GPU Comparison

# The Verdict

The data presents a curious split decision. The NVIDIA GeForce GT 745M and the NVIDIA GeForce MX110 each claim one benchmark victory, yet the magnitude of those wins is wildly asymmetrical. In the Geekbench Vulkan test, the GT 745M surges ahead by 61.2%, posting a score of 5502 against the MX110's 3413. Conversely, in Geekbench OpenCL, the MX110 counters with a 15.9% advantage, scoring 4255 versus 3580. The average benchmark scores tell a nuanced story: the GT 745M averages 3953 across its three tests, while the MX110 averages 3834 across its two, a difference of roughly 3%.

Both GPUs sit at the 23rd percentile among all GPUs, meaning neither escapes the entry-level tier. However, the character of their performance differs sharply. The MX110's OpenCL win suggests compute-oriented workloads favor its architecture, while the GT 745M's Vulkan dominance points to graphics API efficiency. For a buyer choosing between these two end-of-life mobile parts, the decision hinges on workload: if the target applications leverage Vulkan, the GT 745M is the clear pick; if OpenCL compute matters more, the MX110 edges ahead. The data does not support a universal winner, it supports a workload-specific one.

# Architecture Differences

The two chips come from different NVIDIA architectural generations, and those differences show up directly in their specifications. The GT 745M uses the GK107 chip built on the Kepler architecture, fabricated by TSMC on a 28 nm process. It packs 1,270 million transistors into a 118 mm² die, yielding a transistor density of 10.8M per mm². The MX110, by contrast, employs the GM108S chip on the Maxwell architecture, also on TSMC's 28 nm node. It contains fewer transistors, 1,020 million, but fits them into a smaller 77 mm² die, achieving a higher density of 13.2M per mm². This density improvement reflects Maxwell's efficiency-oriented design philosophy.

The memory subsystems diverge significantly. The GT 745M features a 128-bit memory bus with 2 GB of GDDR5, delivering a bandwidth of 64.00 GB/s at 1000 MHz (4 Gbps effective). The MX110 narrows the bus to 64-bit while keeping 2 GB of GDDR5, but its memory runs at 1253 MHz (5 Gbps effective), yielding only 40.10 GB/s. That is a 37% bandwidth deficit for the MX110, a substantial gap that likely explains several benchmark outcomes. The MX110 compensates with higher clock speeds: its base clock is 978 MHz with a 1006 MHz boost, whereas the GT 745M's base and boost clocks are not specified in the data.

Compute resources also differ. The GT 745M fields 384 shading units, 32 texture mapping units (TMUs), and 16 raster output units (ROPs). The MX110 halves those counts to 256 shaders, 16 TMUs, and 8 ROPs. Despite fewer cores, the MX110's Maxwell architecture achieves higher raw throughput: its FP32 performance is 515.1 GFLOPS versus 421.6 GFLOPS for the GT 745M, a 22% advantage. Texture rate favors the GT 745M slightly at 17.57 GTexel/s versus 16.10 GTexel/s, while pixel rate goes decisively to the MX110 at 8.048 GPixel/s against 4.392 GPixel/s, an 83% lead. The MX110 also runs cooler on paper, with a 30 W TDP versus 45 W for the GT 745M, and connects via PCIe 3.0 x4 rather than the GT 745M's PCIe 3.0 x16 interface.

# FAQ

Q: Which GPU has the higher average benchmark score?

A: The GT 745M averages 3953 across its three benchmark tests (Geekbench Metal, OpenCL, and Vulkan), while the MX110 averages 3834 across its two tests (OpenCL and Vulkan). That puts the GT 745M roughly 3% ahead on average.

Q: Why does the MX110 win the OpenCL test despite having fewer shading units?

A: The MX110 has 256 shading units versus 384 in the GT 745M, but its Maxwell architecture delivers higher FP32 throughput at 515.1 GFLOPS compared to 421.6 GFLOPS. It also boosts to 1006 MHz, which helps compensate for its smaller core count in compute-heavy workloads.

Q: Is the GT 745M's wider memory bus a decisive advantage?

A: The data suggests it matters in Vulkan but not in OpenCL. The GT 745M's 128-bit bus provides 64.00 GB/s of bandwidth versus 40.10 GB/s for the MX110's 64-bit bus. In Vulkan, the GT 745M wins by 61.2%, but in OpenCL, the MX110 still wins by 15.9%, indicating bandwidth is not the sole determinant.

Q: What do their percentile rankings indicate?

A: Both GPUs sit at the 23rd percentile among all GPUs, meaning they are entry-level parts. Their nearest rivals include the AMD Radeon R5 M420 (averaging 3956, within 0.1% of the GT 745M) and the NVIDIA GeForce GTX 650 (averaging 3823, within 0.3% of the MX110).

Q: Which GPU supports newer API versions?

A: Both support DirectX 12 (11_0) and OpenGL 4.6. The MX110 supports Vulkan 1.4, while the GT 745M supports Vulkan 1.2.175. The MX110's newer Vulkan version does not translate into Vulkan benchmark success, however, as the GT 745M wins that test decisively.

Q: Are these GPUs still in production?

A: No. Both are marked end-of-life. The GT 745M was released in early 2013 as part of the GeForce 700M generation, while the MX110 arrived in late 2017 under the GeForce MX (1xx) branding.

# Specification Differences

| Specification | NVIDIA GeForce GT 745M | NVIDIA GeForce MX110 |

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

| Chip | GK107 | GM108S |

| Architecture | Kepler | Maxwell |

| Process Node | 28 nm | 28 nm |

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

| Die Size | 118 mm² | 77 mm² |

| Transistor Density | 10.8M / mm² | 13.2M / mm² |

| Base Clock | Not specified | 978 MHz |

| Boost Clock | Not specified | 1006 MHz |

| Memory Clock | 1000 MHz (4 Gbps effective) | 1253 MHz (5 Gbps effective) |

| Memory Size | 2 GB | 2 GB |

| Memory Type | GDDR5 | GDDR5 |

| Memory Bus Width | 128 bit | 64 bit |

| Memory Bandwidth | 64.00 GB/s | 40.10 GB/s |

| Shading Units | 384 | 256 |

| TMUs | 32 | 16 |

| ROPs | 16 | 8 |

| Pixel Rate | 4.392 GPixel/s | 8.048 GPixel/s |

| Texture Rate | 17.57 GTexel/s | 16.10 GTexel/s |

| FP32 Performance | 421.6 GFLOPS | 515.1 GFLOPS |

| TDP | 45 W | 30 W |

| Power Connectors | Not specified | None |

| Bus Interface | PCIe 3.0 x16 | PCIe 3.0 x4 |

| Vulkan Version | 1.2.175 | 1.4 |

| Release Date | 2013-03-31 | 2017-11-16 |

# Head-to-Head Benchmarks

The head-to-head data gives each GPU one win, but the margins tell opposing stories. In the Geekbench Vulkan test, the GT 745M achieves a score of 5502 against the MX110's 3413, a delta of 61.2%. This is a commanding victory. The GT 745M's 128-bit memory bus and 64.00 GB/s bandwidth likely provide the throughput that Vulkan applications crave, while the MX110's narrower 64-bit bus and 40.10 GB/s bandwidth become a bottleneck. The result is stark: the older Kepler part outperforms the newer Maxwell part by nearly two-thirds in this API.

The Geekbench OpenCL test flips the script. The MX110 scores 4255 versus the GT 745M's 3580, a 15.9% margin in favor of the newer chip. This is a solid but not overwhelming win. The MX110's higher FP32 throughput (515.1 GFLOPS) and superior pixel rate (8.048 GPixel/s) likely drive this result, even though it has half the ROPs and TMUs of the GT 745M. The MX110's higher boost clock of 1006 MHz also helps keep its shader array fed despite the smaller core count.

Notably, the GT 745M has a third benchmark result, Geekbench Metal at 2777, that the MX110 does not have. This missing data point complicates direct comparison. The GT 745M's average score of 3953 includes that Metal result, while the MX110's 3834 average is drawn from only two tests. The delta in average scores is 119 points, or roughly 3%, which falls within the noise of their nearest rivals: the GT 745M's closest competitor, the AMD Radeon R5 M420, averages 3956 (a -0.1% delta), while the MX110's closest competitor, the NVIDIA GeForce GTX 650, averages 3823 (a 0.3% delta).

# Where Each One Wins

The GT 745M is the clear victor in Vulkan-based graphics workloads. Its 61.2% lead in the Vulkan benchmark suggests that games or applications built on that API will run noticeably better on the older Kepler chip. The 128-bit memory bus provides 64.00 GB/s of bandwidth, which is 60% more than the MX110's 40.10 GB/s. For texture-heavy scenes or high-resolution rendering, that bandwidth advantage is likely decisive. The GT 745M also offers a PCIe 3.0 x16 interface, which may reduce data transfer bottlenecks in systems that support it.

The MX110 stakes its claim in OpenCL compute tasks. Its 15.9% OpenCL victory, combined with 515.1 GFLOPS of FP32 performance, makes it the better choice for general-purpose GPU compute, video encoding, and other parallel workloads that leverage OpenCL. The MX110's higher pixel rate (8.048 GPixel/s versus 4.392 GPixel/s) also suggests better fill-rate-bound performance in certain rendering paths, even if the overall Vulkan results do not reflect that. Its lower 30 W TDP is another point in its favor for thin-and-light laptops where thermal headroom is scarce.

The data also hints at architectural efficiency: the MX110 achieves its wins with fewer transistors (1,020 million versus 1,270 million) and a smaller die (77 mm² versus 118 mm²). Maxwell's design extracts more performance per transistor, but Kepler's wider memory bus gives the GT 745M a resilience in bandwidth-sensitive scenarios that Maxwell's clock-speed advantage cannot fully overcome. For a user prioritizing modern API graphics, the GT 745M is the pick. For compute throughput and power efficiency, the MX110 leads. Neither GPU escapes the bottom quartile, so expectations should remain modest.