AMD Radeon RX 9070 vs NVIDIA P106-100 Comparison

The AMD Radeon RX 9070 and the NVIDIA P106-100 occupy entirely different eras of GPU design, and the benchmark data reflects a generational chasm. The RX 9070 wins all three shared benchmark tests decisively, with its most dominant showing in 3DMark Steel Nomad DX12, where it scores 6290 against the P106-100’s 899—a 599.7% advantage. This is not a marginal victory; it is a complete rout that underscores how far GPU architecture has progressed since the P106-100’s mining-oriented release.

Head-to-Head Benchmarks

The most striking disparity appears in 3DMark Steel Nomad DX12, a modern, compute-heavy workload. The RX 9070’s score of 6290 crushes the P106-100’s 899, a delta of 599.7%. This test scales with raw shading power, memory bandwidth, and driver optimization—all areas where the RDNA 4.0 part holds overwhelming superiority. The P106-100, built for cryptocurrency mining rather than gaming, lacks the architectural features to keep pace in a DirectX 12 Ultimate workload.

Geekbench OpenCL tells a similar story, albeit with a narrower gap. The RX 9070 posts 131,539 points versus 35,951 for the P106-100, a 265.9% difference. OpenCL exercises general-purpose compute, and the RX 9070’s 36.13 TFLOPS FP32 throughput dwarfs the P106-100’s 4.375 TFLOPS. The 53,900 million transistors on the RX 9070’s 357 mm² die provide a massive compute reservoir that the older 4,400-million-transistor GP106 chip cannot approach.

The Vulkan benchmark is the closest contest, but the RX 9070 still wins decisively. Its score of 58,705 beats the P106-100’s 32,897 by 78.5%. Vulkan’s lower-level API reduces driver overhead, which narrows the gap somewhat, but the RX 9070’s superior hardware resources—3584 shading units versus 1280, and 56 RT cores versus none—ensure a comfortable margin. The P106-100’s Pascal architecture, while competent for its era, simply lacks the instruction-level efficiency of RDNA 4.0.

Across all three tests, the RX 9070’s average benchmark score of 23,877 places it just 0.6% above the NVIDIA GeForce GTX TITAN Z, while the P106-100’s 23,249 average sits 0.1% below the AMD Radeon Pro Vega 16. The RX 9070’s percentile rank of 69 vs all GPUs is nearly identical to the P106-100’s 68, but that metric is skewed by the inclusion of non-gaming accelerators; within their respective peer groups, the performance gap is enormous.

Where Each One Wins

The RX 9070 wins in every measurable category from the data. Its advantages are most pronounced in modern, feature-rich workloads: DirectX 12 Ultimate title with ray tracing, compute-heavy OpenCL tasks, and Vulkan-based engines. The 3DMark Steel Nomad result (599.7% lead) indicates the RX 9070 is built for current-generation gaming, where techniques like mesh shaders and variable rate shading are standard. Its 16 GB GDDR6 memory with 644.6 GB/s bandwidth provides 3.35 times the bandwidth of the P106-100, which matters in high-resolution textures and large scene data.

The P106-100 has no benchmark wins in the head-to-head data. Its only theoretical advantage lies in power efficiency—a 120 W TDP versus 220 W—but that is a physical property, not a performance attribute. In practice, the P106-100’s 6 GB GDDR5 memory and 192-bit bus limit it to 1080p-era workloads, and its lack of display outputs means it cannot even drive a monitor without a companion GPU. The RX 9070, by contrast, supports 1x HDMI 2.1b and 3x DisplayPort 2.1a outputs, making it a complete standalone solution.

For users running legacy DirectX 9 or 11 titles, the P106-100’s Pascal architecture retains some relevance, but the RX 9070’s passmark scores—343 in DX9, 281 in DX11—show it handles those APIs well too. The P106-100 does not provide passmark scores in the data, so no direct comparison is possible for those older APIs, but the RX 9070’s 25,381 G3D score versus its 14,737 GPU compute score indicates balanced performance across both rasterization and compute tasks.

Architecture Differences

The architectural gulf between these two GPUs is vast. The RX 9070 uses the Navi 48 chip on TSMC’s 4 nm process, packing 53,900 million transistors into a 357 mm² die—a transistor density of 151.0M per mm². The P106-100 uses the GP106 chip on TSMC’s 16 nm process, with 4,400 million transistors on a 200 mm² die, yielding just 22.0M per mm². The RX 9070 achieves nearly seven times the transistor density, enabling far more complex logic in a similar physical footprint.

The RX 9070’s RDNA 4.0 architecture introduces dedicated ray tracing hardware: 56 RT cores, a feature entirely absent from the P106-100’s Pascal design. This explains the massive 3DMark Steel Nomad delta, as that test includes ray-traced effects. The RX 9070 also doubles down on shader throughput with 3584 shading units, 224 TMUs, and 128 ROPs, versus 1280 shading units, 80 TMUs, and 48 ROPs on the P106-100. The pixel rate of 322.6 GPixel/s versus 82.03 GPixel/s demonstrates a 4x fill-rate advantage.

Memory architecture differs fundamentally. The RX 9070 uses 16 GB GDDR6 on a 256-bit bus at 20.1 Gbps effective, achieving 644.6 GB/s. The P106-100 uses 6 GB GDDR5 on a 192-bit bus at 8 Gbps effective, limited to 192.2 GB/s. The RX 9070 also supports PCIe 5.0 x16, while the P106-100 is stuck on PCIe 1.0 x16—a bus standard from 2003 that severely bottlenecks data transfer. The FP16 compute ratio also tells a story: the RX 9070 runs FP16 at a 1:1 ratio with FP32 (36.13 TFLOPS), while the P106-100 offers a paltry 1:64 ratio (68.36 GFLOPS), making it unsuitable for modern machine learning tasks.

FAQ

Q: Which GPU has higher raw compute performance?

A: The AMD Radeon RX 9070 delivers 36.13 TFLOPS FP32, while the NVIDIA P106-100 manages 4.375 TFLOPS—an 8.3x difference in theoretical peak performance.

Q: Can the NVIDIA P106-100 output video to a display?

A: No. The P106-100 has no display outputs, making it strictly a compute or mining card. The RX 9070 offers 1x HDMI 2.1b and 3x DisplayPort 2.1a.

Q: How do their memory bandwidths compare?

A: The RX 9070 provides 644.6 GB/s over a 256-bit GDDR6 interface, while the P106-100 provides 192.2 GB/s over a 192-bit GDDR5 interface—a 3.35x advantage for the RX 9070.

Q: Which card supports ray tracing?

A: Only the RX 9070, which includes 56 RT cores in its RDNA 4.0 architecture. The P106-100 has no ray tracing hardware whatsoever.

Q: What are their production statuses?

A: The RX 9070 is listed as Active and was released on 2025-03-05. The P106-100 is End-of-life, released on 2017-06-18.

Q: How large is the die size difference?

A: The RX 9070’s Navi 48 die measures 357 mm², while the P106-100’s GP106 die measures 200 mm²—despite being smaller, the RX 9070 packs 49,500 million more transistors.

Specification Differences

The two GPUs diverge on nearly every specification. The process node differs drastically: 4 nm for the RX 9070 versus 16 nm for the P106-100. Transistor count is 53,900 million versus 4,400 million, a 12.25x difference. Die size is 357 mm² versus 200 mm². The RX 9070’s base clock runs at 1330 MHz with a 2520 MHz boost, while the P106-100 runs at 1506 MHz base and 1709 MHz boost—the older card actually has a higher base clock, but lower boost.

Memory capacity is 16 GB GDDR6 versus 6 GB GDDR5, with bus widths of 256-bit versus 192-bit. Bandwidth is 644.6 GB/s versus 192.2 GB/s. The RX 9070 packs 3584 shading units, 224 TMUs, and 128 ROPs; the P106-100 has 1280 shading units, 80 TMUs, and 48 ROPs. The RX 9070 adds 56 RT cores; the P106-100 has none. Pixel rate is 322.6 GPixel/s versus 82.03 GPixel/s, and texture rate is 564.5 GTexel/s versus 136.7 GTexel/s.

Power requirements differ: the RX 9070 draws 220 W with 2x 8-pin connectors and a 550 W suggested PSU, while the P106-100 draws 120 W with 1x 6-pin and a 300 W suggested PSU. The bus interface is PCIe 5.0 x16 versus PCIe 1.0 x16. The RX 9070 supports DirectX 12 Ultimate (12_2) and Vulkan 1.4; the P106-100 supports DirectX 12 (12_1) and Vulkan 1.4. The RX 9070 measures dual-slot width with no length listed; the P106-100 measures 250 mm (9.8 inches) in length. The RX 9070 launched with an MSRP of 549 USD; the P106-100 has no recorded launch MSRP.