AMD Radeon R9 M375X vs NVIDIA GeForce GTX 680M Comparison

Head-to-Head Benchmarks

The database records only one shared benchmark test between the AMD Radeon R9 M375X and the NVIDIA GeForce GTX 680M: Geekbench OpenCL. In that test, the NVIDIA GeForce GTX 680M posts a score of 9230, while the AMD Radeon R9 M375X records 8273. This yields a delta of -10.4% for the AMD part, meaning the NVIDIA card is roughly a tenth faster in this single compute workload. It is a clean, if narrow, win for the older Kepler-based mobile GPU.

The AMD Radeon R9 M375X does not win any of the recorded head-to-head tests. Its 8273 OpenCL score sits in the middle of its own nearest rival cluster: it trails the NVIDIA GeForce GTX 675MX (8427) by 1.2%, trails the AMD Radeon 880M (8436) by 1.3%, and trails the NVIDIA GeForce MX330 (8458) by 1.6%. It only edges out the NVIDIA Quadro K1200 (8265) by 0.7%. This places the R9 M375X at the lower end of a tightly packed group of mobile and workstation GPUs, none of which are separated by more than about two percentage points.

The NVIDIA GeForce GTX 680M, by contrast, shows a different competitive picture. Its average benchmark score across all recorded tests is 7023, which is dragged down by a much weaker Geekbench Metal result of 4815. Its OpenCL score of 9230 is the stronger data point. Against its nearest rivals, the GTX 680M is nearly tied with the NVIDIA T600 (7035, delta -0.2%), slightly ahead of the AMD Radeon R5 M240 (6975, delta 0.7%), ahead of the NVIDIA GeForce GTX 675M (6946, delta 1.1%), and behind the NVIDIA GeForce GTX 970 (7157, delta -1.9%). The spread is again small, but the GTX 680M sits in a different league than the R9 M375X when comparing their respective peer groups.

Looking at the raw scores, the GTX 680M OpenCL result (9230) is 957 points higher than the R9 M375X (8273). That is a meaningful gap for two mobile GPUs separated by three years of release dates. The R9 M375X was released in May 2015, the GTX 680M in June 2012. The newer AMD chip cannot overcome the older NVIDIA part's larger silicon and wider memory interface in this compute test.

The percentile data reinforces the separation. The AMD Radeon R9 M375X sits at the 43rd percentile among all GPUs in the database, while the NVIDIA GeForce GTX 680M sits at the 39th percentile. This is counterintuitive at first glance: the NVIDIA card wins the head-to-head OpenCL test, yet its overall percentile is lower. The explanation lies in the Metal score. The GTX 680M's Metal result of 4815 is far below its OpenCL score, and that low outlier pulls down its average benchmark score to 7023. The R9 M375X has no Metal score recorded, only OpenCL (8273) and Vulkan (8377), which average to 8325. The AMD card's average is therefore higher, pushing it up the percentile ladder despite losing the one direct comparison.

The Verdict

The data points to a straightforward conclusion for compute-heavy OpenCL workloads: the NVIDIA GeForce GTX 680M is the faster GPU, with a 10.4% lead over the AMD Radeon R9 M375X in the only shared benchmark. Anyone prioritizing OpenCL compute performance should favor the GTX 680M, even though it is an older design.

However, the overall averages tell a different story. The AMD Radeon R9 M375X has an average benchmark score of 8325, which is 18.5% higher than the GTX 680M's 7023. This is almost entirely due to the NVIDIA card's weak Metal showing. The GTX 680M's Metal score of 4815 is roughly half its OpenCL score, an unusual disparity that suggests the Kepler architecture did not age well under Apple's Metal API. The R9 M375X, with no Metal benchmark recorded, avoids this penalty entirely.

The percentile ranks reflect this: the R9 M375X at the 43rd percentile outranks the GTX 680M at the 39th. So the verdict depends on which metric matters more. For OpenCL specifically, choose the GTX 680M. For a balanced multi-API average, the R9 M375X looks better in the database, though this is partly a function of missing data rather than genuine superiority.

The production status of both chips is end-of-life, so neither is a forward-looking purchase. The GTX 680M's launch MSRP is not recorded in the database, and neither is the R9 M375X's, so no cost comparison is possible from the available facts. What the data does show is a generational clash: a 2012 flagship mobile GPU with a 294 mm² die and 3,540 million transistors versus a 2015 midrange mobile GPU with a 123 mm² die and 1,500 million transistors. The larger, older chip still wins the compute test, which says something about how much raw silicon matters even as architectures advance.

Where Each One Wins

The NVIDIA GeForce GTX 680M wins the only direct head-to-head benchmark, Geekbench OpenCL, with a score of 9230 against 8273. This is the clear, singular victory for the NVIDIA part. It also has a higher pixel rate (21.22 GPixel/s versus 16.24 GPixel/s), a higher texture rate (84.90 GTexel/s versus 40.60 GTexel/s), and roughly 57% more FP32 throughput (2.038 TFLOPS versus 1,299.2 GFLOPS). These are theoretical specifications, but they align with the OpenCL result: the GTX 680M has more shading units (1344 versus 640), more texture mapping units (112 versus 40), and more ROPs (32 versus 16). It also doubles the memory bandwidth (115.2 GB/s versus 72.00 GB/s) on a 256-bit bus versus 128-bit, and offers 4 GB of GDDR5 versus 2 GB.

The AMD Radeon R9 M375X wins the average benchmark score comparison, 8325 versus 7023, and holds the higher percentile rank at 43 versus 39. It also wins the Vulkan API version comparison: the R9 M375X supports Vulkan 1.2.170, while the GTX 680M supports Vulkan 1.2.175, a minor but real difference in API versioning. In OpenCL, the AMD card supports DirectX 12 (11_1), which is one small step above the GTX 680M's DirectX 12 (11_0). The AMD chip also has a lower TDP on paper is not recorded, so no power comparison can be made from the data.

The use-case split is therefore clean. For raw compute throughput, texture-heavy rendering, and memory-intensive workloads, the GTX 680M has the hardware advantage and the benchmark win to back it up. For a more consistent multi-API average and a higher overall percentile, the R9 M375X comes out ahead, largely because its Vulkan score (8377) is close to its OpenCL score (8273), while the GTX 680M's Metal score (4815) is a severe outlier.

FAQ

Q: Which GPU has the higher Geekbench OpenCL score?

A: The NVIDIA GeForce GTX 680M scores 9230, while the AMD Radeon R9 M375X scores 8273. The NVIDIA card leads by 10.4%.

Q: Why does the AMD Radeon R9 M375X have a higher average benchmark score despite losing the OpenCL test?

A: The R9 M375X averages 8325 across its two recorded tests (OpenCL 8273 and Vulkan 8377). The GTX 680M averages 7023 across its two recorded tests (Metal 4815 and OpenCL 9230). The NVIDIA card's Metal score is far lower than its OpenCL score, which drags down its average.

Q: How does the GTX 680M compare to its nearest rivals?

A: The GTX 680M is nearly tied with the NVIDIA T600 (7035, delta -0.2%), slightly ahead of the AMD Radeon R5 M240 (6975, delta 0.7%) and the NVIDIA GeForce GTX 675M (6946, delta 1.1%), and behind the NVIDIA GeForce GTX 970 (7157, delta -1.9%).

Q: What is the percentile ranking of each GPU?

A: The AMD Radeon R9 M375X is at the 43rd percentile among all GPUs, while the NVIDIA GeForce GTX 680M is at the 39th percentile.

Q: Which GPU has more memory and bandwidth?

A: The GTX 680M has 4 GB of GDDR5 on a 256-bit bus with 115.2 GB/s bandwidth. The R9 M375X has 2 GB of GDDR5 on a 128-bit bus with 72.00 GB/s bandwidth.

Q: What are the release dates for these two GPUs?

A: The AMD Radeon R9 M375X was released on May 4, 2015. The NVIDIA GeForce GTX 680M was released on June 3, 2012.

Architecture Differences

The two GPUs come from different designers and different architectural generations. The AMD Radeon R9 M375X uses the Tropo chip built on GCN 1.0 architecture, part of the Gem System (R9 M300) generation. The NVIDIA GeForce GTX 680M uses the GK104 chip built on Kepler architecture, part of the GeForce 600M generation. Both are manufactured by TSMC on a 28 nm process, which is where the similarities end.

The transistor counts differ dramatically. The Tropo chip contains 1,500 million transistors on a 123 mm² die, giving a transistor density of 12.2 million per square millimeter. The GK104 chip contains 3,540 million transistors on a 294 mm² die, giving a density of 12.0 million per square millimeter. The density is nearly identical, which makes sense for the same process node, but the NVIDIA chip is more than twice the physical size and houses more than twice the transistor count. The AMD chip achieves its density on a much smaller die, indicating a more modest design.

The compute configurations diverge sharply. The R9 M375X has 640 shading units, 40 texture mapping units, and 16 ROPs. The GTX 680M has 1344 shading units, 112 texture mapping units, and 32 ROPs. The NVIDIA part has more than double the shading units, nearly three times the TMUs, and double the ROPs. This explains the large differences in pixel rate (16.24 GPixel/s versus 21.22 GPixel/s) and texture rate (40.60 GTexel/s versus 84.90 GTexel/s).

Neither GPU has ray tracing cores or tensor cores, and neither has a recorded FP16 throughput. Both support DirectX 12, but with different feature levels: the R9 M375X supports 12 (11_1), while the GTX 680M supports 12 (11_0). Both support OpenGL 4.6. The Vulkan versions differ slightly: 1.2.170 for AMD, 1.2.175 for NVIDIA.

The memory subsystems are also distinct. The R9 M375X uses 2 GB of GDDR5 on a 128-bit bus with a bandwidth of 72.00 GB/s. The GTX 680M uses 4 GB of GDDR5 on a 256-bit bus with a bandwidth of 115.2 GB/s. The NVIDIA card has double the capacity and 60% more bandwidth. The memory clocks differ as well: the R9 M375X runs at 1125 MHz (4.5 Gbps effective), while the GTX 680M runs at 900 MHz (3.6 Gbps effective). The AMD card has a higher clock speed, but the NVIDIA card's wider bus wins on total bandwidth.

Specification Differences

The two GPUs differ in nearly every measurable specification. The R9 M375X has a base clock of 925 MHz and a boost clock of 1015 MHz, while the GTX 680M has a base clock of 719 MHz and a boost clock of 758 MHz. The AMD chip runs at higher clocks, but the NVIDIA chip compensates with far more execution units.

The memory configurations differ as noted: 2 GB versus 4 GB, 128-bit versus 256-bit, 72.00 GB/s versus 115.2 GB/s. The memory types are both GDDR5, but the effective data rates differ (4.5 Gbps versus 3.6 Gbps).

The FP32 throughput is 1,299.2 GFLOPS for the R9 M375X and 2.038 TFLOPS for the GTX 680M. The pixel rates are 16.24 GPixel/s versus 21.22 GPixel/s, and the texture rates are 40.60 GTexel/s versus 84.90 GTexel/s. The NVIDIA card wins all three throughput metrics.

The bus interfaces differ: the R9 M375X uses PCIe 3.0 x16, while the GTX 680M uses MXM-B (3.0). The GTX 680M also has a TDP of 100 W, whereas the R9 M375X has no TDP recorded. The GTX 680M is described as an MXM Module with no power connectors and portable device dependent display outputs. The R9 M375X has no slot width, power connector, or display output information recorded.

The release dates are three years apart: May 2015 for the AMD part, June 2012 for the NVIDIA part. The predecessor and successor chains differ too: the R9 M375X follows the Solar System generation and precedes Polaris Mobile, while the GTX 680M follows GeForce 500M and precedes GeForce 700M. Both are end-of-life products. Neither has a recorded launch MSRP, so no pricing data exists in the database.