NVIDIA GeForce GTX 670MX vs NVIDIA GeForce GTX 760M Comparison

The NVIDIA GeForce GTX 670MX and the NVIDIA GeForce GTX 760M are two mobile graphics solutions from successive generations of the same product family. The GTX 670MX belongs to the GeForce 600M generation, released in late September 2012, while the GTX 760M is part of the GeForce 700M generation, arriving in late May 2013. Both are built on the Kepler architecture using a 28 nm process at TSMC, and both are now end-of-life products. The recorded benchmark data shows that the older GTX 670MX holds a consistent lead over its successor in every measured test, a result that runs counter to what one might expect from generational naming. This analysis walks through the database entries for both parts, comparing their architecture, specifications, and recorded performance to determine where each one stands.

FAQ

Q: Which GPU has the higher average benchmark score?

A: The NVIDIA GeForce GTX 670MX has an average benchmark score of 5721, placing it in the 33rd percentile of all GPUs. The NVIDIA GeForce GTX 760M has an average score of 5236, placing it in the 31st percentile.

Q: How much faster is the GTX 670MX in the recorded head-to-head tests?

A: The GTX 670MX wins both head-to-head tests. It scores 6125 versus 5604 in Geekbench OpenCL, a 9.3% advantage, and 5316 versus 4868 in Geekbench Vulkan, a 9.2% advantage.

Q: What are the memory configurations of the two cards?

A: The GTX 670MX has 3 GB of GDDR5 memory on a 192-bit bus, with 67.20 GB/s of bandwidth. The GTX 760M has 2 GB of GDDR5 memory on a 128-bit bus, with 64.13 GB/s of bandwidth.

Q: How do their transistor counts and die sizes compare?

A: The GTX 670MX uses the GK104 chip with 3,540 million transistors on a 294 mm² die. The GTX 760M uses the GK106S chip with 2,540 million transistors on a 221 mm² die.

Q: What is the thermal design power difference?

A: The GTX 670MX has a TDP of 75 W, while the GTX 760M has a lower TDP of 55 W. The GTX 760M is also listed with a slot width of MXM Module.

Q: Which card has better Vulkan performance relative to its OpenCL performance?

A: The GTX 670MX maintains a similar lead in both APIs. Its Vulkan score of 5316 is 9.2% ahead of the GTX 760M's 4868, while its OpenCL score of 6125 is 9.3% ahead of the 5604 recorded for the GTX 760M.

Where Each One Wins

The data is unambiguous in the two recorded benchmark tests: the GTX 670MX wins both. There is no test in the database where the GTX 760M comes out ahead. The wins for the GTX 670MX are consistent across different compute APIs, suggesting that the advantage is not specific to one workload type but reflects a general performance gap.

The GTX 670MX wins in Geekbench OpenCL with a score of 6125 against 5604, a 9.3% margin. It also wins in Geekbench Vulkan with 5316 against 4868, a 9.2% margin. These are the only two benchmark entries in the head-to-head comparison, so the GTX 670MX takes both possible wins, giving it a 2 to 0 record in this matchup.

Where the GTX 760M does have a claim is in efficiency and physical footprint. It carries a 55 W TDP compared to the 75 W TDP of the GTX 670MX, a difference of 20 W. It is also specified as an MXM Module, indicating a standardized mobile form factor, whereas the GTX 670MX has no slot width listed. For a system builder prioritizing lower power draw and a modular design, the GTX 760M holds an advantage, but this is not reflected in the compute benchmark scores.

Architecture Differences

Both GPUs are built on NVIDIA's Kepler architecture and use the same 28 nm TSMC manufacturing process, but they are based on different physical chips. The GTX 670MX uses the GK104 die, which contains 3,540 million transistors across a 294 mm² area. The GTX 760M uses the GK106S die, a smaller and less complex chip with 2,540 million transistors on a 221 mm² die. The transistor density is slightly higher on the GTX 670MX at 12.0M per mm², versus 11.5M per mm² for the GTX 760M.

The execution resources differ substantially. The GTX 670MX has 960 shading units, 80 texture mapping units, and 24 ROPs. The GTX 760M has fewer of each: 768 shading units, 64 TMUs, and 16 ROPs. This means the GTX 670MX has 192 more shading units, 16 more TMUs, and 8 more ROPs. These are direct architectural differences that explain much of the performance gap.

Clock behavior also differs. The GTX 670MX has a base clock of 601 MHz and a boost clock of 601 MHz, meaning there is no boost headroom recorded; it runs at a fixed frequency. The GTX 760M has a base clock of 628 MHz and a boost clock of 719 MHz, offering a 91 MHz boost range. Despite running at lower absolute clocks, the GTX 670MX still outperforms the GTX 760M in compute tests, which points to its wider execution resources compensating for the lower clock speed.

The memory architecture is another key divergence. The GTX 670MX uses a 192-bit memory bus with 3 GB of GDDR5, while the GTX 760M uses a narrower 128-bit bus with 2 GB of GDDR5. The effective memory speed also differs: the GTX 670MX memory runs at 2.8 Gbps effective, while the GTX 760M memory runs at 4 Gbps effective. The faster memory on the GTX 760M does not overcome the wider bus on the GTX 670MX, as the bandwidth figures show 67.20 GB/s for the GTX 670MX versus 64.13 GB/s for the GTX 760M.

Specification Differences

The two cards share several attributes: both are NVIDIA parts with Kepler architecture, both use a 28 nm TSMC process, both have PCIe 3.0 x16 interfaces, both support DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.2.175, both have no power connectors, and both have portable-device-dependent display outputs. The differences are where the contrast lies.

The GTX 670MX has a larger GK104 chip with 3,540 million transistors and a 294 mm² die size. The GTX 760M has a smaller GK106S chip with 2,540 million transistors and a 221 mm² die. Transistor density is 12.0M per mm² for the GTX 670MX and 11.5M per mm² for the GTX 760M.

Clock speeds: the GTX 670MX runs at 601 MHz base and 601 MHz boost, with memory at 700 MHz and 2.8 Gbps effective. The GTX 760M runs at 628 MHz base and 719 MHz boost, with memory at 1002 MHz and 4 Gbps effective.

Memory: the GTX 670MX has 3 GB of GDDR5 on a 192-bit bus with 67.20 GB/s bandwidth. The GTX 760M has 2 GB of GDDR5 on a 128-bit bus with 64.13 GB/s bandwidth.

Compute resources: the GTX 670MX has 960 shading units, 80 TMUs, and 24 ROPs. The GTX 760M has 768 shading units, 64 TMUs, and 16 ROPs.

Rates: the GTX 670MX has a pixel rate of 12.02 GPixel/s and a texture rate of 48.08 GTexel/s. The GTX 760M has a pixel rate of 11.50 GPixel/s and a texture rate of 46.02 GTexel/s. FP32 performance is 1,153.9 GFLOPS for the GTX 670MX and 1,104.4 GFLOPS for the GTX 760M.

Power and form factor: the GTX 670MX has a 75 W TDP, while the GTX 760M has a 55 W TDP. The GTX 760M is listed as an MXM Module, while the GTX 670MX has no slot width given. Neither card uses power connectors. Release dates differ, with the GTX 670MX releasing in September 2012 and the GTX 760M in May 2013. The GTX 670MX predecessor is GeForce 500M and its successor is GeForce 700M; the GTX 760M predecessor is GeForce 600M and its successor is GeForce 800M.

Head-to-Head Benchmarks

The recorded head-to-head data contains two tests, and the GTX 670MX wins both. In Geekbench OpenCL, the GTX 670MX scores 6125 against 5604 for the GTX 760M. The delta is 9.3%, meaning the GTX 670MX is nearly a tenth faster in this compute workload. In Geekbench Vulkan, the GTX 670MX scores 5316 against 4868, a delta of 9.2%. The margins are almost identical across the two APIs, which suggests a consistent underlying performance advantage rather than API-specific optimization.

Looking at the nearest rivals provides context. The GTX 670MX average score of 5721 puts it just ahead of the Intel Iris Pro Graphics P6300, which averages 5712, a delta of 0.2%. It is also 2.1% ahead of the NVIDIA Quadro M500M, which averages 5604, and 0.5% ahead of the AMD Radeon HD 8790M, which averages 5691. The only rival that edges it out is the NVIDIA GeForce GTX 550 Ti, which averages 5731, a delta of -0.2% for the GTX 670MX. These are very close margins, placing the GTX 670MX in a tight cluster of comparable mobile and older desktop parts.

The GTX 760M, by contrast, has an average score of 5236. Its nearest rival is the NVIDIA Quadro 4000M at 5211, a delta of 0.5% in favor of the GTX 760M. The NVIDIA GeForce 940M averages 5284, which is 0.9% higher than the GTX 760M, and the NVIDIA GeForce GTX 980M averages 5308, which is 1.3% higher. The AMD Radeon R7 M260X averages 5161, putting it 1.5% behind the GTX 760M. This places the GTX 760M in a lower performance tier than the GTX 670MX, with the latter sitting roughly 9% higher in average score.

The percentile rankings reinforce the gap. The GTX 670MX is in the 33rd percentile of all GPUs, while the GTX 760M is in the 31st percentile. While both are in the lower half of the overall GPU landscape, the GTX 670MX is ranked higher.

The fixed clock behavior of the GTX 670MX is worth noting against the variable boost of the GTX 760M. The GTX 760M can boost from 628 MHz to 719 MHz, a 14.5% increase over its base clock, while the GTX 670MX stays at 601 MHz. Even with this boost capability, the GTX 760M cannot close the gap in the recorded tests. The wider memory bus and larger shading unit count of the GTX 670MX appear to be decisive factors.

The Verdict

The data points to a clear conclusion: the NVIDIA GeForce GTX 670MX is the faster GPU in compute performance. It wins both recorded head-to-head benchmarks, has a higher average score (5721 versus 5236), and sits in a higher percentile rank (33rd versus 31st). Its advantages in shading units, TMUs, ROPs, memory bus width, and memory capacity all contribute to this outcome, despite its lower clock speeds and older release date.

A user prioritizing raw compute performance in OpenCL or Vulkan workloads should choose the GTX 670MX. It delivers 9.3% higher OpenCL performance and 9.2% higher Vulkan performance, and its 3 GB memory capacity on a 192-bit bus provides a bandwidth edge of 67.20 GB/s over 64.13 GB/s. Its nearest rivals include the Intel Iris Pro Graphics P6300 and the NVIDIA GeForce GTX 550 Ti, both within a fraction of a percent, so it sits in competitive company for its era.

The GTX 760M, while slower in every recorded benchmark, is not without a role. It draws 20 W less power, with a 55 W TDP against 75 W, and it comes in an MXM Module form factor, which may suit certain mobile chassis designs. Its higher memory clock of 4 Gbps effective does not translate into higher bandwidth due to the narrower 128-bit bus, but the lower power envelope is a measurable advantage for thermally constrained systems. Its nearest rivals are the NVIDIA Quadro 4000M and the AMD Radeon R7 M260X, with the GeForce 940M and GeForce GTX 980M slightly ahead of it.

For a buyer choosing between these two end-of-life parts strictly on benchmark performance, the GTX 670MX is the one to pick. For a buyer whose primary constraint is power consumption or modular form factor, the GTX 760M has the edge. The database shows a 2 to 0 win record for the GTX 670MX, and the numbers behind that record are consistent across both APIs.