NVIDIA GeForce GTX 680M vs NVIDIA GeForce GTX 750 Comparison

The NVIDIA GeForce GTX 750 and NVIDIA GeForce GTX 680M represent two distinct design philosophies from the same manufacturer, separated by nearly two years of architectural evolution. The desktop-oriented GTX 750, built on the Maxwell architecture, delivers a higher average benchmark score of 7222 compared to the mobile GTX 680M’s 7023, yet the laptop part wins the only head-to-head test where both share a benchmark. This creates a nuanced picture: the GTX 750 is the more efficient and modern part for desktop use, while the GTX 680M offers superior raw throughput in certain legacy APIs, particularly when its larger memory pool and wider bus come into play. The data shows a near-tie in overall performance, but the underlying specifications tell a story of two very different engineering trade-offs.

FAQ

Q: Which GPU has the higher average benchmark score?

A: The NVIDIA GeForce GTX 750 leads with an average benchmark score of 7222, placing it in the 40th percentile of all GPUs. The GTX 680M trails at 7023, sitting in the 39th percentile, a difference of roughly 2.8% in aggregate performance.

Q: How do the two GPUs compare in the Geekbench Metal test?

A: The GTX 680M wins decisively, scoring 4815 against the GTX 750’s 4274. This represents an 11.2% advantage for the mobile part in Apple’s Metal API, which is the largest single-benchmark gap between the two cards.

Q: What about OpenCL performance?

A: The GTX 750 takes the OpenCL test with a score of 9315, narrowly edging out the GTX 680M’s 9230. The 0.9% margin makes this effectively a statistical tie, but the desktop card still holds the top spot.

Q: Which GPU has more shading units?

A: The GTX 680M features 1344 shading units, more than double the GTX 750’s 512. However, the GTX 750 compensates with substantially higher clock speeds, running at 1020 MHz base and 1085 MHz boost versus the GTX 680M’s 719 MHz base and 758 MHz boost.

Q: What are the memory specifications of each card?

A: The GTX 680M comes with 4 GB of GDDR5 memory on a 256-bit bus, delivering 115.2 GB/s of bandwidth. The GTX 750 has 1024 MB of GDDR5 on a 128-bit bus, providing 80.19 GB/s. The GTX 680M also has faster effective memory at 3.6 Gbps versus 5 Gbps effective on the GTX 750, though the latter’s higher clock rate does not overcome the bus width disadvantage.

Q: Which card consumes less power?

A: The GTX 750 has a TDP of 55 W and requires no power connectors, while the GTX 680M draws 100 W and is designed as an MXM module for laptops. The GTX 750’s power efficiency is a direct result of its Maxwell architecture and smaller transistor count of 1,870 million versus the GTX 680M’s 3,540 million.

Architecture Differences

The two GPUs are built on the same 28 nm process node at TSMC, but their internal designs diverge sharply. The GTX 750 uses the GM107 chip with a die size of 148 mm² and 1,870 million transistors, resulting in a transistor density of 12.6M per mm². The GTX 680M employs the GK104 chip, a much larger 294 mm² die housing 3,540 million transistors, yielding a slightly lower density of 12.0M per mm². This size difference reflects their target platforms: the GTX 750 is a compact desktop card, while the GTX 680M is a mobile module.

Architecturally, the GTX 750 is based on Maxwell, the successor to Kepler, which is the architecture powering the GTX 680M. Maxwell introduced significant efficiency improvements, which is evident in the GTX 750’s 55 W TDP despite having fewer but faster cores. The GTX 680M’s Kepler design relies on a massive parallel array of 1344 shading units and 112 texture mapping units, compared to the GTX 750’s 512 shading units and 32 TMUs. The GTX 680M also doubles the render output units, with 32 ROPs versus the GTX 750’s 16.

Memory subsystems differ fundamentally. The GTX 680M offers 4 GB of GDDR5 on a 256-bit interface with 115.2 GB/s bandwidth, while the GTX 750 has 1024 MB on a 128-bit bus with 80.19 GB/s. The GTX 680M’s memory clock is 900 MHz (3.6 Gbps effective), whereas the GTX 750 runs at 1253 MHz (5 Gbps effective). Despite the lower clock, the GTX 680M’s wider bus provides superior bandwidth, a critical factor for high-resolution textures and compute workloads.

API support is nearly identical, with both cards offering DirectX 12 (11_0), OpenGL 4.6, and Vulkan. However, the GTX 750 supports Vulkan 1.4, while the GTX 680M is limited to Vulkan 1.2.175. The GTX 750 also has a faster pixel rate of 17.36 GPixel/s and texture rate of 34.72 GTexel/s, but the GTX 680M counters with 21.22 GPixel/s and 84.90 GTexel/s, reflecting its higher core count. The GTX 680M’s FP32 performance is 2.038 TFLOPS, nearly double the GTX 750’s 1,111.0 GFLOPS.

The Verdict

The data points to a clear but conditional recommendation. For desktop users building a low-power system, the GTX 750 is the logical choice. It achieves a higher average benchmark score of 7222 against 7023, wins the OpenCL test, and does so with a 55 W TDP that requires no auxiliary power connectors. Its Maxwell architecture delivers modern Vulkan 1.4 support and a smaller physical footprint at 145 mm in length. The GTX 750 also has a launch MSRP of 119 USD, making it a straightforward drop-in upgrade for legacy systems.

For laptop users or those requiring maximum compute throughput, the GTX 680M is the stronger option despite its lower aggregate score. It wins the Metal benchmark by 11.2%, a significant margin that indicates better optimization for Apple’s ecosystem. Its 4 GB memory capacity and 256-bit bus provide double the bandwidth of the GTX 750, which is advantageous for large datasets or high-resolution textures. The GTX 680M’s 1344 shading units and 2.038 TFLOPS FP32 performance make it a raw compute powerhouse, even if its 100 W TDP and MXM form factor limit it to mobile platforms.

The head-to-head results show a split decision: the GTX 680M wins one test, the GTX 750 wins the other. The tiebreaker is context. If the workload favors Metal, the GTX 680M is superior. If it favors OpenCL or general efficiency, the GTX 750 holds the edge. Neither card is a universal winner, and the choice hinges entirely on the intended use case and platform constraints.

Specification Differences

The two GPUs differ across nearly every major specification category. The GTX 750 uses the GM107 chip on a 148 mm² die with 1,870 million transistors, while the GTX 680M uses the GK104 chip on a 294 mm² die with 3,540 million transistors. Clock speeds favor the GTX 750, with a base of 1020 MHz and boost of 1085 MHz, versus the GTX 680M’s 719 MHz base and 758 MHz boost. Memory differs in capacity, bus width, and bandwidth: the GTX 680M has 4 GB on a 256-bit bus with 115.2 GB/s, while the GTX 750 has 1024 MB on a 128-bit bus with 80.19 GB/s.

Core counts are dramatically different, with the GTX 680M featuring 1344 shading units, 112 TMUs, and 32 ROPs, compared to the GTX 750’s 512 shading units, 32 TMUs, and 16 ROPs. Pixel and texture rates follow suit: the GTX 680M achieves 21.22 GPixel/s and 84.90 GTexel/s, while the GTX 750 manages 17.36 GPixel/s and 34.72 GTexel/s. FP32 performance is 2.038 TFLOPS for the GTX 680M versus 1,111.0 GFLOPS for the GTX 750.

Power and physical specifications also diverge. The GTX 750 has a 55 W TDP, is single-slot, and requires no power connectors, with a suggested PSU of 250 W. The GTX 680M has a 100 W TDP and is an MXM module with no suggested PSU listed. Bus interfaces differ: the GTX 750 uses PCIe 3.0 x16, while the GTX 680M uses MXM-B (3.0). Display outputs are another contrast, with the GTX 750 offering 2x DVI and 1x mini-HDMI 1.4a, whereas the GTX 680M’s outputs are portable device dependent. The GTX 750 supports Vulkan 1.4, the GTX 680M only Vulkan 1.2.175. Finally, the GTX 750 was released on 2014-02-17, following the GeForce 600 and preceding the GeForce 900, while the GTX 680M launched on 2012-06-03, following the GeForce 500M and preceding the GeForce 700M.

Head-to-Head Benchmarks

The two shared benchmarks reveal a clear split. In Geekbench Metal, the GTX 680M wins with a score of 4815 against the GTX 750’s 4274, a delta of -11.2% for the desktop card. This is the most substantial performance gap between the two, suggesting that the GTX 680M’s higher core count and memory bandwidth provide a tangible benefit in Metal-optimized workloads. The 4 GB memory capacity likely helps with larger frame buffers, and the 256-bit bus reduces bandwidth bottlenecks.

In Geekbench OpenCL, the GTX 750 reverses the result, scoring 9315 versus the GTX 680M’s 9230, a slender 0.9% margin. This near-tie is remarkable given the GTX 680M’s 2.5x advantage in shading units and 1.8x advantage in FP32 throughput. The GTX 750’s victory in OpenCL can be attributed to its higher clock speeds—1085 MHz boost versus 758 MHz—and the architectural efficiency of Maxwell over Kepler. The 55 W TDP also suggests better power management, which can influence sustained performance in compute tasks.

Overall, the GTX 750 wins one benchmark and the GTX 680M wins the other, resulting in a 1-1 tie in head-to-head results. The GTX 750’s average benchmark score of 7222 is 2.8% higher than the GTX 680M’s 7023, but the GTX 680M’s percentile ranking of 39 versus 40 is nearly identical. This indicates that in real-world mixed workloads, the two cards perform at a similar level, with the choice depending more on platform and software optimization than raw capability.

Where Each One Wins

The GTX 750 emerges as the winner in scenarios prioritizing power efficiency and modern API support. Its 55 W TDP eliminates the need for power connectors, making it ideal for small form factor desktops or upgrades to pre-built systems with limited power supplies. The Maxwell architecture’s efficiency is reflected in its higher average benchmark score of 7222, achieved with less than half the transistors of the GTX 680M. The GTX 750 also supports Vulkan 1.4, a more recent version than the GTX 680M’s Vulkan 1.2.175, which future-proofs it for newer games and applications. Its 145 mm length and single-slot design make it a versatile fit for compact cases.

The GTX 680M wins in scenarios requiring maximum memory capacity and raw compute power. Its 4 GB VRAM is four times larger than the GTX 750’s 1024 MB, and its 256-bit bus delivers 115.2 GB/s bandwidth—a 43.7% improvement over the GTX 750. This makes it better suited for high-resolution gaming or compute tasks that exceed the GTX 750’s memory limit. The GTX 680M’s 1344 shading units and 2.038 TFLOPS FP32 provide superior throughput for parallel workloads, and its 84.90 GTexel/s texture rate is more than double the GTX 750’s. In the Metal benchmark, the GTX 680M’s 11.2% lead demonstrates its strength in Apple’s ecosystem, making it the preferred choice for Mac users or developers targeting Metal. Its MXM form factor also makes it the only option for laptop upgrades, despite the 100 W TDP.

In summary, the GTX 750 is the pick for desktop builders who value efficiency, modern APIs, and a compact footprint. The GTX 680M is the pick for mobile users or those needing large memory pools and peak compute performance, particularly in Metal-based applications. The data does not crown a single champion; it presents a balanced trade-off where each card excels in its designated environment.