NVIDIA GeForce GT 640 vs NVIDIA GeForce GT 735M Comparison

# NVIDIA GeForce GT 735M vs NVIDIA GeForce GT 640

The NVIDIA GeForce GT 735M and NVIDIA GeForce GT 640 are both end-of-life Kepler-generation parts from NVIDIA, but they target fundamentally different use cases. The GT 735M is an integrated graphics processor (IGP) built for portable devices, while the GT 640 is a single-slot desktop card with a broader set of display outputs. Benchmark data shows the GT 640 holds a narrow edge in raw compute performance, but the two cards are remarkably close in overall standing, occupying the 21st and 20th percentiles respectively among all GPUs. The GT 735M ships with a single Geekbench OpenCL score of 3616, while the GT 640 posts a slightly higher OpenCL result of 3736, alongside additional Metal and Vulkan scores that reveal its wider API testing coverage.

Where Each One Wins

The GT 640 wins the only head-to-head benchmark available, edging out the GT 735M in Geekbench OpenCL by 3.2%. That single victory translates to a 1-0 win tally in favor of the GT 640. However, the practical meaning of this win is modest: the delta is small, and both cards sit within a hair of each other in raw compute terms.

The GT 735M’s advantage is not in performance but in form factor and power profile. With a 33 W TDP and an IGP slot width, it is designed to be embedded into laptops and portable systems where power draw and space are at a premium. The GT 640, by contrast, is a 65 W single-slot desktop card that requires a 250 W suggested power supply — a clear signal that it is meant for a more traditional desktop environment.

Where the GT 640 wins more decisively is in memory bandwidth and pixel throughput. Its 128-bit memory bus delivers 28.51 GB/s, nearly double the GT 735M’s 14.40 GB/s from a 64-bit bus. Pixel rate also favors the GT 640 at 7.216 GPixel/s versus 5.024 GPixel/s, and texture rate follows suit at 28.86 GTexel/s versus 20.10 GTexel/s. These are not benchmark scores but architectural throughput figures that hint at where the GT 640 would perform better in fill-rate-bound workloads.

The GT 735M counters with a smaller die and lower transistor count — 87 mm² and 1,020 million transistors versus 118 mm² and 1,270 million — which suggests it is a more power-efficient design per unit of silicon area. Its transistor density of 11.7M / mm² is slightly higher than the GT 640’s 10.8M / mm², indicating a tighter packing of logic on the smaller chip.

Architecture Differences

The two GPUs are built on the same 28 nm TSMC process node, but they use different Kepler implementations. The GT 735M is based on the GK208 chip with Kepler 2.0 architecture, while the GT 640 uses the GK107 chip with the original Kepler architecture. This generational split within the Kepler family explains several spec-level differences.

Both cards feature 384 shading units and 32 texture mapping units, so the core compute configuration is identical in those respects. The GT 640 pulls ahead with 16 ROPs versus the GT 735M’s 8 ROPs, which directly contributes to its higher pixel fill rate. The memory subsystem also diverges: both use 2 GB of DDR3, but the GT 640’s 128-bit bus width is exactly double the GT 735M’s 64-bit bus, resulting in the significant bandwidth gap noted earlier.

Clock speeds tell another story. The GT 735M has defined base and boost clocks of 575 MHz and 628 MHz, respectively, while the GT 640 does not list base or boost clocks in the data — only a memory clock of 891 MHz (1782 Mbps effective). This omission is notable: the GT 640’s higher FP32 throughput of 692.7 GFLOPS versus 482.3 GFLOPS for the GT 735M suggests it runs at a higher core clock, but the exact figures are not provided.

API support is identical across both cards: DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.2.175. This parity means software compatibility should be the same for modern titles and applications. The bus interface differs, with the GT 735M using PCIe 3.0 x8 and the GT 640 using PCIe 3.0 x16 — a bandwidth advantage for the desktop card in data-heavy workloads.

The GT 640 is the older product, released in June 2012, while the GT 735M followed in March 2013. The GT 640’s predecessor is GeForce 500 and its successor is GeForce 700, while the GT 735M sits between GeForce 600M and GeForce 800M in the mobile lineup.

Head-to-Head Benchmarks

The only direct benchmark comparison in the data is Geekbench OpenCL, where the GT 640 scores 3736 against the GT 735M’s 3616. That is a 3.2% difference, with the GT 640 as the winner. In percentile terms, this places the GT 735M at the 21st percentile and the GT 640 at the 20th percentile — a nearly identical standing despite the GT 640’s higher raw score.

Contextualizing these scores against their nearest rivals reveals how close these parts are to other GPUs. The GT 735M’s OpenCL score of 3616 is just 0.3% behind the NVIDIA GeForce GTX 1050 (3629) and 0.6% ahead of both the NVIDIA RTX 5000 Mobile Ada Generation (3596) and the NVIDIA GeForce GT 545 (3594). It trails the AMD Radeon HD 6770 (3649) by 0.9%. These are remarkably small gaps, suggesting the GT 735M sits in a cluster of GPUs with nearly identical compute performance.

The GT 640’s average benchmark score of 3210 is lower than its OpenCL score because it also includes Metal (2086) and Vulkan (3809) results, which vary widely. Its nearest rivals include the Intel Arc Pro B60 (3182, with the GT 640 ahead by 0.9%), the NVIDIA Quadro P1000 (3163, ahead by 1.5%), the NVIDIA GeForce 920M (3287, behind by 2.3%), and the NVIDIA GeForce GT 730M (3316, behind by 3.2%). This mix of rivals spans very different market segments, from professional workstation GPUs to low-end mobile parts, underscoring the GT 640’s broad performance envelope.

The GT 735M’s single benchmark score means no Vulkan or Metal data exists for comparison. This is a data gap that limits direct cross-API analysis. The GT 640’s Vulkan score of 3809 is notably higher than its OpenCL score, suggesting it may perform better in Vulkan-based workloads, though no Vulkan result exists for the GT 735M to compare against.

The Verdict

The data points to a clear but narrow victory for the GT 640 in raw compute performance. It wins the only head-to-head benchmark, posts a higher FP32 throughput (692.7 GFLOPS versus 482.3 GFLOPS), and doubles the memory bandwidth and pixel rate of the GT 735M. For anyone running compute-heavy tasks or fill-rate-bound workloads, the GT 640 is the stronger choice based on these measurable differences.

However, the GT 735M is not without its own case. Its 33 W TDP is roughly half the GT 640’s 65 W, and its IGP form factor means it is designed for portable devices where the GT 640 cannot physically exist. The GT 735M’s smaller die (87 mm²) and lower transistor count (1,020 million) suggest it achieves near-identical OpenCL performance with significantly less silicon and power. The benchmark gap of 3.2% is small enough that real-world differences in mobile versus desktop environments could easily flip the practical outcome.

The GT 640’s launch MSRP is 99 USD, which is worth noting as a historical reference point, but no pricing information exists for the GT 735M. The GT 640 also offers a richer display output set — 1x DVI, 1x HDMI 1.4a, and 1x DisplayPort 1.2 — while the GT 735M’s outputs are listed as "Portable Device Dependent," meaning they vary by laptop design.

Who should pick which? Strictly from the data, the GT 640 is the better-performing desktop GPU with a higher ceiling in bandwidth, fill rate, and FP32 compute. The GT 735M is the better choice for portable implementations where power efficiency and size matter more than peak throughput. The GT 640’s 250 W suggested PSU requirement is a practical consideration for builders, while the GT 735M’s lack of power connectors makes it a drop-in solution for mobile designs.

FAQ

Q: Which GPU has a higher average benchmark score?

A: The GT 735M has an average benchmark score of 3616, while the GT 640 averages 3210. However, this is because the GT 640’s average includes Metal and Vulkan scores, which drag down its average despite a higher OpenCL score.

Q: How much faster is the GT 640 in OpenCL performance?

A: The GT 640 scores 3736 in Geekbench OpenCL versus 3616 for the GT 735M, a 3.2% difference in favor of the GT 640.

Q: Do both GPUs support the same graphics APIs?

A: Yes, both support DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.2.175, making their API compatibility identical.

Q: What is the memory bandwidth difference between the two?

A: The GT 640 has a 128-bit memory bus delivering 28.51 GB/s, while the GT 735M has a 64-bit bus delivering 14.40 GB/s — the GT 640 offers roughly double the bandwidth.

Q: Which GPU is more power-efficient on paper?

A: The GT 735M has a 33 W TDP versus 65 W for the GT 640, making it the lower-power part. It also uses a smaller die (87 mm² versus 118 mm²) and fewer transistors (1,020 million versus 1,270 million).

Q: What form factors do these GPUs come in?

A: The GT 735M is an IGP (integrated graphics processor) designed for portable devices, while the GT 640 is a single-slot desktop card measuring 145 mm (5.7 inches) in length with display outputs for DVI, HDMI 1.4a, and DisplayPort 1.2.

Specification Differences

| Specification | NVIDIA GeForce GT 735M | NVIDIA GeForce GT 640 |

|---|---|---|

| Chip | GK208 | GK107 |

| Architecture | Kepler 2.0 | Kepler |

| Generation | GeForce 700M | GeForce 600 |

| Die Size | 87 mm² | 118 mm² |

| Transistors | 1,020 million | 1,270 million |

| Transistor Density | 11.7M / mm² | 10.8M / mm² |

| Base Clock | 575 MHz | Not listed |

| Boost Clock | 628 MHz | Not listed |

| Memory Clock | 900 MHz (1800 Mbps effective) | 891 MHz (1782 Mbps effective) |

| Memory Bus Width | 64 bit | 128 bit |

| Memory Bandwidth | 14.40 GB/s | 28.51 GB/s |

| ROPs | 8 | 16 |

| Pixel Rate | 5.024 GPixel/s | 7.216 GPixel/s |

| Texture Rate | 20.10 GTexel/s | 28.86 GTexel/s |

| FP32 | 482.3 GFLOPS | 692.7 GFLOPS |

| TDP | 33 W | 65 W |

| Slot Width | IGP | Single-slot |

| Suggested PSU | None listed | 250 W |

| Bus Interface | PCIe 3.0 x8 | PCIe 3.0 x16 |

| Display Outputs | Portable Device Dependent | 1x DVI, 1x HDMI 1.4a, 1x DisplayPort 1.2 |

| Dimensions | Not listed | 145 mm (5.7 inches) length |

| Release Date | 2013-03-31 | 2012-06-04 |

| Predecessor | GeForce 600M | GeForce 500 |

| Successor | GeForce 800M | GeForce 700 |

| Launch MSRP | Not listed | 99 USD |