AMD Radeon RX 560 XT vs NVIDIA P104-100 Comparison

AMD Radeon RX 560 XT and NVIDIA P104-100 are both end-of-life graphics cards, but they target fundamentally different workloads, and the benchmark data reflects that split. The NVIDIA P104-100 is the clear winner in raw compute performance, taking both available head-to-head benchmarks with significant margins, while the AMD Radeon RX 560 XT offers a more conventional feature set with display outputs and a standard PCIe interface. The data shows a 2-0 win tally for the NVIDIA card, but the AMD card holds its own in specific API workloads and offers practical usability that the mining-focused NVIDIA card lacks.

Where Each One Wins

The NVIDIA P104-100 dominates in synthetic compute benchmarks, which aligns with its mining GPU heritage. In Geekbench OpenCL, the NVIDIA card scores 52,368 against the AMD card’s 31,387, a 40.1% advantage that places it firmly ahead in raw number-crunching tasks. This is not a marginal win; it is a decisive margin that suggests the NVIDIA card is substantially better suited for compute-heavy applications like cryptocurrency mining or general-purpose GPU compute. The NVIDIA card also wins in Geekbench Vulkan with a score of 45,165 versus 36,879, a smaller but still clear 18.3% lead.

The AMD Radeon RX 560 XT, despite losing both head-to-head tests, wins in practical versatility. It features display outputs (1x HDMI 2.0b and 3x DisplayPort 1.4a), making it usable for gaming or workstation tasks that require a monitor connection. The NVIDIA P104-100 has no display outputs at all, which disqualifies it from any traditional graphics workload. The AMD card also uses a standard PCIe 3.0 x16 interface, while the NVIDIA card is limited to PCIe 1.0 x4, a severe bandwidth constraint that could bottleneck data transfer in many scenarios. In terms of API support, the AMD card offers Vulkan 1.3 and DirectX 12 (12_0), while the NVIDIA card supports Vulkan 1.4 and DirectX 12 (12_1), so the NVIDIA card has a slight edge in API versioning, but the AMD card’s connectivity makes it the only choice for users who need a functional graphics output.

Architecture Differences

The two cards come from different architectural generations and foundries. The AMD Radeon RX 560 XT uses the Ellesmere chip built on GCN 4.0 architecture, manufactured on a 14 nm process at GlobalFoundries. The NVIDIA P104-100 uses the GP104 chip on Pascal architecture, built on a 16 nm process at TSMC. The NVIDIA chip is physically larger and more complex: it packs 7,200 million transistors on a 314 mm² die, compared to the AMD chip’s 5,700 million transistors on a 232 mm² die. Interestingly, the AMD card has a higher transistor density at 24.6M per mm² versus NVIDIA’s 22.9M per mm², despite the older process node.

Core configurations differ significantly. The NVIDIA P104-100 has 1,920 shading units, 120 texture mapping units, and 64 raster operations pipelines. The AMD Radeon RX 560 XT has 1,792 shading units, 112 TMUs, and only 32 ROPs. This means the NVIDIA card has roughly 7% more shaders, 7% more TMUs, and double the ROP count. The ROP difference is particularly stark and explains the NVIDIA card’s massive lead in pixel rate: 110.9 GPixel/s versus 39.23 GPixel/s. Texture rate also favors NVIDIA at 208.0 GTexel/s versus 137.3 GTexel/s. FP32 performance follows the same pattern, with NVIDIA delivering 6.655 TFLOPS against AMD’s 4.394 TFLOPS.

Memory subsystems also diverge. Both cards have 4 GB of VRAM and a 256-bit bus width, but the NVIDIA card uses faster GDDR5X memory running at 10 Gbps effective, yielding 320.3 GB/s bandwidth. The AMD card uses GDDR5 at 7 Gbps effective, giving 224.0 GB/s bandwidth. That is a 43% bandwidth advantage for NVIDIA, which contributes heavily to its compute lead. The NVIDIA card also has a much higher base and boost clock: 1607 MHz base and 1733 MHz boost, versus AMD’s 1074 MHz base and 1226 MHz boost. FP16 performance is a notable exception: the AMD card delivers 4.394 TFLOPS at 1:1 ratio, while the NVIDIA card is severely limited at 104.0 GFLOPS (1:64), meaning the AMD card is 42 times faster in half-precision compute.

Head-to-Head Benchmarks

The Geekbench OpenCL test shows the NVIDIA P104-100 at 52,368 points, a full 40.1% ahead of the AMD Radeon RX 560 XT’s 31,387 points. This is the largest margin in any benchmark and reflects the NVIDIA card’s superior shader count, clock speeds, and memory bandwidth. The AMD card’s nearest rivals include the NVIDIA RTX A2000 12 GB (34,154) and RTX A1000 (34,207), which are only 0.1% and 0.2% ahead respectively, showing that the AMD card sits in a competitive mid-range bracket. The NVIDIA P104-100, by contrast, is clustered with mobile workstation GPUs like the NVIDIA T600 Mobile (32,849) and T550 Mobile (33,161), which are 0.4% behind and 0.5% ahead respectively, placing it in a different performance tier.

In Geekbench Vulkan, the NVIDIA card wins again with 45,165 points versus 36,879, a 18.3% margin. This is a smaller gap than OpenCL, suggesting the AMD card’s GCN architecture handles Vulkan workloads relatively better than its raw compute capabilities would indicate. Still, the NVIDIA card’s lead is substantial. The AMD card’s Vulkan score is closer to its OpenCL score (36,879 vs 31,387), while the NVIDIA card’s Vulkan score is actually lower than its OpenCL score (45,165 vs 52,368), indicating that NVIDIA’s advantage narrows in API-specific workloads. The overall average benchmark scores reflect this: the AMD card averages 34,133 points with a 79th percentile ranking among all GPUs, while the NVIDIA card averages 32,982 points with a 77th percentile ranking. This is counterintuitive given NVIDIA’s head-to-head wins, but it means the AMD card is more consistent across a broader range of tests, likely due to its more flexible feature set.

FAQ

Q: Which card has higher raw compute performance?

A: The NVIDIA P104-100 wins both head-to-head benchmarks. It scores 52,368 in Geekbench OpenCL versus 31,387 for the AMD Radeon RX 560 XT (40.1% ahead), and 45,165 in Geekbench Vulkan versus 36,879 (18.3% ahead).

Q: Can the NVIDIA P104-100 be used for gaming or display output?

A: No. The NVIDIA P104-100 has no display outputs, making it unsuitable for any task requiring a monitor connection. The AMD Radeon RX 560 XT offers 1x HDMI 2.0b and 3x DisplayPort 1.4a outputs.

Q: What is the memory bandwidth difference?

A: The NVIDIA P104-100 has 320.3 GB/s bandwidth using GDDR5X memory, while the AMD Radeon RX 560 XT has 224.0 GB/s using GDDR5 memory. Both have 4 GB capacity and a 256-bit bus.

Q: Which card has better half-precision (FP16) performance?

A: The AMD Radeon RX 560 XT is dramatically better, delivering 4.394 TFLOPS at a 1:1 ratio. The NVIDIA P104-100 delivers only 104.0 GFLOPS at a 1:64 ratio, making the AMD card roughly 42 times faster in FP16 workloads.

Q: How do the cards compare in terms of PCIe interface?

A: The AMD card uses PCIe 3.0 x16, a standard full-bandwidth slot. The NVIDIA card uses PCIe 1.0 x4, which is severely bandwidth-limited and may bottleneck performance in data-intensive tasks.

Q: What is the transistor count and die size difference?

A: The NVIDIA P104-100 has 7,200 million transistors on a 314 mm² die (16 nm TSMC). The AMD Radeon RX 560 XT has 5,700 million transistors on a 232 mm² die (14 nm GlobalFoundries). NVIDIA’s die is 35% larger physically.

Specification Differences

| Specification | AMD Radeon RX 560 XT | NVIDIA P104-100 |

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

| Chip | Ellesmere | GP104 |

| Architecture | GCN 4.0 | Pascal |

| Process Node | 14 nm (GlobalFoundries) | 16 nm (TSMC) |

| Transistors | 5,700 million | 7,200 million |

| Die Size | 232 mm² | 314 mm² |

| Transistor Density | 24.6M / mm² | 22.9M / mm² |

| Base Clock | 1074 MHz | 1607 MHz |

| Boost Clock | 1226 MHz | 1733 MHz |

| Memory Type | GDDR5 | GDDR5X |

| Memory Speed | 7 Gbps effective | 10 Gbps effective |

| Memory Bandwidth | 224.0 GB/s | 320.3 GB/s |

| Shading Units | 1792 | 1920 |

| TMUs | 112 | 120 |

| ROPs | 32 | 64 |

| Pixel Rate | 39.23 GPixel/s | 110.9 GPixel/s |

| Texture Rate | 137.3 GTexel/s | 208.0 GTexel/s |

| FP32 Performance | 4.394 TFLOPS | 6.655 TFLOPS |

| FP16 Performance | 4.394 TFLOPS (1:1) | 104.0 GFLOPS (1:64) |

| TDP | 150 W | Not specified |

| Power Connectors | 1x 6-pin | 1x 8-pin |

| Suggested PSU | 450 W | 200 W |

| Bus Interface | PCIe 3.0 x16 | PCIe 1.0 x4 |

| Display Outputs | 1x HDMI 2.0b, 3x DisplayPort 1.4a | No outputs |

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

| Vulkan Support | 1.3 | 1.4 |

| Length | 241 mm (9.5 inches) | 267 mm (10.5 inches) |

| Release Date | 2019-03-12 | 2017-12-11 |

The Verdict

The data points to a clear split: the NVIDIA P104-100 is the superior compute card, winning both benchmarks with 40.1% and 18.3% margins, and it offers higher clocks, more shaders, double the ROPs, and significantly more memory bandwidth. It also has a newer DirectX version (12_1 vs 12_0) and Vulkan 1.4 support. However, the NVIDIA card is a mining-specific product with no display outputs and a crippled PCIe 1.0 x4 interface, which makes it impractical for any general-purpose graphics use. Its average benchmark score of 32,982 is actually lower than the AMD card’s 34,133, and its 77th percentile ranking trails the AMD card’s 79th percentile, suggesting that in a broader benchmark suite, the AMD card is more balanced.

For users who need a functional graphics card with display outputs, standard PCIe connectivity, and competitive FP16 performance, the AMD Radeon RX 560 XT is the only viable option. It also wins on transistor density (24.6M vs 22.9M per mm²) and has a lower power draw at 150 W, though the NVIDIA card’s TDP is unspecified. The AMD card’s 1:1 FP16 ratio is a unique advantage that the NVIDIA card cannot match, making it relevant for certain compute workloads. For pure compute tasks where display output and PCIe bandwidth are irrelevant — such as dedicated mining rigs — the NVIDIA P104-100 is the clear choice based on raw score advantages. The 2-0 benchmark sweep for NVIDIA is decisive, but the AMD card’s versatility and higher average benchmark score make it the more well-rounded product. Choose NVIDIA for raw compute, choose AMD for anything that requires a functional GPU.