years before the GV100, and its Maxwell architecture lacks tensor cores and FP16 support. Its 300 W TDP is higher than the GV100's 250 W, and it requires a 700 W suggested PSU versus the GV100's 600 W. None of its recorded benchmark scores approach the GV100's numbers.

However, the M60 is not without a niche. Its average benchmark score of 30,490 places it near the NVIDIA CMP 70HX and AMD Radeon RX 6700, meaning it still holds up reasonably well in general compute workloads. For server deployments where display output is unnecessary and multiple GPUs are virtualized for cloud workloads, the M60's simpler architecture and 8 GB capacity might suffice. But for any task where raw performance is the primary criterion, the GV100 wins outright.

The percentile data reinforces this: the GV100 at the 80th percentile versus the M60 at the 75th percentile. The GV100's nearest rivals are modern mobile and workstation parts, while the M60's nearest rivals are mid-range desktop GPUs. This places the two cards in different market segments entirely.

In summary, the Quadro GV100 is the stronger, more capable, and more modern GPU. The Tesla M60 is an older, lower-performance part that should only be considered for legacy server deployments or workloads where its specific feature set (no display outputs, dual-slot form factor) is required. For everyone else, the benchmark data is unambiguous: choose the Quadro GV100.