NVIDIA GeForce GT 735M vs NVIDIA GeForce GT 745M Comparison

The NVIDIA GeForce GT 745M and GT 735M are both end-of-life Kepler mobile GPUs from the GeForce 700M generation, but the benchmark data shows they are far from interchangeable. While the GT 735M edges out a narrow victory in the single available head-to-head test, the architectural differences reveal a more nuanced picture where the GT 745M holds significant theoretical advantages in memory bandwidth that the GT 735M partially offsets with higher clock speeds and compute rates.

Head-to-Head Benchmarks

The only direct benchmark comparison available is the Geekbench OpenCL test, and the result is surprisingly close. The GT 735M scores 3616, while the GT 745M scores 3580. This yields a delta of -1% from the perspective of the GT 745M, meaning the GT 735M is approximately 1% faster in this specific compute workload. In practical terms, this is a statistical tie—a difference of only 36 points out of roughly 3,600. The data indicates that for general-purpose OpenCL compute, neither card provides a meaningful advantage over the other.

However, this near-parity in OpenCL performance is notable because the two cards achieve it through entirely different hardware configurations. The GT 735M’s victory here is its only win in the head-to-head comparison, giving it a 1-0 record against the GT 745M. But the broader benchmark context shows both cards are clustered tightly with similar rivals. The GT 745M’s average benchmark score is 3953, which places it at the 23rd percentile of all GPUs. Its nearest rival, the AMD Radeon R5 M420, scores 3956 with a delta of -0.1%, while the NVIDIA GeForce 830M scores 3957 with the same -0.1% delta. The GT 745M also sits just 0.3% behind the NVIDIA Quadro K2000 (3964) and 0.6% ahead of the NVIDIA Quadro 2000D (3930). This demonstrates that the GT 745M is operating in a very crowded performance tier where single-digit point differences separate competitors.

The GT 735M, by contrast, has an average benchmark score of 3616, placing it at the 21st percentile. Its nearest rival is the NVIDIA GeForce GTX 1050, which scores 3629 with a delta of -0.3%, meaning the GTX 1050 is only 0.3% faster. Interestingly, the NVIDIA RTX 5000 Mobile Ada Generation scores 3596, putting it 0.6% behind the GT 735M, and the NVIDIA GeForce GT 545 scores 3594, also 0.6% behind. The AMD Radeon HD 6770 completes the rival group at 3649, which is 0.9% ahead. The GT 735M’s nearest rivals span from a modern flagship mobile workstation GPU to a decade-old desktop card, underscoring that its performance level is defined more by compute capability than by generation.

Architecture Differences

The GT 745M is built on the GK107 chip using the original Kepler architecture, while the GT 735M uses the GK208 chip with the Kepler 2.0 architecture. Both are manufactured by TSMC on a 28 nm process node, but the chip configurations diverge significantly. The GT 745M packs 1,270 million transistors into a die size of 118 mm², yielding a transistor density of 10.8M per mm². The GT 735M is a smaller chip with 1,020 million transistors on an 87 mm² die, resulting in a higher transistor density of 11.7M per mm².

The memory subsystem presents the starkest contrast. The GT 745M features 2 GB of GDDR5 memory on a 128-bit bus, delivering a bandwidth of 64.00 GB/s at a memory clock of 1000 MHz (4 Gbps effective). The GT 735M also has 2 GB of memory but uses DDR3 on a 64-bit bus, which drastically reduces bandwidth to just 14.40 GB/s at a memory clock of 900 MHz (1800 Mbps effective). This represents a 4.4x bandwidth advantage for the GT 745M, a factor that should heavily favor it in memory-intensive workloads.

Despite the memory disparity, the GT 735M compensates with higher clock speeds. The GT 745M has no listed base or boost clock in the data, while the GT 735M runs at a base clock of 575 MHz and a boost clock of 628 MHz. In terms of raw compute, the GT 735M achieves 482.3 GFLOPS of FP32 performance, whereas the GT 745M delivers 421.6 GFLOPS—meaning the GT 735M is approximately 14% higher in theoretical peak compute. Similarly, the GT 735M’s pixel rate is 5.024 GPixel/s versus 4.392 GPixel/s for the GT 745M, a 14% advantage, and its texture rate is 20.10 GTexel/s versus 17.57 GTexel/s, an advantage of roughly 14% as well.

Both cards share identical shading unit counts at 384, and both have 32 TMUs. However, the ROP count differs: the GT 745M has 16 ROPs, while the GT 735M has only 8 ROPs. This halving of ROPs in the GT 735M is partially mitigated by its higher clock speed, which explains why its pixel rate is still higher than the GT 745M’s despite having half the ROPs. The power envelope also differs, with the GT 745M rated at a 45 W TDP and the GT 735M at a 33 W TDP, making the GT 735M the more power-efficient option. The bus interface also differs: the GT 745M uses PCIe 3.0 x16, while the GT 735M uses PCIe 3.0 x8.

The Verdict

The data presents a clear trade-off between memory bandwidth and compute throughput. The GT 745M is the superior choice for any workload that depends on memory bandwidth, given its 64.00 GB/s versus the GT 735M’s 14.40 GB/s. The GDDR5 memory type and 128-bit bus provide a four-fold bandwidth advantage that cannot be overcome by the GT 735M’s higher clocks in memory-bound scenarios. For gaming at higher resolutions or with higher texture detail, the GT 745M’s bandwidth advantage is decisive.

Conversely, the GT 735M wins on raw compute throughput. Its 482.3 GFLOPS FP32 performance, higher pixel rate, and higher texture rate give it an edge in compute-bound workloads that fit within its memory constraints. The Geekbench OpenCL result, where the GT 735M scores 3616 versus the GT 745M’s 3580, reflects this compute advantage in practice. The GT 735M also consumes less power at 33 W versus 45 W, making it better suited for thinner laptops with limited cooling.

For most users, the GT 745M is the better all-rounder due to its memory bandwidth. The GT 735M’s compute lead is marginal (14% in FP32) and will rarely translate to visible real-world gains, whereas the GT 745M’s bandwidth advantage is massive and will prevent stuttering in memory-heavy applications. The GT 735M is only preferable when power efficiency is the primary concern and the workload is compute-dominant.

FAQ

Q: Which GPU has higher memory bandwidth?

A: The NVIDIA GeForce GT 745M has significantly higher memory bandwidth at 64.00 GB/s, compared to the GT 735M’s 14.40 GB/s. This is due to the GT 745M’s GDDR5 memory on a 128-bit bus versus the GT 735M’s DDR3 on a 64-bit bus.

Q: Which GPU is faster in the Geekbench OpenCL test?

A: The NVIDIA GeForce GT 735M scores 3616, slightly ahead of the GT 745M’s 3580, a delta of -1% from the GT 745M’s perspective. This is a narrow margin that falls within typical run-to-run variance.

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

A: The GT 745M has a TDP of 45 W, while the GT 735M is rated at 33 W. The GT 735M is therefore the more power-efficient option, making it more suitable for thinner and lighter laptops.

Q: Do both GPUs support the same APIs?

A: Yes, both support DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.2.175. They also share identical display outputs described as "Portable Device Dependent."

Q: Which GPU has more ROPs?

A: The GT 745M has 16 ROPs, double the GT 735M’s 8 ROPs. Despite this, the GT 735M achieves a higher pixel rate of 5.024 GPixel/s versus 4.392 GPixel/s due to its higher clock speeds.

Q: What is the transistor count difference?

A: The GT 745M contains 1,270 million transistors on a 118 mm² die, while the GT 735M contains 1,020 million transistors on an 87 mm² die. The GT 735M has a higher transistor density at 11.7M per mm² versus 10.8M per mm².

Where Each One Wins

The GT 745M wins decisively in memory-bound scenarios. Its 64.00 GB/s bandwidth is a 4.4x advantage over the GT 735M, which will manifest in higher frame rates at 1080p gaming, smoother texture streaming, and better performance in applications that frequently access large datasets. The 128-bit memory bus and GDDR5 memory type are the key enablers here. The GT 745M also benefits from having 16 ROPs, which aids in fill-rate-heavy operations like anti-aliasing and high-resolution rendering, even though its lower clock speed reduces the theoretical pixel rate.

The GT 735M wins in compute-bound and power-constrained scenarios. Its 482.3 GFLOPS FP32 performance is 14% higher than the GT 745M’s 421.6 GFLOPS, and its texture rate of 20.10 GTexel/s surpasses the GT 745M’s 17.57 GTexel/s by a similar margin. The Geekbench OpenCL score of 3616 confirms this compute advantage. At 33 W TDP, the GT 735M is also the better fit for ultraportable laptops where thermal headroom is minimal. The GT 735M’s higher base and boost clocks (575 MHz and 628 MHz) compensate for its narrower 64-bit memory bus in workloads that are not bandwidth-limited.

For users who primarily run compute workloads like OpenCL-based rendering or physics simulations that fit within 2 GB of memory, the GT 735M’s higher compute throughput will be noticeable. For users who play modern games or use GPU-accelerated video editing, the GT 745M’s bandwidth advantage is the more critical factor. The GT 735M’s nearest rival list includes the GTX 1050 (0.3% faster) and the RTX 5000 Mobile Ada Generation (0.6% slower), indicating its compute performance is competitive with much newer and more expensive hardware, but its memory bandwidth remains a severe limitation.

Specification Differences

| Specification | NVIDIA GeForce GT 745M | NVIDIA GeForce GT 735M |

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

| Chip | GK107 | GK208 |

| Architecture | Kepler | Kepler 2.0 |

| Process Node | 28 nm | 28 nm |

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

| Die Size | 118 mm² | 87 mm² |

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

| Base Clock | Not listed | 575 MHz |

| Boost Clock | Not listed | 628 MHz |

| Memory Clock | 1000 MHz (4 Gbps effective) | 900 MHz (1800 Mbps effective) |

| Memory Type | GDDR5 | DDR3 |

| Memory Bus Width | 128 bit | 64 bit |

| Memory Bandwidth | 64.00 GB/s | 14.40 GB/s |

| ROPs | 16 | 8 |

| Pixel Rate | 4.392 GPixel/s | 5.024 GPixel/s |

| Texture Rate | 17.57 GTexel/s | 20.10 GTexel/s |

| FP32 Performance | 421.6 GFLOPS | 482.3 GFLOPS |

| TDP | 45 W | 33 W |

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

| Power Connectors | Not listed | None |

| Average Benchmark Score | 3953 | 3616 |

| Percentile | 23rd | 21st |