AMD Radeon R7 M370 vs NVIDIA GeForce GTX 950M Comparison

The Verdict

The NVIDIA GeForce GTX 950M is the clear choice for anyone who needs stronger compute performance in a laptop. Across the two recorded benchmark tests, it wins both, with a 38% advantage in Geekbench OpenCL and a narrower 0.9% edge in Geekbench Vulkan. Its average benchmark score of 8135 places it at the 42nd percentile of all GPUs, while the AMD Radeon R7 M370 sits at 6764 average, in the 38th percentile. If you prioritize raw throughput for OpenCL workloads or Vulkan API tasks, the GTX 950M is the safer pick.

The AMD Radeon R7 M370 is not without merit, but the data does not support choosing it over the NVIDIA part for raw score chasing. It loses both head-to-head tests, and its nearest rivals include the NVIDIA GeForce GT 1010 (1% behind) and the AMD Radeon R7 M460 (2.3% ahead), which shows it sits in a lower performance tier. The GTX 950M, by contrast, trades blows with the AMD Radeon R9 M360 (0.1% behind) and the NVIDIA GeForce GTX 980 (0.4% behind), meaning its average score is nearly identical to much larger desktop-class parts. For a laptop GPU, that is a strong position.

Pick the GTX 950M if you want the higher benchmark ceiling, the larger 4 GB memory buffer, and the better OpenCL showing. Pick the R7 M370 only if you specifically need its GCN feature set or its 2 GB GDDR5 memory type, but be aware that its Vulkan score is nearly tied while its OpenCL score trails by a wide margin.

Architecture Differences

The two GPUs come from different architectural generations. The NVIDIA GeForce GTX 950M uses the GM107 chip built on the Maxwell architecture, part of the GeForce 900M generation. It is manufactured by TSMC on a 28 nm process, with 1,870 million transistors on a 148 mm² die, giving a transistor density of 12.6M per mm². The AMD Radeon R7 M370 uses the Litho chip based on GCN 1.0, belonging to the Gem System (R7 M300) generation. It is also built by TSMC on 28 nm, but with just 950 million transistors on a 77 mm² die, for a density of 12.3M per mm². The NVIDIA chip is physically larger and packs roughly twice the transistor count, which explains its higher shading unit count and texture throughput.

Core configuration differs significantly. The GTX 950M has 640 shading units, 40 texture mapping units, and 16 raster operation units. The R7 M370 has 384 shading units, 24 TMUs, and only 8 ROPs. That difference in ROPs is notable: the NVIDIA part can output 17.98 GPixel/s, while the AMD part manages 7.68 GPixel/s. Similarly, texture rate is 44.96 GTexel/s versus 23.04 GTexel/s. Floating point performance (FP32) is 1,438.7 GFLOPS for the GTX 950M and 737.3 GFLOPS for the R7 M370. The NVIDIA part is roughly twice as fast in every raw throughput metric.

Memory architecture also diverges. The GTX 950M comes with 4 GB of DDR3 on a 128 bit bus, yielding 28.80 GB/s of bandwidth. The R7 M370 has 2 GB of GDDR5 on the same 128 bit bus, but bandwidth jumps to 57.60 GB/s. That is a 2x bandwidth advantage for AMD, which could help in certain memory-heavy workloads despite its lower compute throughput. Clock speeds differ as well: the GTX 950M runs at 993 MHz base and 1124 MHz boost, while the R7 M370 runs at 875 MHz base and 960 MHz boost. The NVIDIA chip also lists a memory clock of 900 MHz (1800 Mbps effective), while AMD lists 900 MHz (3.6 Gbps effective), reflecting the different memory types.

API support shows a slight edge for AMD on DirectX: the R7 M370 supports DirectX 12 (11_1), while the GTX 950M supports DirectX 12 (11_0). Both support OpenGL 4.6, but the Vulkan versions differ: NVIDIA lists Vulkan 1.4, AMD lists Vulkan 1.2.170. For modern API compatibility, NVIDIA has the newer Vulkan revision. Power draw is only listed for NVIDIA (75 W TDP), while AMD does not provide a TDP figure in the database.

Head-to-Head Benchmarks

The Geekbench OpenCL test shows the largest gap. The GTX 950M scores 9745, while the R7 M370 scores 7063, a 38% difference in favor of NVIDIA. This is a substantial lead and reflects the much higher shading unit count and FP32 throughput. In practical terms, any OpenCL compute task, such as video encoding, image processing, or physics simulation, will run noticeably faster on the GTX 950M. The R7 M370’s higher memory bandwidth cannot compensate for its lower compute core count in this workload.

The Geekbench Vulkan test is much closer. The GTX 950M scores 6525, and the R7 M370 scores 6465, a difference of only 0.9%. This near tie suggests that Vulkan workloads are less sensitive to raw shading unit counts, or that the R7 M370’s GCN architecture handles certain Vulkan operations efficiently. Still, NVIDIA wins this one too, so the overall record stands at 2 wins for the GTX 950M and 0 for the R7 M370.

Looking at the average benchmark scores, the GTX 950M’s 8135 places it just 0.1% behind the AMD Radeon R9 M360 (8129) and 0.4% behind the NVIDIA GeForce GTX 980 (8167). It is also 0.4% ahead of the NVIDIA GeForce 945M (8099) and 0.7% ahead of the NVIDIA GRID K2 (8080). The R7 M370’s 6764 average sits 1% behind the AMD FirePro M5100 (6830) and 2.6% behind the NVIDIA GeForce GTX 675M (6946), while being 1% ahead of the NVIDIA GeForce GT 1010 (6698) and 2.3% ahead of the AMD Radeon R7 M460 (6612). The GTX 950M competes in a higher performance bracket, while the R7 M370 sits among lower-tier mobile parts.

FAQ

Q: Which GPU has the higher average benchmark score?

A: The NVIDIA GeForce GTX 950M, with an average score of 8135, compared to the AMD Radeon R7 M370’s 6764. That is roughly a 20% gap in average performance.

Q: Is the AMD Radeon R7 M370 competitive in Vulkan?

A: Yes, in the Geekbench Vulkan test the R7 M370 scores 6465 versus the GTX 950M’s 6525, a difference of only 0.9%. It is nearly tied, but still loses.

Q: Does the AMD Radeon R7 M370 have any advantage in memory bandwidth?

A: Yes, its 2 GB GDDR5 memory on a 128 bit bus delivers 57.60 GB/s, while the GTX 950M’s 4 GB DDR3 on the same bus width provides 28.80 GB/s. The AMD part has double the bandwidth.

Q: Which GPU has more shading units?

A: The NVIDIA GeForce GTX 950M has 640 shading units, while the AMD Radeon R7 M370 has 384. That is a 67% higher count for NVIDIA.

Q: What is the transistor count difference?

A: The GTX 950M has 1,870 million transistors on a 148 mm² die, while the R7 M370 has 950 million transistors on a 77 mm² die. The NVIDIA chip uses roughly twice the transistors.

Q: Which GPU supports a newer Vulkan version?

A: The NVIDIA GeForce GTX 950M lists Vulkan 1.4, while the AMD Radeon R7 M370 lists Vulkan 1.2.170. NVIDIA’s implementation is newer in the database.

Where Each One Wins

The NVIDIA GeForce GTX 950M wins in all recorded benchmarks, but its strongest showing is in OpenCL, where it leads by 38%. This makes it the better choice for compute-heavy applications that rely on OpenCL, such as GPU-accelerated rendering, scientific simulation, or any workload that uses the GPU as a general-purpose processor. Its higher shading unit count (640 vs 384), higher pixel rate (17.98 GPixel/s vs 7.68 GPixel/s), and higher texture rate (44.96 GTexel/s vs 23.04 GTexel/s) all point to superior raw throughput. The 4 GB memory capacity also gives it more headroom for large datasets, even though the DDR3 memory type limits bandwidth to 28.80 GB/s.

The AMD Radeon R7 M370 does not win any benchmark in this comparison, but it does have one clear technical advantage: memory bandwidth. Its 57.60 GB/s GDDR5 setup is exactly double the GTX 950M’s 28.80 GB/s. In scenarios where bandwidth matters more than compute throughput, such as certain image processing filters, some compression algorithms, or specific Vulkan workloads, the R7 M370 could be less of a bottleneck. The Vulkan test shows this: only 0.9% behind, despite having far fewer shading units. So if you are locked into an AMD platform and your workload is Vulkan-based, the R7 M370 is not a poor choice, but it is still behind.

For rasterization and pixel filling, the GTX 950M dominates due to its 16 ROPs versus 8 on the R7 M370. That directly affects fill-rate-bound scenarios like high resolution gaming with heavy post-processing. The GTX 950M also has a higher boost clock (1124 MHz vs 960 MHz), which helps in burst workloads. The R7 M370’s lower TDP is not listed, so no power comparison is possible, but the GTX 950M is rated at 75 W, which is typical for a mobile part.

Specification Differences

| Specification | NVIDIA GeForce GTX 950M | AMD Radeon R7 M370 |

| --- | --- | --- |

| Chip | GM107 | Litho |

| Architecture | Maxwell | GCN 1.0 |

| Generation | GeForce 900M | Gem System (R7 M300) |

| Process Node | 28 nm | 28 nm |

| Transistors | 1,870 million | 950 million |

| Die Size | 148 mm² | 77 mm² |

| Transistor Density | 12.6M / mm² | 12.3M / mm² |

| Base Clock | 993 MHz | 875 MHz |

| Boost Clock | 1124 MHz | 960 MHz |

| Memory Clock | 900 MHz (1800 Mbps effective) | 900 MHz (3.6 Gbps effective) |

| Memory Size | 4 GB | 2 GB |

| Memory Type | DDR3 | GDDR5 |

| Memory Bus Width | 128 bit | 128 bit |

| Memory Bandwidth | 28.80 GB/s | 57.60 GB/s |

| Shading Units | 640 | 384 |

| TMUs | 40 | 24 |

| ROPs | 16 | 8 |

| Pixel Rate | 17.98 GPixel/s | 7.680 GPixel/s |

| Texture Rate | 44.96 GTexel/s | 23.04 GTexel/s |

| FP32 Performance | 1,438.7 GFLOPS | 737.3 GFLOPS |

| TDP | 75 W | Not listed |

| Bus Interface | PCIe 3.0 x8 | PCIe 3.0 x8 |

| DirectX | 12 (11_0) | 12 (11_1) |

| OpenGL | 4.6 | 4.6 |

| Vulkan | 1.4 | 1.2.170 |

| Production Status | End-of-life | End-of-life |

| Release Date | 2015-03-12 | 2015-05-04 |

| Predecessor | GeForce 800M | Solar System |

| Successor | GeForce 10 Mobile | Polaris Mobile |

The key differences are clear: NVIDIA has twice the transistors, twice the shading units, twice the TMUs, twice the ROPs, and roughly double the compute throughput. AMD responds with double the memory bandwidth and a slightly newer DirectX feature level. The GTX 950M also has double the memory capacity (4 GB vs 2 GB), which matters for modern workloads. The R7 M370 is a smaller, less powerful chip that relies on GDDR5 to stay relevant in bandwidth-sensitive tasks. For most users, the GTX 950M is the superior part based on the recorded data.