AMD Radeon R7 M380 vs NVIDIA GeForce GTX 950M Comparison

The Verdict

The recorded data presents a clear but narrow overall picture. In the only shared benchmark, Geekbench OpenCL, the NVIDIA GeForce GTX 950M scores 9745 against the AMD Radeon R7 M380's 9313, a 4.4% advantage. This single result, however, masks a more complex comparison when considering each card's position within the broader database.

The AMD Radeon R7 M380 holds a higher percentile ranking at 46, compared to the NVIDIA GeForce GTX 950M's 42. Its average benchmark score of 9313 is also substantially higher than the GTX 950M's 8135. This discrepancy is explained by the database's inclusion of an additional Vulkan test for the NVIDIA card, where it scores 6525, which drags its average down. Therefore, the GTX 950M is the clear winner in the specific OpenCL workload they share, but the M380 shows a stronger overall benchmark profile across the tests recorded for each.

From a practical standpoint, the data suggests that for users prioritizing OpenCL compute performance, the GTX 950M is the superior choice. For those looking at a broader average of recorded benchmark scores, the Radeon R7 M380 appears more consistent. The GTX 950M also offers a higher pixel rate and texture rate, indicating potentially better raw rasterization throughput. The choice depends on the specific workload: the GTX 950M for OpenCL, and the M380 if considering the aggregate of all its recorded benchmarks.

Where Each One Wins

The benchmark data grants each GPU a distinct, though asymmetric, claim to victory. The NVIDIA GeForce GTX 950M decisively wins the only head-to-head test, the Geekbench OpenCL benchmark. Its score of 9745 surpasses the AMD Radeon R7 M380's 9313 by 4.4%. This is a direct, measured performance advantage in a compute-centric application.

The AMD Radeon R7 M380's "win" is more nuanced and derived from database aggregates. Its average benchmark score is 9313, which is 14.5% higher than the GTX 950M's average of 8135. This difference is not due to superior performance in a shared test, but rather because the NVIDIA card's average is lowered by its additional Geekbench Vulkan score of 6525, a test the AMD card does not have recorded. Furthermore, the M380's percentile rank of 46 places it above the GTX 950M's 42, indicating it outperforms a larger percentage of all GPUs in the database based on its recorded average.

In terms of theoretical throughput, the GTX 950M also holds the lead. Its pixel rate of 17.98 GPixel/s and texture rate of 44.96 GTexel/s are higher than the M380's 14.64 GPixel/s and 36.60 GTexel/s, respectively. This suggests the NVIDIA part has a higher ceiling for fill-rate-bound tasks. The M380's only clear hardware advantage is memory bandwidth, with 32.00 GB/s versus the GTX 950M's 28.80 GB/s.

Architecture Differences

The two GPUs represent fundamentally different architectural designs from their respective manufacturers, built on the same manufacturing process. Both are fabricated by TSMC on a 28 nm node, but their internal designs diverge significantly.

The AMD Radeon R7 M380 is built on the Graphics Core Next 1.0 architecture, using the Tropo chip. It packs 1,500 million transistors onto a 123 mm² die, resulting in a transistor density of 12.2 million per mm². The NVIDIA GeForce GTX 950M, in contrast, uses the Maxwell architecture with the GM107 chip. This chip contains 1,870 million transistors on a larger 148 mm² die, achieving a slightly higher transistor density of 12.6 million per mm².

Both GPUs feature identical core configurations for shading, texturing, and rasterization: 640 shading units, 40 texture mapping units, and 16 render output units. The key architectural difference lies in their clock speeds and memory subsystems. The GTX 950M operates at a base clock of 993 MHz with a boost clock of 1124 MHz, significantly higher than the M380's 900 MHz base and 915 MHz boost. This clock advantage directly translates into its higher compute throughput: 1,438.7 GFLOPS for NVIDIA versus 1,171.2 GFLOPS for AMD.

Memory configurations are similar in capacity and type, with both using 4 GB of DDR3 on a 128-bit bus. However, the AMD part runs its memory at 1000 MHz (2 Gbps effective), yielding a bandwidth of 32.00 GB/s. The NVIDIA part's memory runs at 900 MHz (1800 Mbps effective), resulting in a lower bandwidth of 28.80 GB/s. The GTX 950M also has a distinct power profile, rated at 75 W TDP, while the M380's TDP is not recorded. The bus interface differs as well, with the M380 using a full PCIe 3.0 x16 link and the GTX 950M using a PCIe 3.0 x8 link.

FAQ

Q: Which GPU is faster in the Geekbench OpenCL test?

A: The NVIDIA GeForce GTX 950M is faster, scoring 9745 compared to the AMD Radeon R7 M380's 9313, a 4.4% difference.

Q: Why does the AMD Radeon R7 M380 have a higher average benchmark score?

A: The M380's average is 9313, while the GTX 950M's is 8135. This is because the database includes a Geekbench Vulkan score of 6525 for the NVIDIA card, which is not recorded for the AMD card and lowers its average.

Q: Which GPU has a higher memory bandwidth?

A: The AMD Radeon R7 M380 has a higher memory bandwidth at 32.00 GB/s, compared to the NVIDIA GeForce GTX 950M's 28.80 GB/s, despite both using 4 GB of DDR3 on a 128-bit bus.

Q: Do the two GPUs have the same number of shading units?

A: Yes, both the AMD Radeon R7 M380 and the NVIDIA GeForce GTX 950M have exactly 640 shading units, along with 40 texture mapping units and 16 render output units.

Q: What are the architecture and process node for each GPU?

A: The AMD Radeon R7 M380 uses the GCN 1.0 architecture on a 28 nm process. The NVIDIA GeForce GTX 950M uses the Maxwell architecture, also on a 28 nm process.

Q: Which GPU has a higher pixel fill rate?

A: The NVIDIA GeForce GTX 950M has a higher pixel rate of 17.98 GPixel/s, while the AMD Radeon R7 M380 achieves 14.64 GPixel/s.

Head-to-Head Benchmarks

The single recorded head-to-head benchmark is the Geekbench OpenCL test, and it provides a definitive, if narrow, result. The NVIDIA GeForce GTX 950M wins this matchup with a score of 9745 against the AMD Radeon R7 M380's 9313. The delta is 4.4%, placing the NVIDIA card ahead. This is the only direct comparison available, so it carries significant weight for OpenCL compute workloads.

This OpenCL result aligns with the theoretical compute figures. The GTX 950M's FP32 performance is 1,438.7 GFLOPS, which is 22.8% higher than the M380's 1,171.2 GFLOPS. The higher clock speeds of the NVIDIA chip (993 MHz base, 1124 MHz boost) versus the AMD chip (900 MHz base, 915 MHz boost) are the primary drivers of this advantage, and the benchmark score reflects this gap, though the 4.4% benchmark delta is smaller than the theoretical FP32 delta.

The GTX 950M also demonstrates superiority in rasterization throughput. Its texture rate of 44.96 GTexel/s is 22.8% higher than the M380's 36.60 GTexel/s, and its pixel rate of 17.98 GPixel/s is 22.8% higher than the M380's 14.64 GPixel/s. These figures suggest that in texture-heavy or fill-rate-bound scenarios, the GTX 950M would have a clear edge, consistent with its OpenCL victory.

The AMD Radeon R7 M380's advantages are confined to memory bandwidth and database ranking. Its 32.00 GB/s bandwidth is 11.1% higher than the GTX 950M's 28.80 GB/s, which could benefit memory-intensive workloads, but this does not translate into a win in the recorded OpenCL test. Its higher percentile rank (46 vs. 42) and higher average score (9313 vs. 8135) are artifacts of the test suite differences. The M380's average is based solely on its strong OpenCL score, while the GTX 950M's average is pulled down by its additional, lower Vulkan score of 6525. Therefore, the head-to-head data clearly favors the GTX 950M in the shared workload, while the M380's other metrics offer context but not a competitive win.