AMD Radeon Pro WX 2100 vs NVIDIA GeForce GTX 660 Comparison

Head-to-Head Benchmarks

The benchmark data presents a clear picture: the NVIDIA GeForce GTX 660 wins both head-to-head comparisons against the AMD Radeon Pro WX 2100, though the margins differ substantially between workloads. In the Geekbench OpenCL test, the GTX 660 scores 11,347 against the WX 2100’s 8,536, a delta of -24.8% from the AMD card’s perspective. That is a commanding lead of roughly a quarter, indicating a significant raw compute advantage for the older NVIDIA part in this particular API.

The Vulkan results tell a different story in terms of magnitude. Here the GTX 660 scores 11,415 while the WX 2100 reaches 10,770, a delta of only -5.7%. The gap narrows dramatically, suggesting that the AMD card’s architectural efficiency in Vulkan helps it close much of the distance opened in OpenCL. Yet the GTX 660 still holds the top spot, meaning that across both available cross-comparison tests, NVIDIA wins two and AMD wins zero.

Looking at each card’s broader benchmark profile, the WX 2100’s average benchmark score is 9,653, while the GTX 660’s average is 9,022. This is an interesting inversion: despite losing both head-to-head tests, the AMD card has a higher average score when all of its benchmark results are considered. The WX 2100 achieves this with an OpenCL score of 8,536 and a Vulkan score of 10,770, while the GTX 660’s profile includes an additional Metal result of 4,305 alongside its OpenCL and Vulkan figures. The GTX 660’s Metal score drags its average down considerably, which explains why its average sits below the WX 2100 despite winning the shared tests.

Percentile rankings place both cards in similar territory. The WX 2100 sits at the 46th percentile among all GPUs, while the GTX 660 is at the 45th percentile. These are nearly identical positions, reinforcing that the two cards occupy the same performance tier even if individual tests favor one or the other. The nearest rivals for each card further contextualize their standing: the WX 2100’s closest competitor is the NVIDIA Quadro P4000 with an average score of 9,665 and a delta of -0.1%, while the GTX 660’s nearest rival is the NVIDIA TITAN V CEO Edition at 9,037 with a -0.2% delta.

Where Each One Wins

The GTX 660’s wins are straightforward: it takes the OpenCL test by a wide margin and the Vulkan test by a narrower one. OpenCL is the more decisive victory, with a 24.8% lead that speaks to the NVIDIA card’s higher peak compute throughput. The GTX 660’s Vulkan win, while smaller, still demonstrates that it retains an edge even in a modern, low-overhead API. For users running OpenCL-heavy workloads—scientific computing, some rendering pipelines, or general compute tasks—the data clearly favors the GTX 660.

The WX 2100’s strengths are more subtle and emerge from the aggregate data rather than head-to-head wins. Its higher average benchmark score of 9,653 versus 9,022 indicates that its performance is more consistent across the tests it supports. The Vulkan result of 10,770 is within 5.7% of the GTX 660, showing that the AMD card is competitive in that API despite its lower raw compute specifications. Furthermore, the WX 2100’s lack of a Metal benchmark means its average is not penalized by a low-scoring test in the way the GTX 660’s average is.

The win distribution is entirely one-sided: 0 wins for the WX 2100 and 2 for the GTX 660. However, the percentile standings suggest that neither card dominates the other in a global sense. The WX 2100 edges out the GTX 660 by one percentile point, which aligns with the average score difference. This indicates that the WX 2100’s overall performance profile is slightly more balanced, even if the GTX 660 wins the specific tests where they directly compete.

In terms of architectural traits that influence these outcomes, the GTX 660’s 960 shading units and 80 texture mapping units provide substantial parallel throughput, which likely explains its OpenCL dominance. The WX 2100 counters with a newer process node and higher clock efficiency, which may account for its more respectable Vulkan showing. The GTX 660’s memory bandwidth of 144.2 GB/s dwarfs the WX 2100’s 48.00 GB/s, a factor that could contribute to its compute lead in memory-intensive workloads.

The Verdict

The data is unambiguous on the head-to-head front: the NVIDIA GeForce GTX 660 outperforms the AMD Radeon Pro WX 2100 in both OpenCL and Vulkan benchmarks. Anyone prioritizing raw compute performance in those APIs should favor the GTX 660, particularly for OpenCL tasks where the 24.8% advantage is substantial. The GTX 660’s Vulkan lead of 5.7% is smaller but still consistent, making it the safer choice for users who expect to work across both APIs.

However, the WX 2100 is not without merit. Its average benchmark score is higher, and its percentile ranking is one point above the GTX 660. This suggests that in a broader set of tests—beyond the two shared benchmarks—the AMD card may perform more evenly. The WX 2100’s Vulkan score of 10,770 is within striking distance of the GTX 660, and its higher average indicates that it does not suffer from the same low-scoring outlier that the GTX 660’s Metal result represents.

For workstation users who value efficiency and a modern feature set, the WX 2100’s architectural advantages are notable. It uses a 14 nm process versus the GTX 660’s 28 nm, and it supports DirectX 12_0 and Vulkan 1.3, compared to the GTX 660’s DirectX 12 (11_0) and Vulkan 1.2.175. The WX 2100 also requires no power connectors and has a 35 W TDP against the GTX 660’s 140 W, making it far easier to integrate into compact or power-constrained systems.

The verdict depends on the use case. The GTX 660 wins every direct comparison in the data, so it is the performance pick for OpenCL and Vulkan workloads. The WX 2100 offers a more balanced overall profile, a newer feature set, and significantly lower power requirements, making it the better choice for users who prioritize those factors over raw benchmark scores. The data does not support calling either card a clear overall winner; it supports calling the GTX 660 the winner in specific tests and the WX 2100 the winner in overall balance.

FAQ

Q: Which card wins the Geekbench OpenCL benchmark?

A: The NVIDIA GeForce GTX 660 wins, scoring 11,347 against the AMD Radeon Pro WX 2100’s 8,536, a delta of -24.8% for the AMD card.

Q: How close is the Vulkan performance between the two cards?

A: The GTX 660 scores 11,415 while the WX 2100 scores 10,770, giving the NVIDIA card a 5.7% lead. This is a much narrower margin than the OpenCL gap.

Q: What are the average benchmark scores for each card?

A: The WX 2100 has an average benchmark score of 9,653, while the GTX 660 has an average of 9,022. The AMD card’s average is higher despite losing both head-to-head tests.

Q: How do the cards compare in terms of overall GPU percentile rankings?

A: The WX 2100 sits at the 46th percentile among all GPUs, while the GTX 660 is at the 45th percentile. The positions are nearly identical.

Q: What is the closest rival to each card based on average score?

A: The WX 2100’s nearest rival is the NVIDIA Quadro P4000 with an average score of 9,665 and a delta of -0.1%. The GTX 660’s nearest rival is the NVIDIA TITAN V CEO Edition at 9,037 with a delta of -0.2%.

Q: Does the GTX 660 have any benchmark results that the WX 2100 lacks?

A: Yes, the GTX 660 includes a Geekbench Metal score of 4,305. The WX 2100 has no Metal benchmark result in the data.

Architecture Differences

The two cards represent distinct architectural generations and design philosophies. The AMD Radeon Pro WX 2100 is built on the Lexa chip using the GCN 4.0 architecture, manufactured on a 14 nm process at GlobalFoundries. The NVIDIA GeForce GTX 660 uses the GK106 chip with the Kepler architecture, produced on a 28 nm process at TSMC. The process node difference is significant: 14 nm versus 28 nm, which contributes to the AMD card’s lower power draw and higher transistor density of 21.4M per mm², compared to the GTX 660’s 11.5M per mm².

Transistor counts are comparable in absolute terms. The WX 2100 contains 2,200 million transistors on a 103 mm² die, while the GTX 660 packs 2,540 million transistors onto a 221 mm² die. The GTX 660’s die is more than twice as large, a direct consequence of the older process node. The WX 2100’s smaller die and higher density reflect the advantages of newer manufacturing technology.

Memory configurations diverge sharply. Both cards feature 2 GB of GDDR5 memory, but the bus widths and resulting bandwidth differ substantially. The WX 2100 uses a 64-bit bus delivering 48.00 GB/s, while the GTX 660 uses a 192-bit bus delivering 144.2 GB/s. This three-fold bandwidth advantage for the NVIDIA card is a major architectural differentiator and likely a key factor in its OpenCL performance lead.

Compute resources also favor the GTX 660 in raw counts. The NVIDIA card has 960 shading units, 80 TMUs, and 24 ROPs, while the WX 2100 has 512 shading units, 32 TMUs, and 16 ROPs. The GTX 660’s texture rate is 82.56 GTexel/s versus the WX 2100’s 39.01 GTexel/s, and its pixel rate is 20.64 GPixel/s versus 19.50 GPixel/s. FP32 performance follows the same pattern: the GTX 660 delivers 1.981 TFLOPS against the WX 2100’s 1,248.3 GFLOPS. The AMD card does support FP16 at a 1:1 ratio, while the GTX 660 has no listed FP16 capability.

Clock speeds and power profiles tell the efficiency story. The WX 2100 has a base clock of 925 MHz and a boost of 1219 MHz, while the GTX 660 runs at 980 MHz base and 1032 MHz boost. Despite lower clocks, the GTX 660’s higher core count gives it the compute edge. The WX 2100’s TDP is 35 W, requiring no power connectors and a 200 W suggested PSU, while the GTX 660 draws 140 W, needs a single 6-pin connector, and recommends a 300 W PSU. The WX 2100 is single-slot and 168 mm long, whereas the GTX 660 is dual-slot and 241 mm long.

API support and display outputs also differ. The WX 2100 supports DirectX 12 (12_0), OpenGL 4.6, and Vulkan 1.3, while the GTX 660 supports DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.2.175. The AMD card offers one DisplayPort 1.4a and two mini-DisplayPort 1.4a outputs, while the NVIDIA card provides two DVI, one HDMI 1.4a, and one DisplayPort 1.2. The WX 2100 uses a PCIe 3.0 x8 interface, while the GTX 660 uses PCIe 3.0 x16. Both cards are end-of-life, with the WX 2100 releasing in 2017 and the GTX 660 in 2012.