NVIDIA GeForce GTX 670MX vs NVIDIA RTX PRO 5000 72 GB Blackwell Comparison

Head-to-Head Benchmarks

The recorded data for these two GPUs is sparse and not directly comparable, as they were tested under different benchmark suites. The RTX PRO 5000 72 GB Blackwell was evaluated with 3DMark Steel Nomad (DX12) and PassMark tests, while the GTX 670MX was measured with Geekbench OpenCL and Vulkan. No shared test exists in the database, so a direct score-for-score comparison is impossible. However, the average benchmark scores provide a meaningful baseline: the RTX PRO 5000 sits at 6407, while the GTX 670MX averages 5721. This represents a 12% advantage for the newer card in aggregate performance.

Looking at the RTX PRO 5000's individual results, the 3DMark Steel Nomad DX12 score of 9579.5 indicates strong modern API performance, while its PassMark G3D score of 24310 is particularly dominant. The PassMark GPU Compute result of 15667 shows substantial compute throughput. In contrast, the GTX 670MX's Geekbench OpenCL score of 6125 and Vulkan score of 5316 are respectable for its era but fall far short of the RTX PRO 5000's capabilities in absolute terms. The RTX PRO 5000 also records a PassMark DirectX 9 score of 311 and DirectX 11 score of 219, while its DirectX 10 and 12 scores are 175 and 78 respectively, showing that legacy API performance is not its focus.

The percentile rankings place the RTX PRO 5000 in the 37th percentile of all GPUs, while the GTX 670MX sits in the 33rd percentile. This narrow percentile gap is surprising given the massive architectural leap, but it reflects the fact that the RTX PRO 5000's benchmark suite includes heavy modern workloads that penalize older cards less than one might expect. The RTX PRO 5000's nearest rivals in the database include the GeForce GTX 460 SE and GTX 580M, both with an average score of 6389 (0.3% lower), and the Radeon Vega 10 Mobile and Quadro M5000M at 6476 and 6481 (1.1% higher). The GTX 670MX's nearest rivals are the Intel Iris Pro Graphics P6300 at 5712 (0.2% lower) and the GeForce GTX 550 Ti at 5731 (0.2% higher), placing it in far older company.

The key takeaway from the benchmark data is that the RTX PRO 5000 delivers roughly 12% higher average performance than the GTX 670MX, but the real story lies elsewhere. The RTX PRO 5000's raw scores, particularly the 24310 PassMark G3D result, dwarf the GTX 670MX's capabilities, yet both cards occupy similar percentile territory because the database's comparison pool is heavily weighted toward modern GPUs. The GTX 670MX, by contrast, holds its own against contemporaries like the GTX 550 Ti, but its 5721 average is firmly in the low-to-mid range of historical hardware.

FAQ

Q: Which GPU has the higher average benchmark score?

A: The NVIDIA RTX PRO 5000 72 GB Blackwell averages 6407, which is 12% higher than the GeForce GTX 670MX's 5721 average.

Q: How does the RTX PRO 5000 compare to its nearest rivals?

A: The RTX PRO 5000 is essentially tied with the GeForce GTX 460 SE and GTX 580M (both 6389, within 0.3%), while the AMD Radeon Vega 10 Mobile and Quadro M5000M are 1.1% higher at 6476 and 6481.

Q: What is the GTX 670MX's closest competitor in the database?

A: The Intel Iris Pro Graphics P6300 is 0.2% lower at 5712, and the GeForce GTX 550 Ti is 0.2% higher at 5731, making them nearly identical in average score.

Q: Does the GTX 670MX support ray tracing or tensor cores?

A: No, the GTX 670MX has null values for both RT cores and tensor cores, meaning it lacks these features entirely.

Q: What is the RTX PRO 5000's best individual benchmark result?

A: The highest recorded score for the RTX PRO 5000 is 24310 in PassMark G3D, followed by 15667 in PassMark GPU Compute.

Q: How do the two GPUs compare in transistor density?

A: The RTX PRO 5000 has a transistor density of 122.9 million per square millimeter, over ten times higher than the GTX 670MX's 12.0 million per square millimeter.

Architecture Differences

The architectural gap between these two GPUs is generational and profound. The RTX PRO 5000 uses the GB202 chip built on Blackwell 2.0 architecture, fabricated on a 5 nm process at TSMC. The GTX 670MX uses the GK104 chip with Kepler architecture, manufactured on a 28 nm process, also at TSMC. The process node difference alone, 5 nm versus 28 nm, explains much of the performance and efficiency gap.

Transistor counts tell a stark story: the RTX PRO 5000 packs 92,200 million transistors onto a 750 mm² die, while the GTX 670MX has just 3,540 million transistors on a 294 mm² die. This works out to a transistor density of 122.9 million per square millimeter for the newer card versus 12.0 million for the older one. The Blackwell chip crams nearly 26 times more transistors into roughly 2.5 times the die area.

The RTX PRO 5000 features 110 RT cores and 440 tensor cores, dedicated hardware that the GTX 670MX completely lacks. The GTX 670MX's shading units number 960, while the RTX PRO 5000 has 14,080, a 14.7-fold increase. Texture mapping units jump from 80 to 440, and render output units from 24 to 160. The RTX PRO 5000's pixel rate is 380.3 GPixel/s versus 12.02 GPixel/s for the GTX 670MX, and its texture rate is 1,045.9 GTexel/s versus 48.08 GTexel/s. Floating point performance scales from 1,153.9 GFLOPS to 66.94 TFLOPS, a 58-fold increase, with the RTX PRO 5000 offering equal FP16 and FP32 throughput at 66.94 TFLOPS each.

Memory architecture also diverges completely. The RTX PRO 5000 uses 72 GB of GDDR7 on a 384-bit bus, delivering 1.34 TB/s of bandwidth. The GTX 670MX has 3 GB of GDDR5 on a 192-bit bus, providing 67.20 GB/s. This is a 20-fold bandwidth advantage for the newer card. API support differs as well: the RTX PRO 5000 supports DirectX 12 Ultimate (12_2), Vulkan 1.4, and OpenGL 4.6, while the GTX 670MX is limited to DirectX 12 (11_0), Vulkan 1.2.175, and OpenGL 4.6.

Specification Differences

The two GPUs differ across nearly every measurable specification. The RTX PRO 5000 has a base clock of 1740 MHz and boost clock of 2377 MHz, while the GTX 670MX runs at a flat 601 MHz for both base and boost. Memory clocks are 1750 MHz (28 Gbps effective) for the RTX PRO 5000 versus 700 MHz (2.8 Gbps effective) for the GTX 670MX.

Memory capacity is 72 GB versus 3 GB, with GDDR7 replacing GDDR5. Bus width is 384-bit versus 192-bit. The RTX PRO 5000 has a TDP of 300 W, while the GTX 670MX draws just 75 W. The newer card requires a 16-pin power connector and a 700 W suggested PSU, whereas the GTX 670MX has no power connectors and no suggested PSU listed. The RTX PRO 5000 is a dual-slot card measuring 267 mm in length, 111 mm in height, and 40 mm in width, while the GTX 670MX has no listed dimensions.

Bus interface differs from PCIe 5.0 x16 on the RTX PRO 5000 to PCIe 3.0 x16 on the GTX 670MX. Display outputs are 4x DisplayPort 2.1b versus "Portable Device Dependent" for the mobile-oriented GTX 670MX. Production status is Active for the RTX PRO 5000 and End-of-life for the GTX 670MX. The release dates span 13 years: October 2025 versus September 2012. The RTX PRO 5000's predecessor is Workstation Ada, while the GTX 670MX's predecessor is GeForce 500M and its successor is GeForce 700M.

Where Each One Wins

The RTX PRO 5000 wins decisively in every modern workload category. Its 3DMark Steel Nomad DX12 score of 9579.5 demonstrates strong performance in current-generation game engines and DirectX 12 titles. The PassMark G3D result of 24310 indicates excellent overall 3D rendering capability, and the PassMark GPU Compute score of 15667 shows robust compute performance for tasks like scientific simulation, AI inference, and content creation. The 72 GB of GDDR7 memory with 1.34 TB/s bandwidth makes it suitable for massive datasets, high-resolution textures, and multi-display professional environments. Its 440 tensor cores and 110 RT cores enable hardware-accelerated ray tracing and AI features that the GTX 670MX cannot execute at all.

The GTX 670MX's wins are narrower and rooted in its historical context. Its Geekbench OpenCL score of 6125 and Vulkan score of 5316 are the only recorded benchmarks, and they show that the card remains functional for basic compute and modern API workloads despite its age. Its 75 W TDP makes it far more power-efficient in absolute terms, requiring no external power connectors and no PSU upgrade. For legacy applications that rely on older driver paths or DirectX 11 (11_0) support, the GTX 670MX remains compatible, though its 960 shading units and 67.20 GB/s bandwidth severely limit its usefulness in any demanding scenario. The card's 3 GB GDDR5 memory is sufficient for 2012-era games but insufficient for modern titles.

The percentile data reinforces this split: the RTX PRO 5000 sits at the 37th percentile of all GPUs, while the GTX 670MX sits at the 33rd. In relative terms, the RTX PRO 5000 is closer to the top of the database than the GTX 670MX, but the gap is smaller than the architectural differences suggest. This is because the GTX 670MX's benchmark pool includes only two tests, both of which are compute-oriented, while the RTX PRO 5000's eight tests include rasterization and legacy API workloads that drag its average down.

The Verdict

The data clearly favors the NVIDIA RTX PRO 5000 72 GB Blackwell for any modern use case. Its 66.94 TFLOPS of FP32 compute, 1.34 TB/s memory bandwidth, and 72 GB of GDDR7 memory place it in a completely different performance class than the GTX 670MX. The 12% advantage in average benchmark score understates the real-world gap, as the RTX PRO 5000's individual scores, particularly the 24310 PassMark G3D result, are orders of magnitude beyond what the older card can achieve. The RTX PRO 5000 is the only choice for professionals requiring ray tracing, tensor core acceleration, or large memory pools.

The GTX 670MX, with its 5721 average score and 33rd percentile ranking, is a historical artifact. It matches the performance of the Intel Iris Pro Graphics P6300 and GeForce GTX 550 Ti, both from the same era, but it cannot compete with any modern GPU. Its 75 W TDP and lack of power connectors make it easy to install in legacy systems, and its Geekbench scores of 6125 and 5316 show it still functions for basic tasks. However, its 3 GB memory capacity and 67.20 GB/s bandwidth are insurmountable limitations for anything beyond casual use.

The verdict is straightforward: choose the RTX PRO 5000 for any serious workload, from professional visualization to AI development. The GTX 670MX is only relevant for retro computing, compatibility testing, or low-power builds where its 75 W TDP is an asset. The database shows a 12% average score difference, but the architectural chasm, 92,200 million transistors versus 3,540 million, tells the real story. The RTX PRO 5000 is a current-generation powerhouse; the GTX 670MX is a legacy part that has reached the end of its useful life.