NVIDIA GeForce GTX 670M vs NVIDIA GeForce GTX 670MX Comparison

NVIDIA GeForce GTX 670M and NVIDIA GeForce GTX 670MX are two mobile graphics processors from the same GeForce 600M generation, but they target different performance tiers. The recorded data shows the GTX 670M edges out the GTX 670MX in the available OpenCL benchmark, while the GTX 670MX counters with a larger memory pool and a more modern architecture. This analysis breaks down the benchmark results, architectural differences, and specification gaps between the two.

Where Each One Wins

The GTX 670M takes the sole head-to-head benchmark victory. In the Geekbench OpenCL test, the GTX 670M scores 6513 points, while the GTX 670MX scores 6125 points. That gives the GTX 670M a 6.3% lead in raw compute performance as measured by OpenCL. The GTX 670M also holds a higher average benchmark score of 6513, compared to the GTX 670MX's 5721 average, which reflects the fact that the GTX 670MX also has a Vulkan score dragging its average down. The GTX 670M's percentile ranking among all GPUs is 38, while the GTX 670MX sits at 33, meaning the GTX 670M outperforms a larger share of the database's GPU population.

The GTX 670MX does not win any direct benchmark in the head-to-head comparison, but it does have a clear advantage in memory capacity. The GTX 670MX ships with 3 GB of GDDR5 memory, double the 1536 MB found on the GTX 670M. For workloads that are memory-capacity bound, such as larger textures or higher resolution assets, the GTX 670MX holds an edge. The GTX 670MX also delivers a higher pixel rate and texture rate, which points to better fill-rate performance in certain rendering scenarios, even if the OpenCL score is lower.

Architecture Differences

The two GPUs come from different NVIDIA architectures. The GTX 670M uses the GF114 chip built on the Fermi 2.0 architecture, manufactured on a 40 nm process at TSMC. The GTX 670MX uses the GK104 chip based on the Kepler architecture, also from TSMC, but on a smaller 28 nm node. The process shrink allows the GTX 670MX to pack significantly more transistors: 3,540 million versus 1,950 million on the GTX 670M. Despite the higher transistor count, the GTX 670MX has a smaller die size at 294 mm² compared to 332 mm² for the GTX 670M. Transistor density tells the story: the GTX 670MX reaches 12.0M transistors per mm², while the GTX 670M only manages 5.9M per mm².

The shader configuration changes dramatically between the two. The GTX 670MX features 960 shading units, nearly three times the 336 shading units on the GTX 670M. Texture mapping units also increase, from 56 on the GTX 670M to 80 on the GTX 670MX, while the render output units remain identical at 24. The clock speeds differ as well: the GTX 670MX runs at a base and boost clock of 601 MHz, while the GTX 670M's base and boost clocks are not listed in the database. The GTX 670MX's memory runs at 700 MHz (2.8 Gbps effective), slightly lower than the GTX 670M's 750 MHz (3 Gbps effective), which explains why the GTX 670MX has lower memory bandwidth at 67.20 GB/s versus 72.00 GB/s.

The bus interface also differs. The GTX 670M uses an MXM-B (3.0) connector, while the GTX 670MX uses PCIe 3.0 x16. The GTX 670MX supports Vulkan 1.2.175 in its API list, while the GTX 670M has no Vulkan support listed. Both support DirectX 12 (11_0) and OpenGL 4.6. The GTX 670MX also has a slightly later release date, coming out in late September 2012, while the GTX 670M was released in March 2012.

Head-to-Head Benchmarks

The only direct benchmark comparison in the database is the Geekbench OpenCL test. The GTX 670M scores 6513, and the GTX 670MX scores 6125, giving the GTX 670M a 6.3% advantage. This is a meaningful gap in compute performance, and it aligns with the GTX 670M's higher average benchmark score and better percentile rank. The GTX 670MX's additional Vulkan score of 5316 is not part of the head-to-head comparison, but it shows that the GTX 670MX can run Vulkan workloads, which the GTX 670M cannot.

Looking at the nearest rivals for each GPU provides context. The GTX 670M sits close to the NVIDIA GeForce GT 555M, which scores 6493, a 0.3% difference. It also edges out the NVIDIA Quadro M5000M (6481, 0.5% ahead) and the AMD Radeon Vega 10 Mobile (6476, 0.6% ahead), while trailing the Intel UHD Graphics P750 (6554, 0.6% behind). The GTX 670MX, by contrast, is nearly tied with the Intel Iris Pro Graphics P6300 (5712, 0.2% ahead), the NVIDIA GeForce GTX 550 Ti (5731, 0.2% behind), and the AMD Radeon HD 8790M (5691, 0.5% ahead), with a larger 2.1% lead over the NVIDIA Quadro M500M (5604). The rival groups show that the GTX 670M competes in a slightly higher performance band than the GTX 670MX, which aligns with their average scores.

FAQ

Q: Which GPU has the higher OpenCL benchmark score?

A: The GTX 670M scores 6513, while the GTX 670MX scores 6125, giving the GTX 670M a 6.3% lead.

Q: Does the GTX 670MX have more memory than the GTX 670M?

A: Yes, the GTX 670MX has 3 GB of GDDR5 memory, while the GTX 670M has 1536 MB.

Q: What are the transistor counts for each GPU?

A: The GTX 670M has 1,950 million transistors, and the GTX 670MX has 3,540 million transistors.

Q: Which GPU supports Vulkan?

A: Only the GTX 670MX lists Vulkan support (version 1.2.175) in the database. The GTX 670M has no Vulkan entry.

Q: How do the pixel rates compare?

A: The GTX 670MX has a pixel rate of 12.02 GPixel/s, compared to 8.372 GPixel/s for the GTX 670M.

Q: What is the memory bandwidth difference?

A: The GTX 670M has a bandwidth of 72.00 GB/s, while the GTX 670MX has 67.20 GB/s, a difference of about 4.8 GB/s.

Specification Differences

The two GPUs differ across nearly every major specification. The process node is 40 nm for the GTX 670M and 28 nm for the GTX 670MX. The chip is GF114 (Fermi 2.0) for the GTX 670M and GK104 (Kepler) for the GTX 670MX. Transistor count jumps from 1,950 million to 3,540 million, while die size shrinks from 332 mm² to 294 mm². Transistor density goes from 5.9M per mm² to 12.0M per mm². The GTX 670MX has a base and boost clock of 601 MHz, while the GTX 670M lists no base or boost clock. Memory clocks are 750 MHz (3 Gbps effective) for the GTX 670M and 700 MHz (2.8 Gbps effective) for the GTX 670MX. Memory size doubles from 1536 MB to 3 GB, but memory bandwidth drops from 72.00 GB/s to 67.20 GB/s due to the lower clock. Shading units increase from 336 to 960, TMUs from 56 to 80, and ROPs remain at 24. Pixel rate rises from 8.372 GPixel/s to 12.02 GPixel/s, and texture rate rises from 33.49 GTexel/s to 48.08 GTexel/s. FP32 performance goes from 803.7 GFLOPS to 1,153.9 GFLOPS. The bus interface changes from MXM-B (3.0) to PCIe 3.0 x16. Vulkan support is absent on the GTX 670M but present at version 1.2.175 on the GTX 670MX. Release dates differ by about six months, with the GTX 670M launching in March 2012 and the GTX 670MX in September 2012.

The Verdict

The data indicates that the GTX 670M wins on compute performance, as shown by its higher OpenCL score, higher average benchmark score, and better percentile rank. For users running OpenCL-based workloads or applications that rely on general compute throughput, the GTX 670M is the stronger choice. Its higher memory bandwidth also helps in bandwidth-sensitive tasks, despite the smaller memory pool.

The GTX 670MX, on the other hand, offers double the memory capacity at 3 GB, which is beneficial for workloads that exceed 1536 MB of video memory. It also delivers higher pixel and texture rates, suggesting better fill-rate performance in graphics rendering. The Kepler architecture brings a smaller process node, more shading units, and Vulkan API support, which the GTX 670M lacks. For users prioritizing memory capacity or needing Vulkan compatibility, the GTX 670MX is the appropriate pick, even though its raw OpenCL score trails by 6.3%.