AMD Radeon R9 M380 vs NVIDIA GeForce GTX 680 Comparison

The AMD Radeon R9 M380 and NVIDIA GeForce GTX 680 represent two distinct approaches to mobile and desktop graphics, respectively, with the data revealing a clear split in workload strengths. While the GTX 680 holds a commanding lead in OpenCL compute tasks, the R9 M380 dominates in Metal performance, ultimately making the choice dependent on the user's specific software ecosystem and performance priorities. The benchmark results show a near-tie in overall win count, but the magnitude of the victories in each test tells a more nuanced story.

Head-to-Head Benchmarks

The most striking result in the head-to-head data is the Geekbench Metal comparison, where the AMD Radeon R9 M380 scores 18,476 against the GTX 680's 7,897. This represents a 134% performance advantage for the AMD part, an enormous margin that highlights a fundamental strength in Apple's Metal API. This score is not merely a slight edge; it is a decisive victory that positions the R9 M380 as a far more capable option for Metal-based applications, including many modern games and professional creative tools. In this test, the GTX 680 is left far behind, and no amount of architectural refinement on the NVIDIA side can close this gap.

Conversely, the Geekbench OpenCL test flips the script entirely. Here, the NVIDIA GeForce GTX 680 scores 16,906, while the AMD Radeon R9 M380 manages 12,565. The delta is -25.7% in favor of the GTX 680, meaning the NVIDIA card is roughly a quarter faster in this compute workload. This is a substantial lead, though not as dramatic as the Metal result. The GTX 680's advantage in OpenCL suggests it handles general-purpose compute tasks more efficiently, which could be relevant for users running OpenCL-accelerated applications like video encoding or scientific simulations.

Looking at the broader benchmark context, the average benchmark score for the R9 M380 is 15,521, placing it in the 58th percentile of all GPUs. The GTX 680's average is 14,150, putting it in the 55th percentile. While the R9 M380 has a higher average, the head-to-head results are split 1-1, indicating that the average is heavily influenced by the massive Metal win. The GTX 680's nearest rivals include the NVIDIA Tesla K10 (average score 14,029, a 0.9% delta) and the GeForce GTX 1070 Ti (average score 14,277, a -0.9% delta), showing it sits in a competitive mid-range compute bracket. The R9 M380, meanwhile, is nearly tied with the GTX 1080 Ti (average score 15,548, a -0.2% delta) and the Radeon Pro W5500 (average score 15,679, a -1% delta), indicating its average performance is artificially boosted by its Metal dominance.

The Verdict

The data supports a clear verdict based on software environment. For users whose primary workload involves Metal, the AMD Radeon R9 M380 is the unequivocal choice. Its 134% lead in the Metal benchmark is a categorical advantage that cannot be overlooked. Any application leveraging Metal will see dramatic performance improvements on the R9 M380, making it the superior option for macOS or iOS development, Metal-based gaming, and creative applications that have adopted this API.

For users relying on OpenCL compute, the NVIDIA GeForce GTX 680 is the better performer. Its 25.7% lead in the OpenCL test demonstrates a robust compute architecture that excels in this specific API. The GTX 680 also offers a higher FP32 throughput at 3.250 TFLOPS compared to the R9 M380's 1.536 TFLOPS, reinforcing its raw compute capabilities. This makes it a more suitable option for OpenCL-accelerated tasks, even though it is an older design.

The overall average benchmark score favors the R9 M380 (15,521 vs 14,150), but this is a misleading statistic given the split nature of the results. Users should not choose based solely on the average; instead, they must identify which API their applications use. The GTX 680 also has the advantage of a longer track record and a known launch MSRP of 499 USD, though this is a historical figure. In summary, the R9 M380 wins on Metal, the GTX 680 wins on OpenCL, and the tie in wins (1-1) reflects a genuine performance split rather than a tie in overall capability.

Architecture Differences

The two GPUs are built on fundamentally different architectures from their respective manufacturers. The AMD Radeon R9 M380 uses the Strato chip based on the GCN 2.0 architecture, while the NVIDIA GeForce GTX 680 uses the GK104 chip based on the Kepler architecture. Both are manufactured on a 28 nm process at TSMC, but their transistor budgets differ significantly. The GTX 680 packs 3,540 million transistors on a 294 mm² die, while the R9 M380 has 2,080 million transistors on a 160 mm² die. This results in a slightly higher transistor density for the AMD chip (13.0M / mm² vs 12.0M / mm² for NVIDIA).

The core configurations are radically different. The GTX 680 features 1,536 shading units, 128 texture mapping units (TMUs), and 32 render output units (ROPs). In contrast, the R9 M380 has only 768 shading units, 48 TMUs, and 16 ROPs. This explains the GTX 680's higher theoretical pixel rate (33.86 GPixel/s vs 16.00 GPixel/s) and texture rate (135.4 GTexel/s vs 48.00 GTexel/s). The NVIDIA card is clearly designed for higher raw throughput, which aligns with its OpenCL advantage.

Memory configurations also diverge. The R9 M380 offers 4 GB of GDDR5 memory on a 128-bit bus, yielding 96.00 GB/s of bandwidth. The GTX 680 has 2 GB of GDDR5 on a 256-bit bus, providing 192.3 GB/s of bandwidth. The GTX 680's wider bus gives it double the memory bandwidth, which is critical for high-resolution textures and compute workloads. Clock speeds are similar, with the R9 M380 boosting to 1000 MHz and the GTX 680 boosting to 1058 MHz, but the memory clocks are nearly identical at 1500 MHz and 1502 MHz, respectively.

API support also differs. The R9 M380 supports DirectX 12 (12_0), while the GTX 680 only reaches DirectX 12 (11_0), indicating a feature-level advantage for AMD. Both support OpenGL 4.6, but the Vulkan versions differ slightly: the R9 M380 has 1.2.170 and the GTX 680 has 1.2.175. The GTX 680 has a known TDP of 195 W and requires a 450 W PSU, while the R9 M380's TDP is not specified, leaving its power requirements ambiguous.

FAQ

Q: Which GPU has a higher average benchmark score?

A: The AMD Radeon R9 M380 has an average benchmark score of 15,521, which is higher than the NVIDIA GeForce GTX 680's average of 14,150. This places the R9 M380 in the 58th percentile of all GPUs, versus the 55th percentile for the GTX 680.

Q: How do the two compare in Geekbench Metal?

A: The AMD Radeon R9 M380 is vastly superior in Geekbench Metal, scoring 18,476 compared to the GTX 680's 7,897. This is a 134% advantage for the AMD card, making it the clear winner for Metal-based applications.

Q: What about Geekbench OpenCL performance?

A: In Geekbench OpenCL, the NVIDIA GeForce GTX 680 takes the lead with a score of 16,906 versus the R9 M380's 12,565. This represents a -25.7% delta, meaning the GTX 680 is about a quarter faster in this API.

Q: Which card has more memory bandwidth?

A: The NVIDIA GeForce GTX 680 has significantly more memory bandwidth at 192.3 GB/s, thanks to its 256-bit bus. The AMD Radeon R9 M380 offers 96.00 GB/s from a 128-bit bus, which is exactly half the bandwidth.

Q: What are the transistor counts for each chip?

A: The NVIDIA GeForce GTX 680's GK104 chip contains 3,540 million transistors, while the AMD Radeon R9 M380's Strato chip contains 2,080 million transistors. The GTX 680 also has a larger die size at 294 mm² compared to 160 mm² for the R9 M380.

Q: Which GPU supports a higher DirectX feature level?

A: The AMD Radeon R9 M380 supports DirectX 12 (12_0), while the NVIDIA GeForce GTX 680 supports DirectX 12 (11_0). This gives the AMD card a feature-level advantage in DirectX 12 titles.

Where Each One Wins

The AMD Radeon R9 M380 wins decisively in Metal-based workloads. With a 134% lead in Geekbench Metal, this GPU is the obvious choice for users running Metal-accelerated games, creative software, or compute tasks on Apple platforms. Its higher average benchmark score (15,521) and superior percentile ranking (58th) also make it a more well-rounded performer in aggregate, though this is driven primarily by the Metal result. The R9 M380's larger 4 GB memory capacity also provides an advantage for applications that require more VRAM, even if the bandwidth is lower.

The NVIDIA GeForce GTX 680 wins in OpenCL compute and raw theoretical throughput. Its 25.7% lead in Geekbench OpenCL, combined with a higher FP32 rating of 3.250 TFLOPS (versus 1.536 TFLOPS for AMD), makes it a stronger candidate for general-purpose GPU computing. Its 192.3 GB/s of memory bandwidth is double the R9 M380's, which is crucial for bandwidth-sensitive tasks. The GTX 680 also has higher pixel and texture rates (33.86 GPixel/s and 135.4 GTexel/s, respectively), making it better suited for traditional rasterization workloads that don't rely on Metal. For users in the OpenCL ecosystem, the GTX 680's performance is simply more robust.

Specification Differences

The two GPUs differ across nearly every major specification category. The chip is different (Strato vs GK104), as is the architecture (GCN 2.0 vs Kepler). The transistor count is higher on the GTX 680 (3,540 million vs 2,080 million), and the die size is larger (294 mm² vs 160 mm²). The shading units favor NVIDIA heavily (1,536 vs 768), as do the TMUs (128 vs 48) and ROPs (32 vs 16). Clock speeds are close, with the GTX 680 having a slightly higher base (1006 MHz vs 900 MHz) and boost (1058 MHz vs 1000 MHz) clock.

Memory is a major differentiator: the R9 M380 has 4 GB on a 128-bit bus, while the GTX 680 has 2 GB on a 256-bit bus, resulting in 96.00 GB/s vs 192.3 GB/s of bandwidth. The FP32 performance is more than double on the GTX 680 (3.250 TFLOPS vs 1.536 TFLOPS), as are the pixel rate (33.86 GPixel/s vs 16.00 GPixel/s) and texture rate (135.4 GTexel/s vs 48.00 GTexel/s). The GTX 680 also has a specified TDP of 195 W, a dual-slot form factor, and 2x 6-pin power connectors, while the R9 M380 lacks these details. The DirectX support is higher on the R9 M380 (12_0 vs 11_0), but the Vulkan version is slightly higher on the GTX 680 (1.2.175 vs 1.2.170). Finally, the release dates differ significantly, with the GTX 680 launching in March 2012 and the R9 M380 in May 2015.