GPU Comparison

NVIDIA P104-100 and NVIDIA GRID M60-1Q are both end-of-life, dual-slot, no-display-output accelerator boards built for compute workloads, but they represent entirely different generations of GPU design. The data shows a single head-to-head benchmark result, alongside each card's broader average benchmark scores, which positions them within a percentile ranking of all GPUs: the P104-100 sits at the 77th percentile, while the GRID M60-1Q trails slightly at the 76th percentile.

Head-to-Head Benchmarks

The only direct comparison available in the dataset is the Geekbench Vulkan test. In this test, the NVIDIA P104-100 scores 45,165, while the NVIDIA GRID M60-1Q scores 31,220. This results in a decisive 44.7% victory for the P104-100. The delta is substantial, indicating that the P104-100 delivers significantly higher raw compute throughput in a modern graphics API workload.

Beyond this direct matchup, the average benchmark scores reinforce the gap. The P104-100's average benchmark score is 32,982, whereas the GRID M60-1Q's average is 31,220. This difference of 1,762 points (approximately 5.6%) is modest compared to the Vulkan delta, but it still favors the P104-100. The P104-100 also has a richer benchmark profile, with scores in 3DMark Steel Nomad DX12 (1,413) and Geekbench OpenCL (52,368), in addition to the Vulkan result. The GRID M60-1Q only has a single recorded Geekbench Vulkan score, which limits direct cross-test comparisons.

The nearest-rivals data provides additional context for the P104-100. Its average score of 32,982 places it within 0.4% of the NVIDIA T600 Mobile (32,849), 0.5% above the NVIDIA T550 Mobile (33,161), 0.6% above the NVIDIA GeForce RTX 3050 Mobile (33,170), and 0.7% above the AMD Radeon Pro 570 (33,207). These are all mobile or lower-tier desktop parts, yet the P104-100's average performance is essentially on par with them, despite being a mining-oriented card with a PCIe 1.0 x4 interface.

For the GRID M60-1Q, the nearest rivals show a different competitive landscape. Its average score of 31,220 is virtually identical to the NVIDIA Quadro M5000 (31,206, 0% delta), slightly ahead of the NVIDIA GeForce RTX 4070 Ti SUPER (31,087, 0.4% delta), and behind the NVIDIA RTX PRO 4500 Blackwell (31,532, -1% delta) and the NVIDIA TITAN RTX (31,676, -1.4% delta). Notably, the GRID M60-1Q's average score is competitive with much newer and more powerful consumer cards, which speaks to the efficiency of its compute configuration in the recorded benchmark.

Where Each One Wins

The P104-100 wins the only head-to-head benchmark, and its additional benchmark results suggest strengths across multiple APIs. Its Geekbench OpenCL score of 52,368 is its highest recorded result, indicating strong general-purpose compute performance. The 3DMark Steel Nomad DX12 score of 1,413 further demonstrates its capability in a modern DirectX 12 workload. These results, combined with the Vulkan win, suggest the P104-100 is the better choice for diverse, contemporary compute tasks.

The GRID M60-1Q, despite losing the Vulkan test, still holds its own in the average benchmark score comparison. Its single Vulkan score of 31,220 is its only data point, but it is enough to place it in the 76th percentile overall. This suggests that while it is not as fast as the P104-100 in raw Vulkan performance, it remains a competent compute device. The absence of OpenCL or DX12 results for the GRID M60-1Q prevents a broader assessment of its strengths; the data only supports a conclusion that it is competitive in Vulkan, but behind the P104-100.

In terms of memory, the P104-100 offers 4 GB of GDDR5X on a 256-bit bus with 320.3 GB/s bandwidth, while the GRID M60-1Q offers only 1024 MB (1 GB) of GDDR5 on a 256-bit bus with 160.4 GB/s bandwidth. This is a major differentiator. The P104-100 has twice the bandwidth and four times the memory capacity, which directly impacts workloads that are memory-bound or require large datasets. The GRID M60-1Q's memory size is a severe limitation for modern workloads, and the data shows the P104-100 is the clear winner in this regard.

Architecture Differences

The two cards are built on different architectures and process nodes. The P104-100 uses the GP104 chip on the Pascal architecture, fabricated on a 16 nm process at TSMC. It contains 7,200 million transistors on a die size of 314 mm², yielding a transistor density of 22.9M per mm². The GRID M60-1Q uses the GM204 chip on the Maxwell 2.0 architecture, fabricated on a 28 nm process at TSMC. It contains 5,200 million transistors on a larger die size of 398 mm², resulting in a lower transistor density of 13.1M per mm². The newer 16 nm process allows the P104-100 to pack more transistors into a smaller area, which is a key factor in its performance advantage.

Clock speeds also differ significantly. The P104-100 has a base clock of 1607 MHz and a boost clock of 1733 MHz, with memory running at 1251 MHz (10 Gbps effective). The GRID M60-1Q has a much lower base clock of 557 MHz, but a boost clock of 1178 MHz, with memory at 1253 MHz (5 Gbps effective). The P104-100's higher base and boost clocks contribute to its performance lead. The memory clock is similar in raw MHz, but the P104-100's GDDR5X memory operates at 10 Gbps effective versus 5 Gbps for the GRID M60-1Q's GDDR5, doubling the effective data rate.

The compute unit configurations are comparable in count but not in generation. The P104-100 has 1920 shading units, 120 TMUs, and 64 ROPs. The GRID M60-1Q has 2048 shading units, 128 TMUs, and 64 ROPs. The GRID M60-1Q actually has more shading units and TMUs, but the P104-100 achieves higher throughput due to its higher clocks and architecture efficiency. This is reflected in the pixel and texture rates: the P104-100 delivers 110.9 GPixel/s and 208.0 GTexel/s, while the GRID M60-1Q delivers 75.39 GPixel/s and 150.8 GTexel/s. The FP32 performance is 6.655 TFLOPS for the P104-100 versus 4.825 TFLOPS for the GRID M60-1Q, a 38% advantage for the P104-100.

Power and interface specifications also differ. The GRID M60-1Q has a TDP of 225 W and a suggested PSU of 550 W, while the P104-100 has no TDP listed but a suggested PSU of 200 W. The P104-100 uses a PCIe 1.0 x4 interface, which is notably older and narrower than the GRID M60-1Q's PCIe 3.0 x16 interface. This is a potential bottleneck for the P104-100 in data transfer-heavy tasks, but the benchmark data does not indicate a negative impact on compute scores. Both cards use a dual-slot form factor, a single 8-pin power connector, and have no display outputs. They also share identical API support for DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.4.

The Verdict

The data clearly favors the NVIDIA P104-100 in direct performance comparisons. It wins the head-to-head Vulkan test by 44.7%, has a higher average benchmark score, and offers significantly more memory and bandwidth. For any workload that relies on Vulkan, OpenCL, or DirectX 12, the P104-100 is the superior choice based on the available benchmarks. Its higher FP32 throughput and faster memory make it a more capable compute accelerator for modern tasks.

The GRID M60-1Q, while losing the direct comparison, is not without merit. Its average benchmark score is competitive with much newer GPUs, as shown in its nearest-rivals list, which includes the GeForce RTX 4070 Ti SUPER and the TITAN RTX. This suggests that in specific compute scenarios, it can hold its own. However, its 1 GB memory capacity is a hard limitation that the data cannot ignore. The P104-100's 4 GB capacity provides a substantial practical advantage for any workload that requires more than 1 GB of working set.

For users prioritizing raw compute performance in modern APIs, the P104-100 is the clear pick. For users with legacy workloads that were optimized for Maxwell architecture or that are constrained by the PCIe 3.0 x16 interface, the GRID M60-1Q may still be relevant, but the benchmark data does not support a performance advantage over the P104-100. The P104-100 is the stronger card in every measured category.

FAQ

Q: Which GPU has a higher Geekbench Vulkan score?

A: The NVIDIA P104-100 scores 45,165, which is 44.7% higher than the NVIDIA GRID M60-1Q's score of 31,220.

Q: What is the average benchmark score difference between the two cards?

A: The P104-100 has an average benchmark score of 32,982, while the GRID M60-1Q has an average of 31,220, a difference of approximately 5.6% in favor of the P104-100.

Q: How much memory does each card have, and what is the bandwidth?

A: The P104-100 has 4 GB of GDDR5X memory with 320.3 GB/s bandwidth. The GRID M60-1Q has 1024 MB (1 GB) of GDDR5 memory with 160.4 GB/s bandwidth.

Q: What are the FP32 performance figures for both cards?

A: The P104-100 delivers 6.655 TFLOPS FP32 performance, while the GRID M60-1Q delivers 4.825 TFLOPS.

Q: Which card has a higher transistor density?

A: The P104-100 has a transistor density of 22.9M per mm², compared to the GRID M60-1Q's 13.1M per mm², due to its 16 nm process versus the 28 nm process.

Q: Do both cards support the same APIs?

A: Yes, both cards support DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.4, according to the data.