AMD Radeon Pro WX 4100 vs NVIDIA GeForce GTX 670M Comparison

The Verdict

The data presents a decisive outcome: the AMD Radeon Pro WX 4100 is the overwhelmingly faster GPU in compute workloads, while the NVIDIA GeForce GTX 670M is effectively a legacy part with a single benchmark score that places it in a similar overall percentile. In the sole head-to-head benchmark available, Geekbench OpenCL, the Radeon Pro WX 4100 scores 17,642 against the GTX 670M’s 6,513, a delta of -63.1% from the AMD card’s perspective. That is not a marginal victory; it is a generational gap. The GTX 670M’s average benchmark score of 6,513 places it at the 38th percentile of all GPUs, while the WX 4100’s average of 6,330 sits at the 37th percentile — a statistical near-tie in aggregate ranking, but that aggregate is pulled down by the WX 4100’s additional legacy DirectX and compute tests. For any user whose priority is raw OpenCL throughput, the WX 4100 is the only rational choice from this dataset. The GTX 670M, with its 2012 release date and end-of-life status, offers no modern feature set to counterbalance its performance deficit. The WX 4100 is a professional card with a launch MSRP of 399 USD, and it delivers on that positioning with a 2.460 TFLOPS FP32 rating — over three times the GTX 670M’s 803.7 GFLOPS. The verdict is unambiguous: the Radeon Pro WX 4100 wins on performance, efficiency, and API support. The GTX 670M is only relevant for legacy system compatibility or as a point of historical comparison.

Architecture Differences

The two GPUs come from entirely different architectural eras. The NVIDIA GeForce GTX 670M is built on the GF114 chip, using the Fermi 2.0 architecture, and manufactured on a 40 nm process at TSMC. The AMD Radeon Pro WX 4100 uses the Baffin chip, based on GCN 4.0, and is fabricated by GlobalFoundries on a 14 nm node. This process difference is stark: the GTX 670M packs 1,950 million transistors into a 332 mm² die, yielding a transistor density of 5.9 million per square millimeter. The WX 4100 crams 3,000 million transistors into just 123 mm², achieving a density of 24.4 million per square millimeter. That is a 4x improvement in density, which explains much of the performance disparity. The GTX 670M’s Fermi architecture is from the GeForce 600M generation, released on 2012-03-21; the WX 4100 is a Radeon Pro Polaris (WX x100) part from the GCN 4.0 generation, released on 2016-11-09. The WX 4100’s predecessor is listed as Radeon Pro GCN, and its successor is Radeon Pro Vega; the GTX 670M’s predecessor is GeForce 500M, and its successor is GeForce 700M. In terms of compute resources, the WX 4100 has 1,024 shading units, 64 texture mapping units, and 16 ROPs. The GTX 670M has 336 shading units, 56 TMUs, and 24 ROPs. The WX 4100 has more than triple the shader count, which directly feeds its FP32 advantage. The GTX 670M has a higher ROP count, but that does little to help in compute-bound OpenCL workloads. The WX 4100 also supports FP16 at a 1:1 ratio with FP32 (2.460 TFLOPS each), while the GTX 670M lists no FP16 capability at all. This is a fundamental architectural shift: Fermi was designed before the compute-focused era that GCN 4.0 embraced.

Head-to-Head Benchmarks

The only direct comparison available is the Geekbench OpenCL test, and it is a landslide. The AMD Radeon Pro WX 4100 scores 17,642, while the NVIDIA GeForce GTX 670M scores 6,513. The delta percentage is -63.1%, meaning the GTX 670M is roughly 63% slower than the WX 4100 in this workload. To put that in perspective, the GTX 670M’s nearest rivals include the NVIDIA GeForce GT 555M (average score 6,493, delta 0.3%), the NVIDIA Quadro M5000M (6,481, delta 0.5%), and the AMD Radeon Vega 10 Mobile (6,476, delta 0.6%). The GTX 670M is essentially tied with these parts, all hovering around the 6,500 mark. The Intel UHD Graphics P750 is slightly ahead at 6,554, a delta of -0.6% from the GTX 670M’s perspective. The WX 4100, in contrast, sits in a different league entirely. Its nearest rivals are the AMD Radeon R7 M350 (average score 6,327, delta 0.1%), the NVIDIA Quadro K620 (6,282, delta 0.8%), and two cards at 6,389 — the NVIDIA GeForce GTX 460 SE and the NVIDIA GeForce GTX 580M, both with a delta of -0.9%. None of these rivals approach the WX 4100’s OpenCL score of 17,642. The WX 4100 also has additional benchmark results that paint a fuller picture: it scores 21,018 in Geekbench Metal, 18,703 in Geekbench Vulkan, and 3,699 in Passmark G3D. Its Passmark compute score is 1,475, and it manages 646 in Passmark G2D. The GTX 670M has no such breadth of data — only the single OpenCL result. This lack of data is itself telling: the GTX 670M is not a part that was subjected to modern benchmark suites. The WX 4100’s dominance in OpenCL is corroborated by its pixel rate of 19.22 GPixel/s and texture rate of 76.86 GTexel/s, versus the GTX 670M’s 8.372 GPixel/s and 33.49 GTexel/s. Every measurable compute metric favors the AMD card.

Specification Differences

The specification sheets diverge on nearly every meaningful field. The process node is a major differentiator: the GTX 670M is 40 nm, the WX 4100 is 14 nm. Transistor counts are 1,950 million for NVIDIA and 3,000 million for AMD. Die sizes are 332 mm² and 123 mm², respectively. The memory subsystems differ substantially. The GTX 670M has 1,536 MB of GDDR5 on a 192-bit bus, yielding 72.00 GB/s of bandwidth. The WX 4100 has 4 GB of GDDR5 on a 128-bit bus, but achieves 96.00 GB/s due to its higher memory clock of 1,500 MHz (6 Gbps effective) versus the GTX 670M’s 750 MHz (3 Gbps effective). The WX 4100 also has a base clock of 1,125 MHz and a boost clock of 1,201 MHz; the GTX 670M lists no base or boost clocks, only its memory speed. The shading unit count is 1,024 for AMD versus 336 for NVIDIA. TMUs are 64 against 56, and ROPs are 16 against 24. The FP32 throughput is 2.460 TFLOPS for the WX 4100 versus 803.7 GFLOPS for the GTX 670M. The WX 4100 also lists FP16 at 2.460 TFLOPS with a 1:1 ratio; the GTX 670M has no FP16 data. Power consumption favors the AMD card despite its higher performance: the WX 4100 has a TDP of 50 W, while the GTX 670M is rated at 75 W. The WX 4100 is a single-slot card measuring 168 mm (6.6 inches) in length and 69 mm (2.7 inches) in height, with a suggested PSU of 250 W. The GTX 670M is an MXM module with no length or height listed, and no suggested PSU. The bus interfaces are PCIe 3.0 x8 for the WX 4100 versus MXM-B (3.0) for the GTX 670M. Display outputs are 4x mini-DisplayPort 1.4a for the WX 4100, versus “Portable Device Dependent” for the GTX 670M — a laptop part. API support is another clear win for AMD: the WX 4100 supports DirectX 12 (12_0), OpenGL 4.6, and Vulkan 1.3. The GTX 670M supports DirectX 12 (11_0) and OpenGL 4.6, but Vulkan is not listed. Neither card has ray tracing or tensor cores. The WX 4100 has a launch MSRP of 399 USD; the GTX 670M has no listed MSRP.

FAQ

Q: Which GPU is faster in OpenCL compute?

A: The AMD Radeon Pro WX 4100 is dramatically faster, scoring 17,642 in Geekbench OpenCL versus the NVIDIA GeForce GTX 670M’s 6,513. The delta is -63.1%, indicating the GTX 670M is about 63% slower.

Q: How do the two cards compare in overall benchmark percentile?

A: The GTX 670M sits at the 38th percentile of all GPUs, while the WX 4100 sits at the 37th percentile. Despite the WX 4100’s much higher OpenCL score, its average benchmark score of 6,330 is lower than the GTX 670M’s 6,513 due to additional legacy tests in its suite.

Q: Which card has better memory bandwidth?

A: The AMD Radeon Pro WX 4100 has 96.00 GB/s of bandwidth from 4 GB of GDDR5 on a 128-bit bus. The GTX 670M has 72.00 GB/s from 1,536 MB of GDDR5 on a 192-bit bus, despite its wider interface.

Q: What is the power consumption difference?

A: The WX 4100 has a TDP of 50 W, while the GTX 670M is rated at 75 W. The AMD card delivers higher performance at lower power, reflecting its 14 nm process advantage over the 40 nm Fermi part.

Q: Which card supports more modern APIs?

A: The WX 4100 supports DirectX 12 (12_0), OpenGL 4.6, and Vulkan 1.3. The GTX 670M supports DirectX 12 (11_0) and OpenGL 4.6, but Vulkan is not listed in its specifications.

Q: What are the physical form factors?

A: The WX 4100 is a single-slot card with a length of 168 mm (6.6 inches) and height of 69 mm (2.7 inches), using a PCIe 3.0 x8 interface. The GTX 670M is an MXM module with an MXM-B (3.0) interface, designed for portable devices with no fixed dimensions listed.