Intel UHD Graphics 750 vs NVIDIA GeForce GTX 670M Comparison

The Verdict

The data separates these two mobile graphics parts into distinct use cases. Intel UHD Graphics 750, an integrated solution from the Rocket Lake generation, wins the only direct comparison in the database. It scores 6743 in Geekbench OpenCL, beating the NVIDIA GeForce GTX 670M's 6513 by 3.5%. That slim margin means the Intel part is the better choice for compute workloads that rely on OpenCL acceleration.

The GTX 670M, however, holds its own in specific legacy scenarios. It has a dedicated 1536 MB of GDDR5 memory on a 192 bit bus with 72.00 GB/s bandwidth, whereas the UHD 750 uses system shared memory with bandwidth that is system dependent. For anyone running older DirectX 11 titles or applications that favor a discrete GPU with its own memory pool, the NVIDIA part remains relevant. The GTX 670M also has a much higher pixel rate at 8.372 GPixel/s and texture rate at 33.49 GTexel/s, though the Intel part counters with a higher pixel rate of 10.40 GPixel/s.

Pick the Intel UHD Graphics 750 if you want a modern, efficient integrated GPU with Vulkan support and the higher raw compute score in the database. Pick the GTX 670M if you need a discrete mobile GPU with dedicated VRAM and the higher shading unit count, and if you are working within the constraints of its Fermi 2.0 architecture.

Architecture Differences

The two GPUs come from completely different eras and design philosophies. Intel UHD Graphics 750 uses the Generation 12.1 architecture built on Intel's 14 nm+++ process node, fabricated in-house at Intel. NVIDIA GeForce GTX 670M uses the Fermi 2.0 architecture on TSMC's 40 nm process with 1,950 million transistors packed into a 332 mm² die. The transistor density tells the story: the GTX 670M has 5.9M transistors per mm², while the Intel part does not list such a figure.

The Intel chip is built around 256 shading units, 16 texture mapping units, and 8 raster output units. The NVIDIA chip has more of all three: 336 shading units, 56 TMUs, and 24 ROPs. Despite fewer execution units, the Intel part achieves a higher pixel rate of 10.40 GPixel/s versus 8.372 GPixel/s on the NVIDIA. The GTX 670M counters with a much higher texture rate of 33.49 GTexel/s versus the Intel's 20.80 GTexel/s.

Memory architecture is fundamentally different. The UHD 750 uses system shared memory, with type, bus width, and bandwidth all listed as system dependent. The GTX 670M has 1536 MB of dedicated GDDR5 with a 192 bit bus and 72.00 GB/s of bandwidth, running at 750 MHz or 3 Gbps effective.

API support differs substantially. Intel supports DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.4. NVIDIA supports DirectX 12 (11_0) and OpenGL 4.6, but has no Vulkan support listed. The NVIDIA part also has no FP16 performance listed, while the Intel part delivers 1,331.2 GFLOPS FP16 at a 2:1 ratio. For FP32, the GTX 670M leads with 803.7 GFLOPS versus 665.6 GFLOPS on the Intel.

Power and form factor also diverge. The UHD 750 has a 15 W TDP and is an IGP with a Ring Bus interface. The GTX 670M has a 75 W TDP, comes as an MXM Module, uses an MXM-B (3.0) bus interface, and requires no power connectors.

Where Each One Wins

The Intel UHD Graphics 750 wins the only head-to-head benchmark recorded in the database. In Geekbench OpenCL, it scores 6743 against the GTX 670M's 6513, a 3.5% advantage. That places the Intel part at the 40th percentile among all GPUs, with an average benchmark score of 7441 across its two recorded tests. Its nearest rivals include the AMD Radeon HD 8850M at 7447 (0.1% higher), the AMD Radeon R7 350 at 7425 (0.2% lower), the NVIDIA GeForce GTX 1650 at 7472 (0.4% higher), and the Intel Arc A310 at 7550 (1.4% higher). The UHD 750 also supports Vulkan 1.4, making it the better choice for modern cross-platform graphics APIs.

The GTX 670M wins in raw shading throughput and memory bandwidth. Its 336 shading units, 56 TMUs, and 24 ROPs give it a texture rate of 33.49 GTexel/s, roughly 61% higher than the Intel part's 20.80 GTexel/s. Its dedicated 72.00 GB/s of memory bandwidth is not dependent on system memory performance, which matters in memory-sensitive workloads. The GTX 670M sits at the 38th percentile among all GPUs, with its nearest rivals including the NVIDIA GeForce GT 555M at 6493 (0.3% higher), the NVIDIA Quadro M5000M at 6481 (0.5% higher), the AMD Radeon Vega 10 Mobile at 6476 (0.6% higher), and the Intel UHD Graphics P750 at 6554 (0.6% lower).

For compute workloads that are not memory-bandwidth-bound, the Intel part's higher pixel rate and Vulkan support give it the edge. For texture-heavy workloads and applications that benefit from dedicated VRAM, the GTX 670M is the stronger choice.

FAQ

Q: Which GPU has the higher OpenCL benchmark score?

A: The Intel UHD Graphics 750 scores 6743 in Geekbench OpenCL, which is 3.5% higher than the GTX 670M's 6513.

Q: Does the GTX 670M support Vulkan?

A: No. The GTX 670M lists no Vulkan support in the database. The Intel UHD Graphics 750 supports Vulkan 1.4.

Q: How much dedicated memory does each GPU have?

A: The GTX 670M has 1536 MB of dedicated GDDR5 memory with a 192 bit bus. The Intel UHD Graphics 750 has no dedicated memory; it uses system shared memory with system dependent bandwidth.

Q: Which GPU has more shading units?

A: The GTX 670M has 336 shading units, while the Intel UHD Graphics 750 has 256 shading units.

Q: What is the TDP difference?

A: The Intel UHD Graphics 750 has a 15 W TDP, while the GTX 670M has a 75 W TDP.

Q: Which GPU has the higher pixel rate?

A: The Intel UHD Graphics 750 has a pixel rate of 10.40 GPixel/s, which is higher than the GTX 670M's 8.372 GPixel/s.

Head-to-Head Benchmarks

The database records exactly one direct comparison between these two GPUs. In Geekbench OpenCL, the Intel UHD Graphics 750 scores 6743 while the NVIDIA GeForce GTX 670M scores 6513. That gives the Intel part a 3.5% victory, and it wins the head-to-head category 1 win to 0.

That 3.5% margin is small in practical terms. It suggests the two GPUs are closely matched in raw OpenCL compute performance, despite the architectural differences. The Intel part achieves this with fewer shading units (256 versus 336) and a lower FP32 throughput (665.6 GFLOPS versus 803.7 GFLOPS). Its advantage likely comes from its newer architecture and higher pixel rate of 10.40 GPixel/s versus 8.372 GPixel/s.

The GTX 670M's higher texture rate of 33.49 GTexel/s and dedicated 72.00 GB/s memory bandwidth do not translate into a win in this particular benchmark. The Intel part also has a higher average benchmark score of 7441, though that figure includes its Geekbench Vulkan score of 8138, a test the GTX 670M does not have recorded in the database.

In the broader competitive context, the Intel UHD Graphics 750 sits close to the AMD Radeon HD 8850M (7447, 0.1% higher), the AMD Radeon R7 350 (7425, 0.2% lower), and the NVIDIA GeForce GTX 1650 (7472, 0.4% higher). The GTX 670M sits close to the NVIDIA GeForce GT 555M (6493, 0.3% higher), the NVIDIA Quadro M5000M (6481, 0.5% higher), and the AMD Radeon Vega 10 Mobile (6476, 0.6% higher). Notably, the GTX 670M's nearest rival list includes the Intel UHD Graphics P750 at 6554, which is 0.6% higher. That P750 part is a sibling of the UHD 750 in the same Rocket Lake family, and the GTX 670M trails it by a similar margin to its loss against the UHD 750.

Specification Differences

The two GPUs differ in nearly every recorded specification category, reflecting their different roles in the mobile market.

Manufacturer and process: Intel makes the UHD 750 on its 14 nm+++ node. NVIDIA makes the GTX 670M on TSMC's 40 nm node. The GTX 670M has 1,950 million transistors on a 332 mm² die, with a transistor density of 5.9M per mm². The Intel part does not list transistor count, die size, or density.

Clock speeds: The UHD 750 has a base clock of 300 MHz and a boost clock of 1300 MHz. The GTX 670M lists no base or boost clock, only a memory clock of 750 MHz (3 Gbps effective).

Memory: The UHD 750 uses system shared memory with a system dependent bandwidth. The GTX 670M uses 1536 MB of GDDR5 on a 192 bit bus with 72.00 GB/s bandwidth.

Execution units: The UHD 750 has 256 shading units, 16 TMUs, and 8 ROPs. The GTX 670M has 336 shading units, 56 TMUs, and 24 ROPs.

Throughput rates: The UHD 750 achieves 10.40 GPixel/s and 20.80 GTexel/s. The GTX 670M achieves 8.372 GPixel/s and 33.49 GTexel/s.

Compute performance: The UHD 750 delivers 665.6 GFLOPS FP32 and 1,331.2 GFLOPS FP16 (2:1). The GTX 670M delivers 803.7 GFLOPS FP32 and lists no FP16 figure.

Power and form factor: The UHD 750 has a 15 W TDP, is an IGP, uses a Ring Bus interface, and has motherboard dependent display outputs. The GTX 670M has a 75 W TDP, is an MXM Module, uses an MXM-B (3.0) bus interface, has no power connectors, and has portable device dependent display outputs.

API support: The UHD 750 supports DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.4. The GTX 670M supports DirectX 12 (11_0) and OpenGL 4.6, with no Vulkan support.

Release timeline: The GTX 670M released on 2012-03-21 and has predecessor GeForce 500M and successor GeForce 700M. The UHD 750 released on 2021-03-29 with no predecessor or successor listed. Both are end-of-life products.