NVIDIA GeForce GT 750M

The NVIDIA GeForce GT 750M is a mobile graphics solution from the Kepler generation, built on a 28 nm process at TSMC. It targets the entry-level segment of the GeForce 700M series, featuring 384 shading units, 32 texture mapping units, and 16 raster output units. With an average benchmark score of 3215, it sits at the 18th percentile of all GPUs, placing it firmly in the lower performance tier, yet its data reveals a specific profile for light gaming and multimedia tasks.

Memory Subsystem

The GT 750M is equipped with 2 GB of GDDR5 memory, which was a notable pairing for its class at launch, as many contemporaries still relied on slower DDR3. The memory operates at 1003 MHz, translating to 4 Gbps effective, across a 128-bit bus interface. This configuration yields a memory bandwidth of 64.19 GB/s, a figure that is modest by modern standards but was adequate for the 1080p resolutions common during its release period.

The 128-bit bus width is a critical limitation for high-resolution workloads. At 1440p or 4K, the available bandwidth becomes a bottleneck, as the GPU must constantly fetch data from VRAM to feed its 384 shading units. The pixel rate of 7.736 GPixel/s and texture rate of 30.94 GTexel/s further indicate that the card is not designed to push large framebuffers. For 1080p gaming, the 2 GB capacity is sufficient for most titles of its era, but modern games with high-resolution texture packs will likely exceed this allocation, causing stuttering or texture pop-in. Benchmark results suggest the memory subsystem provides balanced throughput for the GPU’s compute capabilities, but it does not offer headroom for future software demands.

Ray Tracing and Feature Set

The GT 750M does not include dedicated ray tracing cores or tensor cores, as these hardware units were introduced in later architectures. Its Kepler-based design focuses on traditional rasterization, with DirectX 12 (11_0) support being the highest API level available. While DirectX 12 is nominally supported, the feature level is 11_0, meaning it cannot leverage the advanced features of DirectX 12 Ultimate, such as variable rate shading or mesh shaders.

On the API front, the GPU supports OpenGL 4.6 and Vulkan 1.2.175, which allows for modern compute and graphics workloads in compatible applications. OpenGL 4.6 provides access to advanced rendering techniques, while Vulkan support enables lower-overhead driver interactions, which can improve performance in well-optimized titles. However, without hardware-accelerated ray tracing, any ray-traced effects must be handled through compute shaders, which would severely tax the GPU’s 742.7 GFLOPS FP32 throughput. The absence of tensor cores also means no AI-accelerated features like DLSS, leaving the card reliant on native resolution rendering. The feature set is thus adequate for legacy titles and productivity apps, but it lacks the hardware foundations for contemporary graphical advancements.

Benchmark Performance

The GT 750M’s average benchmark score of 3215 places it within a tightly contested cluster of low-end GPUs. The data shows a marginal delta of -0.2% relative to the NVIDIA Quadro P620, which scores 3221. This effectively puts the two cards at performance parity, with differences well within run-to-run variance. The Quadro P620 is a workstation-oriented card, so the GT 750M matching it in raw compute suggests the mobile chip holds its own in general-purpose tasks.

Against the Intel HD Graphics 4600, the GT 750M leads by a razor-thin 0.2%, with the Intel solution scoring 3208. This is a surprising result, as integrated graphics typically lag discrete GPUs by a wider margin. The data implies that the HD 4600’s performance is competitive in synthetic benchmarks, though real-world gaming may favor the GT 750M’s dedicated memory bandwidth and driver optimizations. The gap widens slightly against the NVIDIA GeForce 920M, where the GT 750M trails by -1.6% (3266 vs. 3215). The 920M, a later entry-level part, edges out the older chip, indicating generational efficiency gains from NVIDIA. Finally, the GT 750M sits -2.7% behind the NVIDIA GeForce GT 740, which scores 3304. The GT 740 is a desktop card, so this delta reflects the inherent clock and thermal constraints of mobile designs.

The most telling statistic is the 18th percentile ranking among all GPUs. This means over 80% of modern graphics cards outperform the GT 750M, even those from integrated families. The average benchmark score of 3215 is the midpoint between the Geekbench Metal score of 2163 and the OpenCL score of 4266, highlighting that the GPU performs significantly better in compute-heavy OpenCL workloads than in graphics-oriented Metal tests. This dichotomy suggests the Kepler architecture’s compute units are capable, but its fixed-function graphics pipeline is dated.

FAQ

Q: What is the memory bandwidth of the GT 750M?

A: The GT 750M has a memory bandwidth of 64.19 GB/s, achieved via 2 GB of GDDR5 memory on a 128-bit bus running at 4 Gbps effective.

Q: Does the GT 750M support ray tracing?

A: No, the GT 750M lacks dedicated ray tracing cores. It relies on the Kepler architecture, which does not include hardware acceleration for ray-traced effects.

Q: How does the GT 750M compare to the Intel HD Graphics 4600?

A: The GT 750M scores 3215, which is 0.2% higher than the HD 4600’s 3208. The difference is negligible in synthetic benchmarks, though the discrete GPU may offer better sustained performance.

Q: What is the FP32 performance of the GT 750M?

A: The GPU delivers 742.7 GFLOPS of FP32 compute throughput, which is typical for its class and generation.

Q: What are the API capabilities of the GT 750M?

A: It supports DirectX 12 (feature level 11_0), OpenGL 4.6, and Vulkan 1.2.175, enabling modern software compatibility albeit without advanced DirectX 12 features.

Q: Is the GT 750M suitable for modern gaming?

A: Benchmark data shows it ranks at the 18th percentile of all GPUs, indicating it will struggle with demanding modern titles, especially at higher resolutions due to its limited memory bandwidth.

How It Compares

NVIDIA Quadro P620: The GT 750M trails the Quadro P620 by a negligible 0.2%, with scores of 3215 and 3221, respectively. This near-identical performance is notable given the P620’s professional focus, suggesting the GT 750M offers comparable raw compute for consumer workloads.

Intel HD Graphics 4600: The GT 750M leads the integrated Intel solution by just 0.2%, a razor-thin margin that challenges expectations. The data implies that for many tasks, the performance difference between a discrete entry-level GPU and a contemporary iGPU is minimal, though driver support and dedicated VRAM may tip real-world scenarios.

NVIDIA GeForce 920M: The newer 920M outperforms the GT 750M by 1.6%, scoring 3266 versus 3215. This generational improvement, while small, shows that even entry-level GPUs advanced modestly in the intervening period, with better efficiency and possibly higher clocks.

NVIDIA GeForce GT 740: The desktop GT 740 leads by 2.7%, achieving 3304 against the GT 750M’s 3215. The delta is modest, but the desktop card benefits from superior thermal headroom and power delivery, allowing it to sustain higher performance without the constraints of a mobile chassis.

Who Should Consider It

The GT 750M’s benchmark profile positions it as a solution for 720p gaming or 1080p at low to medium settings in older titles. Its 2 GB VRAM and 64.19 GB/s bandwidth are just adequate for such workloads, but the 18th percentile ranking means it will choke on modern AAA releases. The OpenCL score of 4266 suggests better aptitude for compute tasks like video encoding or basic 3D rendering, where the 384 shading units can be fully utilized without the memory pressure of gaming. Users with a Geekbench Metal score of 2163 should avoid Metal-accelerated creative applications, as the GPU clearly underperforms in that API.

For esports titles like CS:GO or League of Legends, the GT 750M can deliver playable framerates at 1080p with reduced settings. The Vulkan 1.2.175 support may help in optimized titles, but the lack of ray tracing and tensor cores eliminates any future-proofing. Those who primarily consume media or use office applications will find the GPU sufficient, but it is not recommendable for any current gaming beyond legacy or lightweight genres. The performance parity with the Intel HD 4600 is a cautionary tale—it suggests that many users may not notice a tangible upgrade from integrated graphics without specific software needs.

Power and Cooling

The GT 750M has a thermal design power (TDP) of 50 W, a modest figure that reflects its mobile orientation. This low power draw means it requires no dedicated power connectors, as it draws all its energy from the MXM module interface. The slot width is specified as MXM Module, indicating it is designed for laptop upgradeability rather than desktop installation. The absence of a suggested PSU rating reinforces that this is not a desktop card; it is intended for systems where the motherboard and battery manage power delivery.

The 28 nm process node from TSMC, with 1,270 million transistors on a 118 mm² die, yields a transistor density of 10.8M per mm². This density is modest by today’s standards but was efficient for its era, contributing to the low TDP. The base clock of 941 MHz and boost clock of 967 MHz are conservative, likely set to maintain stable thermals within thin laptop chassis. The lack of a power connector simplifies integration, but it also means the GPU cannot be overclocked beyond its factory settings without risking instability. Cooling will be handled by the laptop’s existing thermal solution, which should suffice given the 50 W envelope, but sustained loads may induce thermal throttling if the chassis has poor airflow. The production status is end-of-life, and the predecessor GeForce 600M and successor GeForce 800M bracket its lifespan, with a release date of January 2013 marking its market entry.