AMD Radeon RX 5700 vs NVIDIA P106-100 Comparison

The NVIDIA P106-100 and AMD Radeon RX 5700 are two graphics cards from different eras and purposes. The P106-100 is a Pascal-based mining GPU with no display outputs, while the RX 5700 is a full-fledged RDNA 1.0 gaming card. Benchmark data shows a clear performance hierarchy, but the P106-100’s unique positioning as a compute-only card makes it an interesting comparison for specific workloads.

Head-to-Head Benchmarks

The data is unambiguous: the AMD Radeon RX 5700 wins all three head-to-head benchmark comparisons. The largest margin is in the 3DMark Steel Nomad DX12 test, where the RX 5700 scores 1862 against the P106-100’s 899. That represents a delta of -51.7%, meaning the P106-100 trails by more than half. This is a synthetic DirectX 12 workload that heavily taxes modern GPU features, and the RX 5700’s newer architecture and higher raw throughput dominate.

In Geekbench OpenCL, the RX 5700 scores 69603, while the P106-100 scores 35951. The delta here is -48.3%, so the P106-100 delivers roughly half the compute performance. This is a significant gap, but OpenCL is a general-purpose compute API, and the RX 5700’s 2304 shading units versus the P106-100’s 1280 explains much of the difference. The texture rate also differs substantially: the RX 5700 hits 248.4 GTexel/s, while the P106-100 manages 136.7 GTexel/s.

The Geekbench Vulkan test shows a similar story. The RX 5700 scores 64166, and the P106-100 scores 32897, a delta of -48.7%. Vulkan is a low-level API that benefits from driver optimization and architectural efficiency, and the RDNA 1.0 architecture in the RX 5700 is clearly better suited to it than the older Pascal design. The P106-100 does have a Vulkan 1.4 API rating, matching the RX 5700, but the actual performance gap is enormous.

Across all three benchmarks, the RX 5700 is roughly twice as fast as the P106-100. The closest margin is in OpenCL, but even there, the P106-100 only achieves 51.7% of the RX 5700’s score. The pixel rate also favors the RX 5700 at 110.4 GPixel/s versus 82.03 GPixel/s, and the FP32 throughput is 7.949 TFLOPS against 4.375 TFLOPS. These are not subtle differences; they are generational leaps.

The Verdict

The AMD Radeon RX 5700 is the clear winner for any standard GPU workload. It is 51.7% faster in 3DMark Steel Nomad, 48.3% faster in OpenCL, and 48.7% faster in Vulkan. The P106-100’s average benchmark score of 23249 is slightly higher than the RX 5700’s 22170, but that single number is misleading. The P106-100’s average is buoyed by its Geekbench scores, which are not directly comparable to the RX 5700’s broader test suite, which includes Metal, DirectX 9/10/11/12, and compute tests.

The RX 5700 also holds a higher percentile rank at 67 versus the P106-100’s 68, but that is within a rounding error. The real story is the head-to-head data. The RX 5700 wins every single test, and the margins are not close. If you are choosing between these two for gaming, rendering, or general compute, the RX 5700 is the only rational choice.

However, the P106-100 has a niche. It has no display outputs, making it useless for a standard desktop. But for a dedicated compute server or a mining rig where video output is unnecessary, it could serve a purpose. The RX 5700, with its 1x HDMI 2.0b and 3x DisplayPort 1.4a outputs, is a complete package. The verdict is straightforward: the RX 5700 wins on performance, features, and usability. The P106-100 only wins if you specifically need a card with no video outputs and are willing to accept half the performance.

Architecture Differences

The two cards are built on fundamentally different architectures. The NVIDIA P106-100 uses the Pascal architecture, which is a 16 nm design fabricated by TSMC. The chip is designated GP106, and it packs 4,400 million transistors on a 200 mm² die. The transistor density is 22.0M per mm². Pascal was designed for the GTX 10 series and is now dated, lacking modern features like hardware ray tracing or tensor cores.

The AMD Radeon RX 5700 uses the RDNA 1.0 architecture, specifically the Navi 10 chip. This is a 7 nm design, also from TSMC, which is a major process advantage. The RX 5700 has 10,300 million transistors on a 251 mm² die, resulting in a transistor density of 41.0M per mm². That is nearly double the density of the Pascal chip, which explains why the RX 5700 can pack 2304 shading units into a similar physical footprint.

The FP16 performance is a stark contrast. The P106-100 achieves 68.36 GFLOPS, which is a 1:64 ratio compared to its FP32 throughput. This is a crippled FP16 path, typical of mining-focused cards that do not need half-precision compute. The RX 5700, by contrast, hits 15.90 TFLOPS FP16 with a 2:1 ratio, meaning it can do two FP16 operations per FP32 operation. This makes the RX 5700 far more versatile for machine learning or other workloads that leverage half-precision.

The memory subsystems also reflect the generational gap. The P106-100 uses 6 GB of GDDR5 on a 192-bit bus, yielding 192.2 GB/s of bandwidth. The RX 5700 uses 8 GB of GDDR6 on a 256-bit bus, delivering 448.0 GB/s. That is more than double the bandwidth, which is critical for high-resolution textures and compute tasks. The RX 5700 also supports PCIe 4.0 x16, while the P106-100 is limited to PCIe 1.0 x16, a severe bottleneck for data transfer.

Specification Differences

The specification sheets show clear divergences. The P106-100 has a base clock of 1506 MHz and a boost clock of 1709 MHz. The RX 5700 has a lower base clock of 1465 MHz but a higher boost clock of 1725 MHz, plus a game clock of 1625 MHz. The memory clocks differ significantly: the P106-100 runs at 2002 MHz (8 Gbps effective), while the RX 5700 runs at 1750 MHz (14 Gbps effective).

The P106-100 has 1280 shading units, 80 TMUs, and 48 ROPs. The RX 5700 has 2304 shading units, 144 TMUs, and 64 ROPs. These are massive differences in raw compute resources. The TDP also differs: the P106-100 draws 120 W, while the RX 5700 draws 180 W. The power connectors are 1x 6-pin for the P106-100 and 1x 6-pin + 1x 8-pin for the RX 5700. The suggested PSU is 300 W for the P106-100 and 450 W for the RX 5700.

The physical dimensions vary too. The P106-100 is 250 mm long (9.8 inches), while the RX 5700 is 268 mm long (10.6 inches). The RX 5700 also has listed height and width of 111 mm and 36 mm, respectively, while the P106-100 lacks those measurements. The P106-100 has no display outputs, while the RX 5700 has 1x HDMI 2.0b and 3x DisplayPort 1.4a. The RX 5700 was released on 2019-07-06, with the P106-100 launching on 2017-06-18.

FAQ

Q: Is the NVIDIA P106-100 any good for gaming?

A: No. It has no display outputs, so you cannot connect a monitor directly. Its 3DMark Steel Nomad score of 899 is 51.7% lower than the RX 5700’s 1862, indicating weak DirectX 12 performance.

Q: Which card has more memory bandwidth?

A: The AMD Radeon RX 5700 has 448.0 GB/s of bandwidth, compared to the P106-100’s 192.2 GB/s. This is due to the RX 5700’s 256-bit GDDR6 memory bus versus the P106-100’s 192-bit GDDR5 bus.

Q: Can the P106-100 be used for compute workloads?

A: Yes, but it is limited. Its Geekbench OpenCL score of 35951 is 48.3% lower than the RX 5700’s 69603. Its FP16 performance is severely restricted at 68.36 GFLOPS, making it poor for half-precision tasks.

Q: What is the power draw difference?

A: The P106-100 has a TDP of 120 W and requires a 300 W PSU. The RX 5700 has a TDP of 180 W and requires a 450 W PSU. The RX 5700 also needs both a 6-pin and an 8-pin power connector, while the P106-100 only needs a single 6-pin.

Q: Which card is more likely to be found in a modern system?

A: The RX 5700, as it was released in 2019 and supports PCIe 4.0 x16. The P106-100, released in 2017, uses PCIe 1.0 x16, which is a significant bottleneck for data transfer.

Q: How do their average benchmark scores compare?

A: The P106-100 has a higher average benchmark score of 23249, while the RX 5700 scores 22170. However, this is misleading because the RX 5700’s suite includes more diverse tests, and the head-to-head results show the RX 5700 winning every shared benchmark.

Where Each One Wins

The AMD Radeon RX 5700 wins in every category that matters for a general-purpose GPU. It is 51.7% faster in 3DMark Steel Nomad, 48.3% faster in OpenCL, and 48.7% faster in Vulkan. It has more memory, higher bandwidth, more shading units, and a more modern architecture. It is the obvious choice for gaming, content creation, and any compute task that requires a display output.

The NVIDIA P106-100 wins in exactly one scenario: when you need a GPU with no display outputs and you are on a strict power budget. Its 120 W TDP is 60 W lower than the RX 5700, and it only needs a 300 W PSU versus 450 W. It is also shorter at 250 mm, which could fit in smaller chassis. But these advantages are niche. For a mining rig or a dedicated compute server where you never plug in a monitor, the P106-100’s lower power draw and smaller footprint are tangible benefits.

The P106-100 also has a slightly higher percentile rank at 68 versus 67, but that is a statistical artifact of its limited benchmark suite. The RX 5700’s nearest rival is the GeForce GTX 1060 6 GB at a 1.4% delta, while the P106-100’s nearest rival is the Radeon Pro Vega 16 at a 0% delta. Neither card is a top performer, but the RX 5700 is clearly the more capable and versatile product. Choose the RX 5700 unless your specific use case demands a card with no video outputs and lower power consumption.