AMD Radeon RX 6800M vs NVIDIA P104-100 Comparison

The Verdict

The recorded data presents a clear split between these two GPUs. The AMD Radeon RX 6800M wins every head-to-head benchmark in the database, taking all three comparisons with decisive margins. Its average benchmark score of 28,874 places it in the 74th percentile of all GPUs, while the NVIDIA P104-100 posts an average of 32,982 and sits in the 77th percentile. The percentile ranking is notable: despite the P104-100 having a higher average score, the RX 6800M wins every direct comparison where both were tested under identical conditions.

The NVIDIA P104-100 is a mining-focused card with no display outputs, which limits its usefulness for conventional desktop workloads. The AMD Radeon RX 6800M is a mobile part with portable-device-dependent outputs, making it the practical choice for any application requiring an actual video signal. For users who need compute performance in an accessible form factor, the RX 6800M is the data-backed selection. The P104-100 remains relevant only for specialized compute tasks where its specific configuration and end-of-life availability are acceptable trade-offs.

Architecture Differences

The two GPUs come from fundamentally different design philosophies. The NVIDIA P104-100 uses the GP104 chip built on the Pascal architecture, fabricated on a 16 nm process at TSMC. It packs 7,200 million transistors onto a 314 mm² die, yielding a transistor density of 22.9 million transistors per square millimeter. The AMD Radeon RX 6800M uses the Navi 22 chip with RDNA 2.0 architecture, also built by TSMC but on a 7 nm process. This newer node allows AMD to fit 17,200 million transistors onto a 335 mm² die, achieving a density of 51.3 million transistors per square millimeter, more than double the density of the Pascal part.

The AMD card belongs to the Radeon RX 6000 series and the Navi Mobile generation, while the NVIDIA card is classified under Mining GPUs. The RX 6800M includes 40 ray tracing cores, a feature entirely absent from the P104-100, which reports no ray tracing hardware. The P104-100 also lacks tensor cores. In terms of compute resources, the RX 6800M fields 2,560 shading units, 160 texture mapping units, and 64 ROPs. The P104-100 counters with 1,920 shading units, 120 TMUs, and the same 64 ROPs.

Memory architecture diverges sharply. The P104-100 uses 4 GB of GDDR5X on a 256-bit bus, delivering 320.3 GB/s of bandwidth. The RX 6800M uses 12 GB of GDDR6 on a narrower 192-bit bus, yet achieves higher bandwidth at 384.0 GB/s thanks to faster 16 Gbps effective memory clocks versus 10 Gbps effective on the NVIDIA card. The RX 6800M also supports DirectX 12 Ultimate (12_2), while the P104-100 is limited to DirectX 12 (12_1). Both support OpenGL 4.6 and Vulkan 1.4.

Head-to-Head Benchmarks

The database contains three direct comparisons between these GPUs, and the AMD Radeon RX 6800M wins all of them. In the 3DMark Steel Nomad DX12 test, the RX 6800M scores 2,238 against the P104-100's 1,413, a difference of 36.9 percent in AMD's favor. This is a modern DirectX 12 workload, and the RX 6800M's newer architecture with hardware ray tracing support likely contributes to the substantial gap.

The Geekbench OpenCL test shows the RX 6800M at 87,621 versus the P104-100 at 52,368, a 40.2 percent advantage. This compute-oriented benchmark reflects the RX 6800M's higher throughput: it delivers 12.24 TFLOPS of FP32 performance compared to 6.655 TFLOPS for the P104-100. The AMD card also has a massive FP16 advantage at 24.47 TFLOPS (2:1 ratio) versus the P104-100's 104.0 GFLOPS (1:64 ratio), which is a stark generational difference in compute capability.

The largest delta appears in Geekbench Vulkan, where the RX 6800M scores 94,766 against 45,165 for the P104-100, a 52.3 percent lead. This result indicates that the AMD architecture extracts significantly more performance from the Vulkan API, which is relevant for both gaming and compute workloads that leverage this cross-platform interface.

The P104-100 does not win any head-to-head test in the database. Its only notable advantages appear in its aggregate benchmark average and percentile placement, which are influenced by the specific test suite applied to each card. The P104-100's OpenCL score of 52,368 and Vulkan score of 45,165 are respectable for a 2017 mining card, but they fall well short of the RX 6800M in every shared workload.

Specification Differences

The two cards differ across nearly every measurable specification. The process node moves from 16 nm on the P104-100 to 7 nm on the RX 6800M. Transistor count jumps from 7,200 million to 17,200 million. Die size increases modestly from 314 mm² to 335 mm², but transistor density more than doubles from 22.9 million per square millimeter to 51.3 million.

Clock speeds favor AMD substantially. The RX 6800M has a base clock of 2116 MHz, a game clock of 2300 MHz, and a boost clock of 2390 MHz. The P104-100 runs at 1607 MHz base and 1733 MHz boost, which is lower despite being a desktop-oriented card. Memory clocks also differ: the P104-100 runs at 1251 MHz with 10 Gbps effective, while the RX 6800M runs at 2000 MHz with 16 Gbps effective.

The shading unit count rises from 1,920 to 2,560, and TMUs rise from 120 to 160, while ROPs remain at 64 for both. Pixel rate improves from 110.9 GPixel/s to 153.0 GPixel/s, and texture rate from 208.0 GTexel/s to 382.4 GTexel/s. FP32 throughput nearly doubles from 6.655 TFLOPS to 12.24 TFLOPS. FP16 performance is not even comparable: 104.0 GFLOPS on the P104-100 versus 24.47 TFLOPS on the RX 6800M.

Power and physical specifications differ as well. The P104-100 is a dual-slot card with a 1x 8-pin power connector and a suggested PSU of 200 W, but reports no TDP figure. The RX 6800M is an IGP (integrated graphics processor for mobile) with no power connectors and a TDP of 145 W. The P104-100 measures 267 mm (10.5 inches) in length and offers no display outputs. The RX 6800M has no listed dimensions and uses portable-device-dependent outputs. The bus interface moves from PCIe 1.0 x4 on the P104-100 to PCIe 4.0 x16 on the RX 6800M, a substantial difference in host connectivity bandwidth.

The release dates are separated by roughly three and a half years: the P104-100 launched on December 11, 2017, and the RX 6800M on May 30, 2021. The RX 6800M lists Polaris Mobile as its predecessor. Both cards are marked as end-of-life in the database.

FAQ

Q: Which GPU wins the most benchmarks in the database?

A: The AMD Radeon RX 6800M wins all three head-to-head tests: 3DMark Steel Nomad DX12, Geekbench OpenCL, and Geekbench Vulkan. The NVIDIA P104-100 records zero wins in direct comparisons.

Q: How large is the performance gap in Vulkan workloads?

A: The RX 6800M scores 94,766 in Geekbench Vulkan versus 45,165 for the P104-100, which represents a 52.3 percent advantage for the AMD card, the largest delta among the recorded head-to-head tests.

Q: Which card has more memory?

A: The AMD Radeon RX 6800M has 12 GB of GDDR6 memory on a 192-bit bus with 384.0 GB/s bandwidth. The NVIDIA P104-100 has 4 GB of GDDR5X on a 256-bit bus with 320.3 GB/s bandwidth.

Q: Does either card support ray tracing?

A: The RX 6800M includes 40 ray tracing cores as part of its RDNA 2.0 architecture. The P104-100 reports no ray tracing cores.

Q: What are the FP32 compute figures for each card?

A: The RX 6800M delivers 12.24 TFLOPS of FP32 performance, while the P104-100 delivers 6.655 TFLOPS. The AMD card is roughly 84 percent higher in this metric based on the recorded values.

Q: Which card has display outputs?

A: The P104-100 has no display outputs, as it was designed for mining operations. The RX 6800M has portable-device-dependent outputs, meaning its display support depends on the host device.

Where Each One Wins

The AMD Radeon RX 6800M wins in every measured category where both cards were tested. Its advantages span modern DirectX 12 workloads (36.9 percent ahead in 3DMark Steel Nomad), general compute via OpenCL (40.2 percent ahead), and Vulkan rendering (52.3 percent ahead). The card also leads in raw specifications: higher clocks, more shading units, more TMUs, double the FP32 throughput, and vastly superior FP16 performance. Its 12 GB memory capacity and higher bandwidth make it the better choice for memory-intensive workloads. The inclusion of ray tracing cores gives it a feature set the P104-100 cannot match. The RX 6800M is the appropriate pick for any workload that benefits from modern APIs, compute throughput, or ray tracing, and its portable-device-dependent outputs mean it can actually drive a display.

The NVIDIA P104-100 holds advantages in a narrower set of circumstances. Its average benchmark score of 32,982 exceeds the RX 6800M's 28,874, and its 77th percentile placement versus the AMD card's 74th suggests that in the broader database context, the P104-100 performs comparably to a different class of GPU. Its dual-slot form factor and 1x 8-pin power connector make it a conventional desktop installation, and its 267 mm length fits standard cases. The 256-bit memory bus, while paired with less total memory, provides a different memory topology that may suit certain compute patterns. The P104-100 also has a lower transistor count and smaller die, which could be relevant for applications sensitive to those physical characteristics. For users who already own this card or have access to it at no cost, it remains a functional compute device, particularly given its end-of-life status. However, based strictly on the head-to-head data, there is no benchmark category where the P104-100 outperforms the RX 6800M.