AMD Radeon Pro WX 3200 vs NVIDIA GeForce GTX 670 Comparison

Head-to-Head Benchmarks

The recorded data contains one direct comparison between these two cards, and it is decisive. In the Geekbench OpenCL test, the NVIDIA GeForce GTX 670 scores 15,341 points against 11,228 points for the AMD Radeon Pro WX 3200. That is a 36.6% advantage for the GTX 670, a substantial margin that places the older NVIDIA card clearly ahead in this compute-oriented workload. The GTX 670 wins the only head-to-head benchmark in the database, with a win count of 1 versus 0 for the WX 3200.

Looking at the broader benchmark averages, the gap narrows but the direction remains the same. The GTX 670 has an average benchmark score of 12,773 across all recorded tests, while the WX 3200 averages 11,228. This places the GTX 670 at the 52nd percentile among all GPUs, with the WX 3200 at the 50th percentile. The percentile difference is small, but the average score difference is roughly 13.8% in favor of the GTX 670, which is consistent with the OpenCL result showing a clear performance tier separation.

The nearest rivals for each card help contextualize these scores. The GTX 670 sits within a tight cluster: its average of 12,773 is only 0.4% below the NVIDIA GeForce GTX 590 (12,830) and 0.4% below the AMD Radeon Pro 455 (12,831), while it is 0.5% above the AMD Radeon RX 7600M XT (12,710) and 0.6% below the AMD FirePro W5100 (12,847). This tells us the GTX 670 is essentially dead-even with a set of mid-range cards from both vendors, with no single rival more than 0.6% away. The WX 3200, by contrast, sits in a similarly tight but lower cluster: its 11,228 average is exactly matched by the AMD FirePro W4300 (11,225, 0% delta), 0.3% below the NVIDIA GeForce GTX 780M (11,261), and 1.3% above both the NVIDIA RTX PRO 6000 Blackwell Max-Q and NVIDIA RTX PRO 6000D Blackwell Max-Q (both 11,088). The WX 3200 is therefore a full tier below the GTX 670 in raw compute throughput, even though both cards are close to their own immediate rivals.

It is importantly the GTX 670 has three recorded benchmark scores: 7,424 in Geekbench Metal, 15,341 in Geekbench OpenCL, and 15,553 in Geekbench Vulkan. The WX 3200 has only one recorded score, the 11,228 in Geekbench OpenCL. The OpenCL result is the only directly comparable test, and it favors the GTX 670 by a wide margin. The Vulkan score for the GTX 670 is even higher than its OpenCL score, suggesting the card is particularly strong in that API, but without a WX 3200 Vulkan result, no direct comparison can be made there. Similarly, the Metal score is lower but still respectable, though again no WX 3200 equivalent exists in the database.

Architecture Differences

The two cards come from different architectural generations and foundries. The NVIDIA GeForce GTX 670 uses the GK104 chip, built on the Kepler architecture, manufactured by TSMC on a 28 nm process. The AMD Radeon Pro WX 3200 uses the Polaris 23 chip, built on GCN 4.0, manufactured by GlobalFoundries on a 14 nm process. The node difference is significant: 28 nm versus 14 nm means the AMD part is two process generations ahead in lithography, which explains the dramatic differences in die size and transistor density.

The GTX 670 packs 3,540 million transistors into a die of 294 mm², giving it a transistor density of 12.0 million per mm². The WX 3200, by contrast, has 2,200 million transistors on a die of only 103 mm², yielding a density of 21.4 million per mm². The AMD chip is therefore much denser, with roughly 78% more transistors per square millimeter, a direct consequence of the smaller process node. However, the GTX 670 has a larger absolute transistor count: about 61% more transistors than the WX 3200, which partially explains its higher compute throughput.

The compute resources differ sharply. The GTX 670 has 1,344 shading units, 112 texture mapping units (TMUs), and 32 raster operation units (ROPs). The WX 3200 has 640 shading units, 32 TMUs, and 16 ROPs. This means the GTX 670 has more than double the shading units, 3.5 times the TMUs, and double the ROPs. The raw throughput numbers reflect this: the GTX 670 achieves 2.634 TFLOPS of FP32 compute, 27.44 GPixel/s pixel rate, and 109.8 GTexel/s texture rate. The WX 3200 delivers 1.658 TFLOPS FP32, 20.72 GPixel/s, and 41.44 GTexel/s. The GTX 670 is ahead by roughly 59% in FP32, 32% in pixel rate, and 165% in texture rate.

Clock speeds differ as well. The GTX 670 has a base clock of 915 MHz and a boost clock of 980 MHz, while the WX 3200 has no recorded base or boost clocks in the database. Memory clocks are comparable, with the GTX 670 at 1502 MHz (6 Gbps effective) and the WX 3200 at 1500 MHz (6 Gbps effective). The memory subsystems, however, are very different. The GTX 670 has 2 GB of GDDR5 on a 256-bit bus, yielding 192.3 GB/s of bandwidth. The WX 3200 has 4 GB of GDDR5 on a 128-bit bus, yielding 96.00 GB/s. The GTX 670 has double the memory bus width and double the bandwidth, while the WX 3200 has double the capacity.

The feature sets also differ. The GTX 670 supports DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.2.175. The WX 3200 supports DirectX 12 (12_0), OpenGL 4.6, and Vulkan 1.3. The AMD card has a higher DirectX feature level and a newer Vulkan version. Additionally, the WX 3200 has FP16 support at a 1:1 ratio (1.658 TFLOPS), while the GTX 670 has no recorded FP16 capability. The power and physical characteristics are also divergent: the GTX 670 has a TDP of 170 W, is dual-slot, requires 2x 6-pin power connectors, and suggests a 450 W PSU. The WX 3200 has a TDP of 65 W, is single-slot, needs no power connectors, and suggests a 250 W PSU. The GTX 670 is 241 mm long, 111 mm tall, and 38 mm wide, while the WX 3200 is 167 mm long and 69 mm tall with no recorded width. The GTX 670 uses PCIe 3.0 x16, while the WX 3200 uses PCIe 3.0 x8. Display outputs also differ: the GTX 670 has 2x DVI, 1x HDMI 1.4a, and 1x DisplayPort 1.2, while the WX 3200 has 4x mini-DisplayPort 1.4a.

FAQ

Q: Which card has the higher average benchmark score?

A: The NVIDIA GeForce GTX 670 has an average benchmark score of 12,773, which is higher than the AMD Radeon Pro WX 3200's average of 11,228. The GTX 670 also sits at the 52nd percentile of all GPUs, two points above the WX 3200's 50th percentile.

Q: How do the cards compare in the only shared benchmark test?

A: In the Geekbench OpenCL test, the GTX 670 scores 15,341 versus 11,228 for the WX 3200, a 36.6% advantage for the NVIDIA card. This is the only head-to-head benchmark recorded in the database, and the GTX 670 wins it outright.

Q: What are the memory capacity and bandwidth differences?

A: The GTX 670 has 2 GB of GDDR5 memory on a 256-bit bus, providing 192.3 GB/s of bandwidth. The WX 3200 has 4 GB of GDDR5 on a 128-bit bus, providing 96.00 GB/s. The GTX 670 has double the bandwidth, while the WX 3200 has double the capacity.

Q: Which card is more power-efficient?

A: The WX 3200 has a TDP of 65 W and requires no power connectors, while the GTX 670 has a TDP of 170 W and requires 2x 6-pin connectors. The WX 3200 also recommends a 250 W PSU versus 450 W for the GTX 670, indicating a much lower power draw.

Q: Do the cards support different API versions?

A: The GTX 670 supports DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.2.175. The WX 3200 supports DirectX 12 (12_0), OpenGL 4.6, and Vulkan 1.3. The WX 3200 has a higher DirectX feature level and a newer Vulkan version, while the GTX 670 has no FP16 support and the WX 3200 has FP16 at 1:1.

Q: What are the release dates and MSRPs?

A: The GTX 670 was released on May 9, 2012, with a launch MSRP of 399 USD. The WX 3200 was released on July 1, 2019, with a launch MSRP of 199 USD. Both cards are now end-of-life.

Specification Differences

| Specification | NVIDIA GeForce GTX 670 | AMD Radeon Pro WX 3200 |

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

| Architecture | Kepler | GCN 4.0 |

| Process Node | 28 nm | 14 nm |

| Foundry | TSMC | GlobalFoundries |

| Transistors | 3,540 million | 2,200 million |

| Die Size | 294 mm² | 103 mm² |

| Transistor Density | 12.0M / mm² | 21.4M / mm² |

| Base Clock | 915 MHz | Not recorded |

| Boost Clock | 980 MHz | Not recorded |

| Memory Size | 2 GB | 4 GB |

| Memory Bus Width | 256 bit | 128 bit |

| Memory Bandwidth | 192.3 GB/s | 96.00 GB/s |

| Shading Units | 1344 | 640 |

| TMUs | 112 | 32 |

| ROPs | 32 | 16 |

| Pixel Rate | 27.44 GPixel/s | 20.72 GPixel/s |

| Texture Rate | 109.8 GTexel/s | 41.44 GTexel/s |

| FP32 | 2.634 TFLOPS | 1.658 TFLOPS |

| FP16 | Not recorded | 1.658 TFLOPS (1:1) |

| TDP | 170 W | 65 W |

| Slot Width | Dual-slot | Single-slot |

| Power Connectors | 2x 6-pin | None |

| Suggested PSU | 450 W | 250 W |

| Bus Interface | PCIe 3.0 x16 | PCIe 3.0 x8 |

| Display Outputs | 2x DVI, 1x HDMI 1.4a, 1x DisplayPort 1.2 | 4x mini-DisplayPort 1.4a |

| DirectX | 12 (11_0) | 12 (12_0) |

| Vulkan | 1.2.175 | 1.3 |

| Length | 241 mm | 167 mm |

| Height | 111 mm | 69 mm |

| Width | 38 mm | Not recorded |

| Release Date | 2012-05-09 | 2019-07-01 |

| Launch MSRP | 399 USD | 199 USD |

The Verdict

The data is unambiguous on raw performance: the NVIDIA GeForce GTX 670 is the faster card. Its 36.6% lead in the only shared benchmark, the Geekbench OpenCL test, is a large margin, and its average benchmark score of 12,773 versus 11,228 for the AMD Radeon Pro WX 3200 confirms that advantage across the broader database. The GTX 670 wins on every measured compute metric: FP32 throughput is 59% higher, texture rate is 165% higher, and pixel rate is 32% higher. It also has more than double the shading units, TMUs, and ROPs, plus double the memory bandwidth.

However, the WX 3200 is not without its own strengths, and these are architectural rather than performance-based. It is built on a much newer 14 nm process with a higher transistor density (21.4M / mm² versus 12.0M / mm²), giving it a smaller die (103 mm² versus 294 mm²) with fewer transistors (2,200 million versus 3,540 million). It has double the memory capacity at 4 GB versus 2 GB, which could matter for workloads with large working sets. It also has a dramatically lower TDP of 65 W versus 170 W, requires no power connectors, is single-slot instead of dual-slot, and is physically smaller. It supports newer API versions, including DirectX 12 (12_0) versus 12 (11_0) and Vulkan 1.3 versus 1.2.175, and it has FP16 support at 1:1, which the GTX 670 lacks.

The verdict depends on the priority. For anyone prioritizing raw compute performance, the GTX 670 is the clear choice based on the recorded benchmarks. For anyone prioritizing efficiency, physical footprint, memory capacity, or modern API support, the WX 3200 has compelling advantages. The GTX 670 is an end-of-life card from 2012, while the WX 3200 is an end-of-life card from 2019, so neither is current, but their strengths are distinct.

Where Each One Wins

NVIDIA GeForce GTX 670 wins on raw compute and throughput. The benchmark data shows a 36.6% advantage in Geekbench OpenCL, and the specification sheet supports this with higher FP32 (2.634 TFLOPS versus 1.658 TFLOPS), higher texture rate (109.8 GTexel/s versus 41.44 GTexel/s), higher pixel rate (27.44 GPixel/s versus 20.72 GPixel/s), and double the memory bandwidth (192.3 GB/s versus 96.00 GB/s). The GTX 670 also has more shading units (1,344 versus 640), more TMUs (112 versus 32), and more ROPs (32 versus 16). This is the card for compute-heavy tasks where raw throughput matters most.

AMD Radeon Pro WX 3200 wins on efficiency, capacity, and modern features. The WX 3200 has a TDP of 65 W versus 170 W, requires no external power connectors versus 2x 6-pin, and suggests a 250 W PSU versus 450 W. It is single-slot versus dual-slot, and its dimensions are smaller (167 mm length, 69 mm height versus 241 mm length, 111 mm height). It has 4 GB of memory versus 2 GB, which could be decisive for datasets that exceed the GTX 670's capacity. It also has newer API support: DirectX 12 (12_0) versus 12 (11_0), Vulkan 1.3 versus 1.2.175, and FP16 at 1:1 versus no recorded FP16. The WX 3200's 4x mini-DisplayPort 1.4a outputs also exceed the GTX 670's mixed output set in terms of modern display connectivity.

The split is clear: choose the GTX 670 for maximum performance per the recorded benchmarks, especially in OpenCL workloads, and choose the WX 3200 for low-power, low-profile deployments, larger memory capacity, or environments that require the latest API versions. The GTX 670 is the stronger card in the head-to-head test, but the WX 3200 is the more modern, efficient, and feature-rich option in the specifications that do not appear in the benchmarks.