NVIDIA GeForce GT 635M vs NVIDIA GeForce GT 735M Comparison

The NVIDIA GeForce GT 635M and GT 735M are both end-of-life mobile graphics solutions from NVIDIA, separated by a generation but aimed at the same entry-level segment. Based on the Geekbench OpenCL benchmark data, the older Fermi-based GT 635M actually holds a slight edge over its Kepler-based successor in raw compute performance, though the architectural differences tell a more nuanced story.

Head-to-Head Benchmarks

The only direct benchmark comparison available is the Geekbench OpenCL test, and it produces a somewhat surprising result. The NVIDIA GeForce GT 635M scores 3740, while the NVIDIA GeForce GT 735M scores 3616. This gives the GT 635M a 3.4% victory in the head-to-head matchup, making it the winner of the single benchmark contest. The delta is modest but consistent—the GT 635M leads, and the GT 735M trails.

Context from the nearest rivals reinforces how close these two parts are. The GT 635M sits at the 22nd percentile of all GPUs, with its nearest competitors including the NVIDIA Quadro 3000M at 3718 (0.6% slower), the GeForce GT 740M at 3717 (0.6% slower), and the GeForce 825M at 3694 (1.2% slower). On the other side, it leads the Intel UHD Graphics 710, which scores 3792, by a -1.4% margin (meaning the Intel part is actually 1.4% faster). The GT 735M, meanwhile, occupies the 21st percentile and finds itself in tight company with far more powerful parts: the GeForce GTX 1050 at 3629 is only 0.3% faster, while the RTX 5000 Mobile Ada Generation at 3596 and GeForce GT 545 at 3594 are both 0.6% slower. The AMD Radeon HD 6770 at 3649 is 0.9% faster.

What this means in practical terms is that the raw compute difference between the GT 635M and GT 735M is nearly negligible—a 3.4% gap in OpenCL performance is well within the range of run-to-run variance on mobile hardware. The GT 635M wins the benchmark, but the margin is thin enough that real-world application performance would likely feel identical. Still, if you are choosing purely on the Geekbench number, the GT 635M is the faster part.

FAQ

Q: Which GPU has the higher Geekbench OpenCL score?

A: The NVIDIA GeForce GT 635M scores 3740, which is 3.4% higher than the GT 735M’s 3616. The GT 635M wins the head-to-head benchmark.

Q: How does the GT 735M compare to a modern desktop GPU like the GTX 1050?

A: The GT 735M scores 3616 in OpenCL, while the GeForce GTX 1050 scores 3629. The GTX 1050 is only 0.3% faster, meaning the GT 735M is effectively on par with it in this specific compute test.

Q: Is the GT 635M significantly faster than its closest rival, the GT 740M?

A: No. The GT 635M scores 3740, and the GT 740M scores 3717, a delta of just 0.6%. The two GPUs are essentially tied in OpenCL performance.

Q: What is the transistor density difference between the two chips?

A: The GT 635M’s GF108 chip packs 585 million transistors into a 116 mm² die, yielding a density of 5.0M transistors per mm². The GT 735M’s GK208 chip has 1,020 million transistors in an 87 mm² die, giving it a density of 11.7M per mm².

Q: Do both GPUs support DirectX 12?

A: Yes, both support DirectX 12 (11_0). They also both support OpenGL 4.6. However, the GT 735M adds Vulkan support at version 1.2.175, while the GT 635M has no Vulkan support listed.

Q: Which GPU has a higher pixel fill rate?

A: The GT 735M has a pixel rate of 5.024 GPixel/s, more than double the GT 635M’s 1.900 GPixel/s. The GT 735M also leads in texture rate at 20.10 GTexel/s versus 7.600 GTexel/s.

Architecture Differences

The two GPUs come from different architectural generations, and that explains much of their divergent behavior. The GT 635M is built on the Fermi architecture (chip GF108), while the GT 735M uses Kepler 2.0 (chip GK208). This is a major generational leap. Fermi was NVIDIA’s first true compute-oriented architecture, but Kepler 2.0 was designed to be more power-efficient and to scale better with additional shading units.

The process node tells a clear story of progress. The GT 635M uses a 40 nm process at TSMC, while the GT 735M moves to a 28 nm process, also at TSMC. This shrink allows the GT 735M to pack far more transistors into a smaller die: 1,020 million transistors in 87 mm² versus 585 million in 116 mm². The transistor density jumps from 5.0M per mm² on the GT 635M to 11.7M per mm² on the GT 735M. That is a 2.34x density improvement, enabling the Kepler chip to fit more execution units in less space.

The compute resources are radically different. The GT 635M has 96 shading units, 16 texture mapping units (TMUs), and 4 render output units (ROPs). The GT 735M, by contrast, has 384 shading units, 32 TMUs, and 8 ROPs. That is a 4x increase in shading units and a 2x increase in both TMUs and ROPs. In theory, this should make the GT 735M vastly more powerful, but the memory subsystem holds it back.

Clock speeds also differ. The GT 635M does not have a listed base or boost clock, but its memory runs at 900 MHz (1800 Mbps effective). The GT 735M has a base clock of 575 MHz and a boost clock of 628 MHz, with the same memory clock of 900 MHz (1800 Mbps effective). The GT 735M’s boost behavior is notable—it can raise clocks above the base when thermals allow, something the GT 635M’s entry does not specify.

Specification Differences

The most consequential difference between these two GPUs is the memory bus width. The GT 635M uses a 128-bit bus, while the GT 735M uses a 64-bit bus. This halves the memory bandwidth: the GT 635M delivers 28.80 GB/s, while the GT 735M manages only 14.40 GB/s. Both have 2 GB of DDR3 memory, but the GT 635M can feed its cores data twice as fast.

The bus interface also differs. The GT 635M uses PCIe 2.0 x16, while the GT 735M uses PCIe 3.0 x8. The newer PCIe 3.0 standard offers more bandwidth per lane, but the x8 link on the GT 735M provides roughly the same total bandwidth as the x16 link on the GT 635M. For an integrated-class mobile GPU, this is rarely a bottleneck.

Compute rates show the GT 735M’s theoretical advantage. The GT 635M has a pixel rate of 1.900 GPixel/s and a texture rate of 7.600 GTexel/s, while the GT 735M reaches 5.024 GPixel/s and 20.10 GTexel/s. FP32 performance is similarly lopsided: the GT 635M delivers 182.4 GFLOPS, while the GT 735M more than doubles that at 482.3 GFLOPS. On paper, the GT 735M is the far more capable compute part.

Power consumption is nearly identical. The GT 635M has a TDP of 35 W, while the GT 735M is rated at 33 W. Both are listed as IGP (integrated graphics processor) slot width, meaning they are soldered onto the motherboard with no power connectors. Display outputs are "Portable Device Dependent" for both, and neither has a launch MSRP listed.

The release dates are about a year apart. The GT 635M launched on 2012-03-21, while the GT 735M launched on 2013-03-31. The GT 635M belongs to the GeForce 600M generation, and the GT 735M belongs to the GeForce 700M generation. The GT 635M’s predecessor is the GeForce 500M and its successor is the GeForce 700M; the GT 735M’s predecessor is the GeForce 600M and its successor is the GeForce 800M.

Where Each One Wins

The GT 635M wins in the only benchmark available, the Geekbench OpenCL test, with a score of 3740 versus 3616. It also has a significant advantage in memory bandwidth—28.80 GB/s versus 14.40 GB/s—thanks to its 128-bit bus. For applications that are memory-bandwidth sensitive, such as certain texture-heavy workloads or older game engines that rely on large data fetches, the GT 635M’s wider bus could compensate for its lower compute throughput. It is also the older part, so it may have more mature driver optimizations for legacy titles from its era.

The GT 735M wins decisively on raw compute specifications. Its 384 shading units, 32 TMUs, and 8 ROPs provide a 4x, 2x, and 2x advantage respectively over the GT 635M. This translates to a pixel rate of 5.024 GPixel/s (2.6x higher) and a texture rate of 20.10 GTexel/s (2.6x higher). FP32 performance of 482.3 GFLOPS is 2.6x the GT 635M’s 182.4 GFLOPS. The GT 735M also has a more modern 28 nm process, which means it achieves this higher performance at a slightly lower TDP (33 W versus 35 W).

Where does each actually win in use? The GT 635M is the better choice for workloads that are bound by memory bandwidth—the 28.80 GB/s figure is double what the GT 735M offers. The GT 735M, on the other hand, is the better choice for compute-heavy tasks like OpenCL acceleration, video encoding, or any workload that can leverage its 384 shading units. The benchmark data, however, contradicts the specification sheet: the GT 635M scores higher in OpenCL despite having fewer compute units, likely because the GT 735M’s halved memory bandwidth starves its larger shader array.

The Verdict

The data presents a clear but counterintuitive picture. The NVIDIA GeForce GT 635M wins the only head-to-head benchmark available, posting a 3.4% higher OpenCL score than the GT 735M. It achieves this with a 128-bit memory bus and 28.80 GB/s of bandwidth, which appears to be the decisive factor. The GT 735M has a vastly superior compute architecture—4x the shading units, 2x the TMUs and ROPs, and 2.6x the FP32 throughput—but its 64-bit memory bus caps bandwidth at 14.40 GB/s, and that bottleneck shows up in the benchmark results.

For a user choosing between these two, the GT 635M is the safer pick for general-purpose compute and older applications, where its memory bandwidth advantage outweighs its smaller shader count. It also edges out the GT 735M in the actual benchmark that matters. The GT 735M is the better choice for anyone who expects to run modern, compute-oriented workloads that can utilize its 384 shading units, or who values the lower TDP and newer process node. However, the 3.4% benchmark delta is small enough that neither GPU will feel dramatically faster in real-world use. Both are end-of-life mobile parts, and the GT 735M’s own nearest rivals include the GeForce GTX 1050 (only 0.3% faster) and the RTX 5000 Mobile Ada Generation (0.6% slower), which puts its performance in perspective—these are all entry-level parts, and the differences between them are minor. If you must pick one, the GT 635M has the better benchmark score, but the GT 735M has the better architecture for future-proofing within its performance class.