AMD Radeon Pro WX 3200 vs NVIDIA P106-090 Comparison

Head-to-Head Benchmarks

The recorded data shows a single head-to-head benchmark comparison between the NVIDIA P106-090 and the AMD Radeon Pro WX 3200, and the result is decisively in favor of the NVIDIA part. In the Geekbench OpenCL test, the P106-090 scores 21,304 points, while the WX 3200 manages 11,228 points. That is an 89.7% advantage for the NVIDIA card, a massive gap that dwarfs the typical deltas seen between adjacent products in the database. For context, the P106-090's average benchmark score of 13,470 puts it just 0.3% behind the GeForce GTX 570 (13,515) and 0.5% ahead of the Radeon Pro 555 (13,407). The WX 3200, with an average score of 11,228, sits essentially level with the FirePro W4300 (11,225) and 1.3% above the RTX PRO 6000 Blackwell Max-Q (11,088).

The 89.7% delta in OpenCL is not a marginal win; it is a category-level separation. The P106-090 delivers nearly double the compute throughput in this workload, which reflects its larger GPU configuration and higher clock behavior. The WX 3200, by contrast, lands in a lower performance tier, with its average score placing it only at the 50th percentile of all GPUs, versus the P106-090 at the 54th percentile. While a single benchmark does not capture every possible workload, the magnitude of this result is consistent with the underlying hardware differences: the P106-090 has more shading units, more texture units, more ROPs, a wider memory bus, and higher raw FP32 throughput.

FAQ

Q: Which GPU wins the only directly comparable benchmark in the database?

A: The NVIDIA P106-090 wins decisively. In Geekbench OpenCL, it scores 21,304 against the AMD Radeon Pro WX 3200's 11,228, an 89.7% advantage.

Q: How does the P106-090 compare to its nearest rivals in average benchmark score?

A: The P106-090's average score of 13,470 is 0.3% below the GeForce GTX 570 (13,515) and 0.5% below the Radeon HD 7770M (13,536), while sitting 0.5% above the Radeon Pro 555 (13,407) and 0.5% above the Radeon RX 9070 XT (13,543).

Q: Where does the WX 3200 rank relative to its closest competitors?

A: The WX 3200's average score of 11,228 is essentially tied with the FirePro W4300 (11,225, 0% delta), 0.3% behind the GeForce GTX 780M (11,261), and 1.3% ahead of both the RTX PRO 6000 Blackwell Max-Q and RTX PRO 6000D Blackwell Max-Q (both 11,088).

Q: What is the percentile ranking of each card among all GPUs?

A: The P106-090 sits at the 54th percentile, while the WX 3200 sits at the 50th percentile, meaning the NVIDIA card outperforms a larger share of the entire GPU population.

Q: Does the WX 3200 have any benchmark win over the P106-090 in the database?

A: No. The head-to-head data records 1 win for the P106-090 and 0 wins for the WX 3200.

Q: What is the launch MSRP of the WX 3200?

A: The launch MSRP for the AMD Radeon Pro WX 3200 is 199 USD. The P106-090 has no recorded launch MSRP in the database.

Architecture Differences

The two cards come from different architectural families and different manufacturing processes. The NVIDIA P106-090 is built on the Pascal architecture using the GP106 chip, fabricated by TSMC on a 16 nm process. The die contains 4,400 million transistors on a 200 mm² slice, yielding a transistor density of 22.0 million per mm². The AMD Radeon Pro WX 3200 uses the GCN 4.0 architecture with the Polaris 23 chip, produced by GlobalFoundries on a 14 nm process. Its die is considerably smaller at 103 mm², holding 2,200 million transistors, which works out to 21.4 million per mm². Despite the smaller die, the AMD part is built on a slightly older (larger) process node, so the transistor densities are close, but the NVIDIA chip carries twice as many transistors.

The compute resources differ sharply. The P106-090 has 768 shading units, 48 texture mapping units, and 48 ROPs. The WX 3200 has 640 shading units, 32 TMUs, and only 16 ROPs. That means the NVIDIA card has 20% more shaders, 50% more texture units, and triple the ROP count. These differences directly explain the performance gap in fill-rate and texture-heavy workloads. The pixel rate for the P106-090 is 73.49 GPixel/s versus 20.72 GPixel/s for the WX 3200, a 3.5x advantage. Texture rate is 73.49 GTexel/s versus 41.44 GTexel/s, a 1.77x lead for NVIDIA.

The FP32 compute figures reinforce the same story: the P106-090 delivers 2.352 TFLOPS, while the WX 3200 delivers 1.658 TFLOPS. Interestingly, the FP16 performance reveals a stark architectural difference. The P106-090 lists FP16 at 36.74 GFLOPS with a 1:64 ratio, meaning its FP16 throughput is dramatically reduced relative to FP32, a common trait for mining-oriented Pascal parts. The WX 3200, on the other hand, offers FP16 at 1.658 TFLOPS with a 1:1 ratio, matching its FP32 output. For workloads that use FP16 arithmetic, the AMD card is far more capable, but for general FP32 compute, the NVIDIA card holds a clear lead.

The API support also differs. Both support DirectX 12 and OpenGL 4.6, but the P106-090 lists DirectX 12_1 and Vulkan 1.4, while the WX 3200 lists DirectX 12_0 and Vulkan 1.3. The NVIDIA card supports a newer DirectX feature level and a newer Vulkan version.

Specification Differences

The two cards diverge on nearly every major specification. Memory configuration is a clear split: the P106-090 has 3 GB of GDDR5 on a 192-bit bus, yielding 192.2 GB/s of bandwidth, while the WX 3200 has 4 GB of GDDR5 on a 128-bit bus, yielding only 96.00 GB/s. The NVIDIA card has double the memory bandwidth despite having less capacity, which is a significant advantage for data-heavy tasks.

Clock speeds also differ. The P106-090 has a base clock of 1354 MHz and a boost clock of 1531 MHz, with memory running at 2002 MHz (8 Gbps effective). The WX 3200 has no base or boost clock listed, but its memory is clocked at 1500 MHz (6 Gbps effective). The NVIDIA part runs its memory substantially faster, contributing to its bandwidth advantage.

Power and physical design are also opposites. The P106-090 has a TDP of 75 W, uses a dual-slot cooler, requires a single 6-pin power connector, and is 250 mm long (9.8 inches). It has no display outputs, which aligns with its mining-oriented generation. The WX 3200 has a TDP of 65 W, uses a single-slot design, requires no power connectors, and is 167 mm long (6.6 inches) with a height of 69 mm (2.7 inches). It offers 4x mini-DisplayPort 1.4a outputs, making it a functional workstation card for display purposes.

The bus interfaces differ as well. The P106-090 uses PCIe 1.0 x1, a severely limited interface, while the WX 3200 uses PCIe 3.0 x8. This means the AMD card can move data between the GPU and system far more efficiently, which is crucial for professional workloads that stream data. The P106-090's x1 interface is a clear bottleneck, even if its compute and memory bandwidth are superior.

Release timing and product lineage also differ. The P106-090 was released on 2017-07-30, while the WX 3200 was released on 2019-07-01. The WX 3200 has a recorded predecessor, the Radeon Pro GCN, and a successor, the Radeon Pro Vega. The P106-090 has no predecessor or successor listed. Both cards are marked as end-of-life in production status.

The Verdict

The benchmark data points to one clear conclusion: the NVIDIA P106-090 is the stronger compute performer in the tested workload. Its 89.7% lead in Geekbench OpenCL, combined with higher average benchmark score (13,470 versus 11,228) and a better percentile ranking (54th versus 50th), makes it the obvious choice for raw compute throughput. The P106-090 also dominates in memory bandwidth (192.2 GB/s versus 96.00 GB/s), pixel rate, texture rate, and FP32 throughput.

However, the WX 3200 is not without its own advantages, and these are not captured by the single head-to-head benchmark. The AMD card offers 4 GB of memory versus 3 GB, which is relevant for workloads that need larger working sets. It has a functional display output configuration with 4x mini-DisplayPort 1.4a, while the P106-090 has no outputs at all. The WX 3200 uses PCIe 3.0 x8, which is far more practical than PCIe 1.0 x1 for system integration. It also has full-rate FP16 compute (1:1 ratio), which the P106-090 lacks entirely.

The verdict depends on the use case. For anyone prioritizing compute performance in OpenCL-style workloads, the P106-090 is the clear winner. For anyone needing a working GPU with display outputs, modern PCIe connectivity, and FP16 capability, the WX 3200 is the only viable option between the two. The data does not support calling the WX 3200 a better performer; it simply serves a different role.

Where Each One Wins

The NVIDIA P106-090 wins in every benchmark category where the two are directly compared. Geekbench OpenCL is the only recorded head-to-head test, and the P106-090 takes it by 89.7%. The raw specification sheet reinforces this dominance: higher FP32 throughput (2.352 TFLOPS versus 1.658 TFLOPS), higher pixel rate (73.49 GPixel/s versus 20.72 GPixel/s), higher texture rate (73.49 GTexel/s versus 41.44 GTexel/s), and double the memory bandwidth (192.2 GB/s versus 96.00 GB/s). The P106-090 also has more shading units (768 versus 640), more TMUs (48 versus 32), and more ROPs (48 versus 16). For any workload that stresses these resources, the P106-090 is the superior part.

The AMD Radeon Pro WX 3200 wins in areas that are not measured by the recorded benchmarks but are visible in the specification sheet. It has a larger memory pool at 4 GB, which is critical for applications that exceed 3 GB of usage. It supports FP16 at full rate (1.658 TFLOPS, 1:1 ratio), making it suitable for workloads that rely on half-precision arithmetic. It has a modern PCIe 3.0 x8 interface, enabling faster host-to-device transfers than the P106-090's PCIe 1.0 x1. It includes 4x mini-DisplayPort 1.4a outputs, allowing multi-monitor setups. It is also a smaller, lower-power card with no external power connector requirement, fitting into more constrained system builds.

The use-case split is clean: the P106-090 is for compute-heavy tasks where performance per benchmark score matters most, and the WX 3200 is for professional environments that need a functional, display-capable GPU with FP16 support and a practical system interface. The database shows no scenario where the WX 3200 outperforms the P106-090 in raw compute, but the WX 3200's feature set makes it the appropriate choice for workstation deployment rather than raw speed.