AMD Radeon HD 8750M vs NVIDIA GeForce GTX 680M Comparison

# Where Each One Wins

The recorded benchmark data shows a single head-to-head comparison between the NVIDIA GeForce GTX 680M and the AMD Radeon HD 8750M, and it is a clean sweep for the NVIDIA part. The GTX 680M wins the only shared test, Geekbench OpenCL, with a score of 9230 against 5970 for the AMD Radeon HD 8750M, a margin of 54.6%. That is the sole direct comparison in the database, so the use-case split is dictated by that result: for any workload that relies on OpenCL compute throughput, the NVIDIA GeForce GTX 680M is the stronger option by a substantial margin.

The AMD Radeon HD 8750M has no recorded win in any benchmark category, which means it does not claim an advantage in compute tasks, memory-bound workloads, or any other test present in our measurements. Its average benchmark score sits at 5970, which places it at the 34th percentile among all GPUs in the database. The NVIDIA GeForce GTX 680M, by contrast, averages 7023 across its recorded tests and sits at the 39th percentile. The percentile gap is modest, but the average score gap is 17.6% in favor of the NVIDIA part when comparing their overall averages directly (7023 versus 5970).

Looking at the nearest rivals for each card clarifies the competitive positioning. The GTX 680M is bracketed by the NVIDIA T600 at 7035, the NVIDIA GeForce GTX 970 at 7157, the AMD Radeon R5 M240 at 6975, and the NVIDIA GeForce GTX 675M at 6946. Its average score of 7023 is within 0.2% of the T600 and 1.1% ahead of the GTX 675M, while trailing the GTX 970 by 1.9%. The Radeon HD 8750M, meanwhile, sits near the NVIDIA Quadro K4000 at 5982, the NVIDIA Quadro K4000M at 5986, the NVIDIA Quadro K620M at 5957, and the AMD Radeon HD 8730M at 5955. The 8750M is 0.2% ahead of the K620M and 0.3% ahead of the HD 8730M, and it trails the K4000 by 0.2% and the K4000M by 0.3%.

In practical terms, the only benchmark that exercises both cards identically is OpenCL, and that test favors the GTX 680M overwhelmingly. There is no gaming benchmark, no DirectX test, and no Vulkan score in the shared database, so any use-case split must be inferred from the compute results and the architectural data available. For users prioritizing general compute acceleration, the NVIDIA part is the clear pick.

# Architecture Differences

The two GPUs come from different architectural families, and the differences are visible in every major structural category. The NVIDIA GeForce GTX 680M is built on the GK104 chip using the Kepler architecture, manufactured on a 28 nm process at TSMC. The die contains 3,540 million transistors on a 294 mm² die, giving a transistor density of 12.0 million transistors per mm². The AMD Radeon HD 8750M uses the Mars chip with the GCN 1.0 architecture, also on a 28 nm process at TSMC, but the die is far smaller at 77 mm² with 950 million transistors, for a density of 12.3 million transistors per mm². The density figures are nearly identical, which reflects the shared process node, but the physical scale of the chips is very different.

The execution resources differ substantially. The GTX 680M has 1344 shading units, 112 texture mapping units, and 32 ROPs. The Radeon HD 8750M has 384 shading units, 24 texture mapping units, and 8 ROPs. That is a 3.5x advantage in shading units, a 4.67x advantage in TMUs, and a 4x advantage in ROPs for the NVIDIA card. These ratios translate directly into throughput limits: the GTX 680M reaches a pixel rate of 21.22 GPixel/s and a texture rate of 84.90 GTexel/s, while the Radeon HD 8750M manages 6.600 GPixel/s and 19.80 GTexel/s. The FP32 compute throughput is 2.038 TFLOPS for the GTX 680M versus 633.6 GFLOPS for the AMD part, a factor of roughly 3.2x.

Memory subsystems also diverge sharply. The GTX 680M is equipped with 4 GB of GDDR5 memory on a 256 bit bus, delivering 115.2 GB/s of bandwidth. The Radeon HD 8750M has 1024 MB of DDR3 memory on a 128 bit bus, yielding only 28.80 GB/s. The effective memory clock for the GTX 680M is stated as 3.6 Gbps, while the AMD part runs at 1800 Mbps effective. The NVIDIA card therefore has a 4x bus width advantage and a 4x bandwidth advantage, which matters for any workload that streams large data sets.

Clock speeds tell a different story. The Radeon HD 8750M runs at a base clock of 775 MHz with an 825 MHz boost, whereas the GTX 680M runs at a lower 719 MHz base and 758 MHz boost. The AMD chip is clocked about 7.8% higher at base and 8.8% higher at boost, but the massive resource and memory advantages of the Kepler part overwhelm the clock deficit. The GTX 680M is rated at 100 W TDP and uses an MXM module form factor with an MXM-B (3.0) bus interface, while the Radeon HD 8750M has no TDP listed and connects over PCIe 3.0 x8.

Both cards support DirectX 12, with the GTX 680M exposing version 12 (11_0) and the Radeon HD 8750M exposing 12 (11_1). Both list OpenGL 4.6. Vulkan support is present on both, with the GTX 680M at version 1.2.175 and the Radeon HD 8750M at 1.2.170, a negligible difference. Neither card has RT cores or tensor cores. The GTX 680M is part of the GeForce 600M generation, while the Radeon HD 8750M belongs to the Solar System (HD 8700M) generation. The NVIDIA part was released on 2012-06-03, and the AMD part on 2013-02-25. Both are end-of-life products.

# Head-to-Head Benchmarks

The database contains exactly one head-to-head benchmark between these two cards. In Geekbench OpenCL, the NVIDIA GeForce GTX 680M scores 9230, and the AMD Radeon HD 8750M scores 5970. The delta is 54.6%, meaning the GTX 680M outperforms the Radeon HD 8750M by more than half of the AMD card's own score. In absolute terms, the NVIDIA part is 3260 points ahead. This is a dominant result, and it aligns with the structural advantages in shading units, memory bandwidth, and FP32 throughput described above.

The average benchmark scores in the database reinforce the same conclusion from a broader perspective. The GTX 680M has an average score of 7023 across its recorded tests, which includes a Geekbench Metal score of 4815 in addition to the OpenCL score of 9230. The Radeon HD 8750M has only a single recorded test, the OpenCL score of 5970, so its average equals that number. The GTX 680M's average is 17.6% higher than the Radeon HD 8750M's average, even when the Metal result, which is lower than the OpenCL result, is included in the NVIDIA average.

Comparing each card to its nearest rivals provides additional context for interpreting these numbers. The GTX 680M's 7023 average places it slightly below the NVIDIA T600 (7035, delta of -0.2%) and the NVIDIA GeForce GTX 970 (7157, delta of -1.9%), but ahead of the AMD Radeon R5 M240 (6975, delta of 0.7%) and the NVIDIA GeForce GTX 675M (6946, delta of 1.1%). This shows the GTX 680M is competitive with a much newer professional card, the T600, and sits just below a desktop-class GTX 970 in this particular metric. The Radeon HD 8750M's 5970 average places it in a lower tier: it is essentially tied with the NVIDIA Quadro K4000 (5982, delta of -0.2%) and the NVIDIA Quadro K4000M (5986, delta of -0.3%), while edging out the NVIDIA Quadro K620M (5957, delta of 0.2%) and the AMD Radeon HD 8730M (5955, delta of 0.3%). The delta percentages among the Radeon HD 8750M's rivals are all within 0.3%, indicating a very tight cluster of performance at that level.

What this means for the head-to-head picture is that the two cards occupy entirely different performance tiers. The GTX 680M is not just ahead of the Radeon HD 8750M; it is competing with cards that score in the low 7000s, while the Radeon HD 8750M sits in the high 5000s. The 54.6% delta in the direct OpenCL comparison is the single most important data point, and it is consistent with the overall average scores.

# FAQ

Q: Which GPU has the higher average benchmark score?

A: The NVIDIA GeForce GTX 680M has an average benchmark score of 7023, while the AMD Radeon HD 8750M has an average of 5970. The GTX 680M is 17.6% higher.

Q: How much faster is the GTX 680M in the shared OpenCL test?

A: In Geekbench OpenCL, the GTX 680M scores 9230 against 5970 for the Radeon HD 8750M, a delta of 54.6% in favor of the NVIDIA card.

Q: Do both GPUs support DirectX 12?

A: Yes, both cards support DirectX 12, but with different feature levels. The GTX 680M supports version 12 (11_0), and the Radeon HD 8750M supports version 12 (11_1).

Q: What are the memory configurations of the two cards?

A: The GTX 680M has 4 GB of GDDR5 memory on a 256 bit bus with 115.2 GB/s bandwidth. The Radeon HD 8750M has 1024 MB of DDR3 memory on a 128 bit bus with 28.80 GB/s bandwidth.

Q: Which card has more shading units?

A: The GTX 680M has 1344 shading units, compared to 384 on the Radeon HD 8750M, a 3.5x advantage for the NVIDIA part.

Q: How do the two cards compare to their nearest rivals?

A: The GTX 680M is within 0.2% of the NVIDIA T600 and 1.1% ahead of the GeForce GTX 675M. The Radeon HD 8750M is within 0.3% of the NVIDIA Quadro K4000, K4000M, K620M, and the AMD Radeon HD 8730M.

# Specification Differences

The following fields differ between the NVIDIA GeForce GTX 680M and the AMD Radeon HD 8750M. The GTX 680M uses the GK104 chip with Kepler architecture, while the Radeon HD 8750M uses the Mars chip with GCN 1.0 architecture. The GTX 680M is in the GeForce 600M generation, and the Radeon HD 8750M is in the Solar System (HD 8700M) generation. Transistor count is 3,540 million versus 950 million, and die size is 294 mm² versus 77 mm². Transistor density is 12.0M per mm² versus 12.3M per mm².

Clock speeds differ: base clock is 719 MHz versus 775 MHz, boost clock is 758 MHz versus 825 MHz. Effective memory speed is 3.6 Gbps versus 1800 Mbps. Memory size is 4 GB versus 1024 MB, memory type is GDDR5 versus DDR3, bus width is 256 bit versus 128 bit, and bandwidth is 115.2 GB/s versus 28.80 GB/s. Shading units are 1344 versus 384, TMUs are 112 versus 24, and ROPs are 32 versus 8. Pixel rate is 21.22 GPixel/s versus 6.600 GPixel/s, texture rate is 84.90 GTexel/s versus 19.80 GTexel/s, and FP32 is 2.038 TFLOPS versus 633.6 GFLOPS.

TDP is 100 W for the GTX 680M, with no TDP listed for the Radeon HD 8750M. Slot width is MXM Module for the NVIDIA card, with no slot width listed for the AMD card. Power connectors are listed as none for the GTX 680M, with no power connector data for the Radeon HD 8750M. Bus interface is MXM-B (3.0) for the GTX 680M and PCIe 3.0 x8 for the Radeon HD 8750M. Display outputs are listed as portable device dependent for the GTX 680M, with no display output data for the AMD card. DirectX support is 12 (11_0) versus 12 (11_1), and Vulkan support is 1.2.175 versus 1.2.170. Release dates are 2012-06-03 for the GTX 680M and 2013-02-25 for the Radeon HD 8750M. The GTX 680M lists predecessor GeForce 500M and successor GeForce 700M, while the Radeon HD 8750M lists predecessor London and successor Gem System.

# The Verdict

The data points to one clear conclusion: the NVIDIA GeForce GTX 680M is the superior GPU in every measured category. It wins the only direct head-to-head benchmark, Geekbench OpenCL, with a 54.6% margin. Its average benchmark score of 7023 is 17.6% higher than the Radeon HD 8750M's 5970. Its architectural resources are overwhelmingly larger: 1344 shading units versus 384, 112 TMUs versus 24, 32 ROPs versus 8, and a memory bandwidth of 115.2 GB/s versus 28.80 GB/s. The GTX 680M also carries 4 GB of GDDR5 against 1024 MB of DDR3, which matters for large working sets.

There is no benchmark category in the database where the AMD Radeon HD 8750M wins. Its only recorded test is the OpenCL score, and it loses that test decisively. Its nearest rivals are clustered in the 5955 to 5986 range, which places it in a lower performance tier than the GTX 680M's rivals, which cluster in the 6946 to 7157 range. The Radeon HD 8750M does have a slightly higher boost clock (825 MHz versus 758 MHz) and a marginally higher transistor density (12.3M per mm² versus 12.0M per mm²), but neither advantage translates into a benchmark win or even a competitive score.

For a user choosing between these two parts, the decision should be based on the compute performance data. The GTX 680M is the choice for anyone who needs OpenCL throughput, higher memory bandwidth, or the larger framebuffer. The Radeon HD 8750M offers no measured advantage in any shared test. The GTX 680M is also positioned closer to newer professional cards like the NVIDIA T600, within 0.2%, while the Radeon HD 8750M is essentially tied with several Quadro parts from an older generation. The verdict is straightforward: the NVIDIA GeForce GTX 680M is the stronger option by every metric recorded in the database.