NVIDIA GeForce GT 635M vs NVIDIA GeForce GTX 650 Comparison

The NVIDIA GeForce GTX 650 and the NVIDIA GeForce GT 635M represent two distinct approaches to graphics processing from the GeForce 600 era. The GTX 650 is a desktop add-in card built on the Kepler architecture, while the GT 635M is a mobile integrated graphics processor based on the older Fermi design. Benchmark data shows a decisive overall performance gap: the GTX 650 averages 3823 across its benchmark suite, compared to 3740 for the GT 635M. The sole head-to-head test, Geekbench OpenCL, shows the GTX 650 scoring 4545 against 3740 for the GT 635M, a 21.5% advantage. However, the data also reveals that both cards sit at the 22nd percentile of all GPUs, meaning neither offers class-leading compute muscle. This analysis breaks down where each part wins, the architectural reasons behind the performance split, and which user should consider each adapter based strictly on the available figures.

Where Each One Wins

The GTX 650 wins unequivocally in raw computational throughput. In the only direct benchmark shared between the two, Geekbench OpenCL, the GTX 650 posts 4545 points against 3740 for the GT 635M — a 21.5% lead. This advantage is consistent with the card’s broader benchmark profile: it also scores 2400 in Geekbench Metal and 4524 in Geekbench Vulkan, giving it an average benchmark score of 3823 across three tests. The GT 635M, by contrast, has only a single recorded benchmark result, the 3740 OpenCL score, which also serves as its average. For any workload that leverages OpenCL compute — such as general-purpose GPU tasks or certain productivity acceleration — the GTX 650 is the clear choice.

The GT 635M’s wins are narrower and more situational. It draws only 35 W of power, compared to 65 W for the GTX 650, and it requires no power connectors, being classified as an IGP (integrated graphics processor) with a portable-device-dependent display output. This makes it suitable for thin-and-light laptops where power draw and physical footprint are primary constraints. Its 2 GB memory configuration doubles the GTX 650’s 1024 MB frame buffer, which could allow it to address larger datasets or higher-resolution textures in memory-bound scenarios, even if the memory itself is slower DDR3. The GT 635M also matches the GTX 650 in API support for DirectX 12 (11_0) and OpenGL 4.6, so software compatibility is not a differentiator. In benchmark comparisons against its own nearest rivals, the GT 635M comes out slightly ahead of the NVIDIA Quadro 3000M (0.6%), the GT 740M (0.6%), and the GeForce 825M (1.2%), while trailing the Intel UHD Graphics 710 by 1.4%. The GTX 650, meanwhile, is nearly tied with the MX110 (0.3% behind) and the UHD 710 (0.8% ahead), but trails the AMD Radeon R5 Graphics by 1.5% and the Quadro 2000 by 1.9%.

FAQ

Q: Which GPU has the higher average benchmark score?

A: The NVIDIA GeForce GTX 650 has an average benchmark score of 3823 across three tests, while the NVIDIA GeForce GT 635M averages 3740 from its single OpenCL result. The GTX 650 leads by 83 points, or roughly 2.2%.

Q: How much faster is the GTX 650 in the shared OpenCL test?

A: In the head-to-head Geekbench OpenCL benchmark, the GTX 650 scores 4545 versus 3740 for the GT 635M, which corresponds to a 21.5% higher score for the desktop card.

Q: Does the GT 635M have any memory capacity advantage?

A: Yes, the GT 635M comes with 2 GB of DDR3 memory on a 128-bit bus, whereas the GTX 650 has 1024 MB of GDDR5 on the same 128-bit bus. The GT 635M has double the capacity, but the GTX 650 has significantly higher bandwidth (80.00 GB/s vs 28.80 GB/s).

Q: Which GPU supports Vulkan?

A: The GTX 650 supports Vulkan 1.2.175, while the GT 635M has no Vulkan support listed. Both support DirectX 12 (11_0) and OpenGL 4.6.

Q: How do the two compare in terms of power requirements?

A: The GTX 650 has a 65 W TDP and requires a single 6-pin power connector with a suggested 250 W power supply. The GT 635M has a 35 W TDP, uses no power connectors, and is classified as an IGP, making it far less demanding on system power.

Q: Are both GPUs at the same performance percentile?

A: Yes, both the GTX 650 and the GT 635M sit at the 22nd percentile when compared to all GPUs in the database, indicating they occupy a similar tier of overall performance despite the GTX 650’s clear lead in the tested compute workloads.

Head-to-Head Benchmarks

The head-to-head comparison contains exactly one benchmark result: Geekbench OpenCL. In this test, the GTX 650 achieves a score of 4545, while the GT 635M manages 3740. The resulting delta is 21.5% in favor of the GTX 650. This is a substantial margin, indicating that the desktop card delivers over one-fifth more raw compute performance in a standard cross-platform compute benchmark. Given that OpenCL is a common measure of general-purpose GPU throughput, this single data point suggests the GTX 650 is meaningfully more capable for tasks that offload parallel work to the graphics processor.

The GTX 650’s additional benchmark scores reinforce its superiority. In Geekbench Metal, it scores 2400, and in Geekbench Vulkan, it scores 4524. These results show that the card performs consistently across different compute APIs, with Vulkan performance nearly matching its OpenCL output. The GT 635M lacks corresponding Metal and Vulkan scores, so a direct comparison in those APIs is impossible from the data. Still, the average benchmark scores — 3823 for the GTX 650 and 3740 for the GT 635M — confirm that the GTX 650 holds the edge in every measurable compute scenario. There are no benchmark tests where the GT 635M wins; the wins tally is 1 for the GTX 650 and 0 for the GT 635M.

Specification Differences

The two GPUs diverge sharply across nearly every hardware specification. The GTX 650 is built on a 28 nm process node, while the GT 635M uses a larger 40 nm process. Transistor counts reflect this: the GTX 650 packs 2,540 million transistors on a 221 mm² die, resulting in a density of 11.5M transistors per mm². The GT 635M has just 585 million transistors on a 116 mm² die, for a density of 5.0M per mm². The GTX 650’s newer process and higher transistor budget give it far more compute resources: 384 shading units, 32 texture mapping units, and 16 ROPs, versus 96 shading units, 16 TMUs, and 4 ROPs on the GT 635M. Pixel rate is 8.464 GPixel/s for the GTX 650 versus 1.900 GPixel/s for the GT 635M, and texture rate is 33.86 GTexel/s versus 7.600 GTexel/s. FP32 performance is 812.5 GFLOPS for the GTX 650, compared to 182.4 GFLOPS for the GT 635M — a 4.5x difference in raw floating-point throughput.

Memory specifications also differ fundamentally. The GTX 650 uses 1024 MB of GDDR5 with a 128-bit bus and 80.00 GB/s bandwidth, while the GT 635M uses 2 GB of DDR3 on a 128-bit bus with 28.80 GB/s bandwidth. The memory clock is 1250 MHz (5 Gbps effective) on the GTX 650 versus 900 MHz (1800 Mbps effective) on the GT 635M. Power and physical characteristics are polar opposites: the GTX 650 is a single-slot card rated at 65 W with a 6-pin connector, a suggested 250 W PSU, PCIe 3.0 x16 interface, and dimensions of 147 mm (5.8 inches) in length. The GT 635M is an IGP with 35 W TDP, no power connectors, no suggested PSU, and a PCIe 2.0 x16 interface. Display outputs reflect their intended use: the GTX 650 offers 1x DVI and 2x DisplayPort 1.2, while the GT 635M has outputs that are “Portable Device Dependent.”

Architecture Differences

The architectural gap between these two GPUs is generational. The GTX 650 is based on the GK106 chip using the Kepler architecture, while the GT 635M uses the GF108 chip with the older Fermi architecture. Kepler was designed for higher efficiency and greater parallelism per transistor, which is evident in the GTX 650’s 384 shading units — exactly four times the 96 shading units of the Fermi-based GT 635M. The Kepler design also supports Vulkan 1.2.175, whereas the GT 635M lists no Vulkan support, despite both supporting DirectX 12 (11_0) and OpenGL 4.6. The manufacturing process differs as well: 28 nm for the GTX 650 versus 40 nm for the GT 635M, both from TSMC. This process shrink contributes to the GTX 650 achieving 11.5M transistors per mm² compared to 5.0M per mm² on the older node, even though the GTX 650’s die is nearly twice as large.

The Kepler architecture also brings a higher transistor density and more balanced execution resources, with 32 TMUs and 16 ROPs versus 16 TMUs and 4 ROPs on the GT 635M. The memory architecture is another key divergence: Kepler’s GDDR5 controller delivers 80.00 GB/s, while Fermi’s DDR3 controller tops out at 28.80 GB/s. This bandwidth gap, combined with the 4.5x difference in FP32 throughput, explains why the GTX 650 dominates in the OpenCL benchmark — compute workloads often saturate both memory and shader resources. The GT 635M’s only architectural advantages are its lower power envelope (35 W vs 65 W) and larger frame buffer (2 GB vs 1024 MB), which are less relevant for raw compute performance. Both cards are end-of-life products, with the GTX 650 released on 2013-11-26 and the GT 635M on 2012-03-21.

The Verdict

The data points to a clear performance hierarchy: the NVIDIA GeForce GTX 650 is the stronger GPU in every benchmarked compute scenario. Its 21.5% lead in Geekbench OpenCL, 4545 versus 3740, is backed by a 4.5x advantage in FP32 throughput (812.5 GFLOPS vs 182.4 GFLOPS) and a 2.8x advantage in memory bandwidth (80.00 GB/s vs 28.80 GB/s). For users running OpenCL-based workloads, or any task that can harness Vulkan or Metal, the GTX 650 is the only viable choice of the two — the GT 635M lacks Vulkan support entirely and has no recorded Metal or Vulkan scores. The GTX 650’s higher average benchmark score of 3823 versus 3740 further confirms this, even though both parts sit at the 22nd percentile of all GPUs.

The GT 635M, however, is not without a purpose. Its 35 W TDP, lack of power connectors, and IGP classification make it appropriate for laptops where the GTX 650’s 65 W draw and 6-pin connector are impractical. Its 2 GB memory capacity, double the GTX 650’s, could also benefit specific memory-hungry applications that do not require high bandwidth. But for any user comparing the two on desktop performance, the GTX 650 wins outright on every benchmark metric in the data. The GT 635M’s nearest rival comparisons show it trading blows within 1.4% of other mobile GPUs like the Quadro 3000M and GT 740M, but that is a separate tier of performance. The verdict from the numbers is straightforward: choose the GTX 650 for compute performance, choose the GT 635M only if the constraints are power, size, and mobile integration.