AMD Radeon Pro 5500M vs NVIDIA GeForce GTX 670 Comparison

Head-to-Head Benchmarks

The benchmark data presents a decisive picture: the AMD Radeon Pro 5500M wins every recorded head-to-head test against the NVIDIA GeForce GTX 670. The database records three direct comparisons, all in compute and graphics API workloads, and the AMD card dominates each one by a substantial margin.

In Geekbench Metal, the AMD Radeon Pro 5500M scores 38,235 against the GTX 670's 7,424. That is a delta of -80.6% from the AMD card's perspective, meaning the NVIDIA card trails by roughly four-fifths of the AMD score. This is the largest gap of the three tests, and it reflects the architectural gulf between the two products, not just a clock speed difference.

The Geekbench OpenCL test narrows the gap somewhat but still favors AMD heavily. The Radeon Pro 5500M posts 31,088, while the GTX 670 manages 15,341. The delta here is -50.7%, so the NVIDIA card delivers approximately half the score of its rival. OpenCL is a more universal compute workload, and the AMD card's lead, while still massive, is less extreme than in Metal.

Geekbench Vulkan shows a similar pattern. The AMD card scores 35,134, and the GTX 670 scores 15,553, a delta of -55.7%. Vulkan is a low-level graphics API, and the Radeon Pro 5500M's RDNA architecture handles it far more efficiently than the older Kepler design.

The overall win tally is 3-0 in favor of the AMD Radeon Pro 5500M. The GTX 670 records zero wins in any head-to-head comparison. The average benchmark score in the database tells the same story: the AMD card averages 11,528 points across all recorded tests, while the NVIDIA card averages 12,773. That is a counterintuitive result, because the head-to-head tests favor AMD so strongly, but the average includes a broader set of workloads where the GTX 670's older driver support or specific optimizations may play a role.

Looking at the nearest rival data provides context for each card's standing. The GTX 670 sits at the 52nd percentile of all GPUs, with an average score of 12,773. Its closest rivals are the NVIDIA GeForce GTX 590 (12,830, -0.4%), the AMD Radeon Pro 455 (12,831, -0.4%), the AMD Radeon RX 7600M XT (12,710, 0.5%), and the AMD FirePro W5100 (12,847, -0.6%). These deltas are all within one percentage point, which indicates the GTX 670 is firmly in a cluster of mid-range cards from its era.

The Radeon Pro 5500M sits at the 51st percentile, with an average score of 11,528. Its nearest rivals include the NVIDIA Tesla K20c (11,479, 0.4%), the AMD Radeon RX 7800 XT (11,627, -0.8%), the NVIDIA GeForce GTX 1660 (11,680, -1.3%), and the NVIDIA GeForce GTX 780M (11,261, 2.4%). The deltas here are also small, but the AMD card's average is lower than the GTX 670's, despite winning every direct comparison. This discrepancy highlights how the two cards excel in different test suites.

Where Each One Wins

The AMD Radeon Pro 5500M wins every recorded head-to-head benchmark, so on paper it is the clear performance leader. But the broader benchmark set in the database reveals a more nuanced picture. The GTX 670's average score across all benchmarks is higher, which suggests it performs better in certain legacy or driver-specific workloads that are not part of the direct comparison suite.

The Radeon Pro 5500M's wins are concentrated in modern, cross-platform compute and graphics APIs. Its Geekbench Metal score of 38,235 is exceptional, and this is likely due to its RDNA 1.0 architecture being designed with contemporary API efficiency in mind. The Vulkan score of 35,134 reinforces this, as does the OpenCL score of 31,088. For workloads that leverage these APIs, the AMD card is dramatically faster.

The GTX 670's strength lies in its consistency across a wider range of older test scenarios. Its average benchmark score of 12,773 is higher than the Radeon Pro 5500M's 11,528, and it sits at a slightly higher percentile (52 versus 51). The GTX 670's nearest rivals all score within 0.6% of its average, which indicates it is well-optimized for the specific tests that contribute to that average. The Radeon Pro 5500M's nearest rivals show more variance, with deltas ranging from -1.3% to +2.4%, suggesting its performance is more workload-dependent.

In practical terms, the Radeon Pro 5500M is the better choice for any application that uses Metal, Vulkan, or OpenCL. The GTX 670 may hold its own in older DirectX-based tests or legacy compute workloads, where its driver maturity and architectural simplicity could be an advantage. The database does not record any head-to-head DirectX results, so any claim about the GTX 670's superiority in that area is speculative. The data only shows that its average score is higher, not where that advantage comes from.

The Verdict

The data is unambiguous: the AMD Radeon Pro 5500M wins every head-to-head benchmark by a wide margin. In Metal, it is 80.6% faster. In OpenCL, it is 50.7% faster. In Vulkan, it is 55.7% faster. For any user whose workload relies on these APIs, the choice is clear.

The GTX 670's higher average benchmark score is the only counterpoint. At 12,773, it exceeds the Radeon Pro 5500M's 11,528 by roughly 10.8%. This suggests that in the broader set of tests recorded in the database, the NVIDIA card performs relatively better. However, none of those tests are part of the head-to-head comparison, so their relevance to real-world modern workloads is uncertain.

For users prioritizing compute performance in Metal, Vulkan, or OpenCL, the AMD Radeon Pro 5500M is the superior choice. Its scores are consistently in the 31,000 to 38,000 range, which places it far above the GTX 670's 7,400 to 15,500 range. The AMD card also offers 8 GB of GDDR6 memory versus 2 GB of GDDR5, which is a significant advantage for larger datasets.

For users who rely on legacy workloads or specific driver optimizations that favor the GTX 670, the NVIDIA card may still be viable. Its average score is higher, and its nearest rivals are all within a fraction of a percent, indicating stable performance across many tests. But the head-to-head data shows that in modern API benchmarks, it is simply outclassed.

The production status of both cards is end-of-life, so neither is a new purchase recommendation. But based purely on the recorded benchmarks, the AMD Radeon Pro 5500M is the faster card in every direct comparison. The GTX 670's only advantage is its higher average score across the full benchmark suite, which is a secondary consideration given the magnitude of the head-to-head losses.

FAQ

Q: Which card wins the most head-to-head benchmarks?

A: The AMD Radeon Pro 5500M wins all three recorded head-to-head tests: Geekbench Metal, Geekbench OpenCL, and Geekbench Vulkan. The NVIDIA GeForce GTX 670 records zero wins.

Q: What is the largest performance gap between the two cards?

A: The largest gap is in Geekbench Metal, where the AMD Radeon Pro 5500M scores 38,235 versus the GTX 670's 7,424, a delta of -80.6%.

Q: How do the average benchmark scores compare?

A: The NVIDIA GeForce GTX 670 has a higher average benchmark score of 12,773, while the AMD Radeon Pro 5500M averages 11,528. The GTX 670 sits at the 52nd percentile, and the Radeon Pro 5500M at the 51st.

Q: What memory configurations do the two cards use?

A: The NVIDIA GeForce GTX 670 has 2 GB of GDDR5 memory on a 256-bit bus, with 192.3 GB/s bandwidth. The AMD Radeon Pro 5500M has 8 GB of GDDR6 memory on a 128-bit bus, with 192.0 GB/s bandwidth.

Q: Which card has higher compute throughput in FP32?

A: The AMD Radeon Pro 5500M delivers 4.454 TFLOPS of FP32 performance, while the NVIDIA GeForce GTX 670 delivers 2.634 TFLOPS. The AMD card also supports FP16 at 8.909 TFLOPS.

Q: What are the power requirements for each card?

A: The NVIDIA GeForce GTX 670 has a TDP of 170 W and requires two 6-pin power connectors, with a suggested 450 W power supply. The AMD Radeon Pro 5500M has a TDP of 85 W and uses no external power connectors.

Architecture Differences

The two cards come from different generations and foundries, which explains their divergent performance profiles. The NVIDIA GeForce GTX 670 uses the GK104 chip, built on the Kepler architecture, and was manufactured by TSMC on a 28 nm process. It contains 3,540 million transistors on a die size of 294 mm², resulting in a transistor density of 12.0 million per square millimeter.

The AMD Radeon Pro 5500M uses the Navi 14 chip, built on the RDNA 1.0 architecture, also manufactured by TSMC but on a 7 nm process. It packs 6,400 million transistors into a die size of 158 mm², which yields a transistor density of 40.5 million per square millimeter. That is more than three times the density of the Kepler chip, a direct result of the smaller process node.

The core configurations differ as well. The GTX 670 has 1,344 shading units, 112 texture mapping units, and 32 raster operations pipelines. The Radeon Pro 5500M has 1,536 shading units, 96 TMUs, and 32 ROPs. The AMD card has more shading units but fewer TMUs, which shifts its workload balance toward compute rather than texture-heavy rendering.

Clock speeds favor the AMD card. The GTX 670 runs at a base clock of 915 MHz with a boost of 980 MHz. The Radeon Pro 5500M runs at a base of 1000 MHz and boosts to 1450 MHz. That higher boost clock, combined with more shading units, explains the large FP32 advantage: 4.454 TFLOPS versus 2.634 TFLOPS.

Memory architecture is another key difference. The GTX 670 uses 2 GB of GDDR5 on a 256-bit bus, achieving 192.3 GB/s bandwidth. The Radeon Pro 5500M uses 8 GB of GDDR6 on a 128-bit bus, achieving 192.0 GB/s. The bandwidth is nearly identical, but the AMD card offers four times the capacity, which is critical for modern workloads that need large framebuffers or datasets.

The bus interfaces also differ. The GTX 670 uses PCIe 3.0 x16, while the Radeon Pro 5500M uses PCIe 4.0 x8. The newer PCIe 4.0 standard offers higher per-lane bandwidth, so the x8 link on the AMD card may provide comparable or better transfer rates than the older x16 link.

API support shows the generational gap. The GTX 670 supports DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.2.175. The Radeon Pro 5500M supports DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.4. The higher DirectX feature level and newer Vulkan version on the AMD card align with its superior Vulkan and Metal benchmark scores.

Power efficiency is a major differentiator. The GTX 670 has a TDP of 170 W and requires two 6-pin power connectors, with a suggested 450 W power supply. The Radeon Pro 5500M has a TDP of 85 W and requires no external power connectors. This makes the AMD card far more suitable for portable or power-constrained systems, which is consistent with its "Radeon Pro Mac (Navi Mobile)" generation designation.

The display outputs also reflect their intended use cases. The GTX 670 offers 2x DVI, 1x HDMI 1.4a, and 1x DisplayPort 1.2, typical for a desktop card. The Radeon Pro 5500M's outputs are listed as "Portable Device Dependent," indicating it is designed for laptops or all-in-one systems where the display connection is not user-configurable.

The release dates are separated by over seven years: the GTX 670 launched in May 2012, and the Radeon Pro 5500M in November 2019. Both are now end-of-life products, but the architectural advancements from Kepler to RDNA 1.0 are clearly visible in the benchmark results.