AMD Radeon RX Vega M GH vs NVIDIA P104-100 Comparison

FAQ

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

A: The NVIDIA P104-100 records an average benchmark score of 32,982, placing it in the 77th percentile of all GPUs. The AMD Radeon RX Vega M GH scores 29,197 on average, which lands in the 74th percentile.

Q: How do the two GPUs compare in the Geekbench OpenCL test?

A: The NVIDIA P104-100 scores 52,368, while the AMD Radeon RX Vega M GH scores 27,125. This gives NVIDIA a 93.1% advantage in that specific workload.

Q: Which GPU has the higher boost clock?

A: The NVIDIA P104-100 boosts to 1733 MHz, whereas the AMD Radeon RX Vega M GH reaches a boost clock of 1190 MHz.

Q: What memory configurations do these cards use?

A: Both have 4 GB of memory, but the NVIDIA P104-100 uses GDDR5X on a 256-bit bus with 320.3 GB/s of bandwidth. The AMD Radeon RX Vega M GH uses HBM2 on a 1024-bit bus with 204.8 GB/s of bandwidth.

Q: Are these GPUs still in production?

A: No, both are listed as end-of-life products. The NVIDIA P104-100 was released in December 2017, and the AMD Radeon RX Vega M GH followed in January 2018.

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

A: The AMD Radeon RX Vega M GH has a higher transistor density at 24.0 million transistors per mm², compared to the NVIDIA P104-100's 22.9 million per mm².

Architecture Differences

The NVIDIA P104-100 is built on the Pascal architecture using the GP104 chip, fabricated on a 16 nm process at TSMC. The AMD Radeon RX Vega M GH uses the GCN 4.0 architecture with the Polaris 22 chip, built on a 14 nm process at GlobalFoundries. These fundamental differences in architecture and manufacturing lead to distinct performance characteristics.

The NVIDIA chip packs 7,200 million transistors into a 314 mm² die, while the AMD chip contains 5,000 million transistors on a 208 mm² die. Despite having fewer transistors, the AMD chip achieves a higher transistor density of 24.0M per mm² versus 22.9M per mm² for NVIDIA. This suggests AMD packed its transistors more tightly, though NVIDIA's larger die allows for more total compute resources.

In terms of processing units, the NVIDIA P104-100 has 1,920 shading units, 120 texture mapping units, and 64 raster operations pipelines. The AMD Radeon RX Vega M GH has 1,536 shading units, 96 TMUs, and also 64 ROPs. NVIDIA maintains a 25% lead in shading units and TMUs, while the ROP count is identical.

Clock speeds differ substantially. The NVIDIA card runs at a base clock of 1607 MHz and boosts to 1733 MHz. The AMD card operates at a lower 1063 MHz base and 1190 MHz boost. These clock advantages contribute significantly to the NVIDIA card's higher theoretical throughput.

Memory architecture presents a stark contrast. The NVIDIA P104-100 uses 4 GB of GDDR5X on a 256-bit bus, delivering 320.3 GB/s of bandwidth. The AMD Radeon RX Vega M GH uses 4 GB of HBM2 on a 1024-bit bus but achieves only 204.8 GB/s. The wider bus on AMD's HBM2 does not compensate for the lower effective memory clock, as NVIDIA's memory runs at 10 Gbps effective versus AMD's 1600 Mbps effective.

Feature support differs in API levels. Both support DirectX 12 and OpenGL 4.6, but the NVIDIA card supports DirectX 12_1 while the AMD card only supports DirectX 12_0. Vulkan support also differs, with NVIDIA offering version 1.4 and AMD offering 1.3.

The physical form factors are entirely different. The NVIDIA P104-100 is a dual-slot card with a 267 mm length, requiring a 1x 8-pin power connector and a 200 W suggested PSU. The AMD Radeon RX Vega M GH is an integrated graphics processor (IGP) with no power connectors and no standalone dimensions, designed for portable devices.

Head-to-Head Benchmarks

The database contains two shared benchmark results between these GPUs, and the NVIDIA P104-100 wins both. The first is Geekbench OpenCL, where the NVIDIA card scores 52,368 against the AMD card's 27,125. This represents a 93.1% advantage for NVIDIA, nearly double the performance in this compute-oriented test. The gap is substantial and indicates that the NVIDIA card's higher clock speeds, more shading units, and greater memory bandwidth translate directly into compute throughput.

The second shared test is Geekbench Vulkan, where the NVIDIA P104-100 scores 45,165 and the AMD Radeon RX Vega M GH scores 31,268. Here the NVIDIA advantage narrows to 44.4%, but it remains a decisive win. The Vulkan test shows that while AMD's GCN architecture can hold its ground better in graphics workloads, it still trails NVIDIA in this particular benchmark.

Overall, the head-to-head record stands at 2 wins for the NVIDIA P104-100 and 0 wins for the AMD Radeon RX Vega M GH. The average benchmark score difference reinforces this picture, with NVIDIA at 32,982 versus AMD at 29,197, a gap of roughly 13%. However, the OpenCL test shows a much larger disparity than the average scores suggest, indicating that the NVIDIA card's advantage is workload-dependent.

Looking at the nearest rivals provides context. The NVIDIA P104-100 sits within 0.7% of the AMD Radeon Pro 570, the NVIDIA GeForce RTX 3050 Mobile, and the NVIDIA T550 Mobile. The AMD Radeon RX Vega M GH sits within 1.1% of the AMD Radeon RX 6800M and the AMD Radeon RX 470. This means each GPU is competitive with a different set of peers, and the performance gap between the two is not representative of a generational divide but rather a specific comparison.

Specification Differences

| Specification | NVIDIA P104-100 | AMD Radeon RX Vega M GH |

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

| Architecture | Pascal | GCN 4.0 |

| Process Node | 16 nm | 14 nm |

| Transistors | 7,200 million | 5,000 million |

| Die Size | 314 mm² | 208 mm² |

| Transistor Density | 22.9M / mm² | 24.0M / mm² |

| Base Clock | 1607 MHz | 1063 MHz |

| Boost Clock | 1733 MHz | 1190 MHz |

| Memory Type | GDDR5X | HBM2 |

| Memory Bus Width | 256 bit | 1024 bit |

| Memory Bandwidth | 320.3 GB/s | 204.8 GB/s |

| Memory Clock | 10 Gbps effective | 1600 Mbps effective |

| Shading Units | 1920 | 1536 |

| TMUs | 120 | 96 |

| ROPs | 64 | 64 |

| Pixel Rate | 110.9 GPixel/s | 76.16 GPixel/s |

| Texture Rate | 208.0 GTexel/s | 114.2 GTexel/s |

| FP32 | 6.655 TFLOPS | 3.656 TFLOPS |

| FP16 | 104.0 GFLOPS (1:64) | 3.656 TFLOPS (1:1) |

| TDP | Not specified | 100 W |

| Slot Width | Dual-slot | IGP |

| Power Connectors | 1x 8-pin | None |

| Suggested PSU | 200 W | None |

| Bus Interface | PCIe 1.0 x4 | IGP |

| Display Outputs | No outputs | Portable Device Dependent |

| DirectX | 12 (12_1) | 12 (12_0) |

| Vulkan | 1.4 | 1.3 |

The Verdict

The recorded data presents a clear performance hierarchy. The NVIDIA P104-100 wins both head-to-head benchmarks with substantial margins, achieving 93.1% higher scores in OpenCL and 44.4% higher scores in Vulkan. Its average benchmark score of 32,982 places it in the 77th percentile, while the AMD Radeon RX Vega M GH sits at 29,197 in the 74th percentile.

The NVIDIA card's advantages are rooted in measurable specifications. It has more shading units (1,920 versus 1,536), higher clocks (1733 MHz boost versus 1190 MHz), and significantly more memory bandwidth (320.3 GB/s versus 204.8 GB/s). Its FP32 throughput of 6.655 TFLOPS nearly doubles the AMD card's 3.656 TFLOPS. The pixel rate of 110.9 GPixel/s and texture rate of 208.0 GTexel/s also far exceed AMD's 76.16 GPixel/s and 114.2 GTexel/s.

However, the AMD Radeon RX Vega M GH is not without merit. It is an integrated processor designed for portable devices, with a 100 W TDP and no power connectors. Its FP16 performance of 3.656 TFLOPS matches its FP32 output, benefiting workloads that use half-precision math. The NVIDIA card's FP16 is severely limited at 104.0 GFLOPS with a 1:64 ratio, making it unsuitable for FP16 compute.

For users who need maximum compute and graphics performance in a discrete card, the NVIDIA P104-100 is the clear choice based on the benchmarks. For those building or using portable systems that require an integrated solution with reasonable performance and lower power draw, the AMD Radeon RX Vega M GH fits that role. The data does not support choosing the AMD card for raw performance, as it loses every shared benchmark.

Where Each One Wins

The NVIDIA P104-100 wins decisively in raw compute performance. Its OpenCL score of 52,368 is nearly double the AMD card's 27,125, making it the preferred option for general-purpose GPU compute tasks that rely on OpenCL. The Vulkan score of 45,165 versus 31,268 also favors NVIDIA, though the margin is smaller.

The NVIDIA card also wins in memory bandwidth-intensive workloads. With 320.3 GB/s of bandwidth compared to AMD's 204.8 GB/s, the NVIDIA card can feed its 1,920 shading units more effectively. Its 6.655 TFLOPS of FP32 performance and 208.0 GTexel/s texture rate suggest it handles high-resolution texture-heavy scenes with less bottleneck.

The AMD Radeon RX Vega M GH wins in FP16 compute scenarios. Its FP16 throughput of 3.656 TFLOPS matches its FP32 performance at a 1:1 ratio, whereas the NVIDIA card's FP16 is limited to 104.0 GFLOPS at a 1:64 ratio. Applications that use half-precision arithmetic would see a dramatic advantage on the AMD card.

The AMD card also wins on integration and power. As an IGP with a 100 W TDP and no external power connectors, it is designed for portable devices. The NVIDIA card requires a 1x 8-pin connector, a 200 W suggested PSU, and occupies a dual-slot form factor with a 267 mm length. For space-constrained or mobile systems, the AMD card is the only viable option despite its performance deficit.

The AMD card's HBM2 memory on a 1024-bit bus offers a different bandwidth profile, though the effective bandwidth is lower. Its 1:1 FP16 capability is a unique advantage that the NVIDIA card cannot match in its current configuration. For Vulkan workloads, the AMD card's closer score suggests it handles graphics APIs relatively better than compute APIs, narrowing the gap to 44.4% versus 93.1% in OpenCL.

Ultimately, the NVIDIA P104-100 is the performance winner across all shared benchmarks, while the AMD Radeon RX Vega M GH serves a distinct niche as an integrated GPU with FP16 capability and low power requirements.