NVIDIA GeForce GTX 850M vs NVIDIA GeForce GTX 960 Comparison

The NVIDIA GeForce GTX 850M and the NVIDIA GeForce GTX 960 are both end-of-life Maxwell parts, but they occupy very different corners of the market. The 850M is a 45 W mobile IGP designed for laptops, while the 960 is a 120 W dual-slot desktop card with a 241 mm length. Despite this, their average benchmark scores are nearly identical: the 850M sits at 9,302, while the 960 posts 9,273, a difference of just 0.3%. The nearest-rival data underscores this parity, with the GTX 465 (9,294) and Radeon R7 M380 (9,313) all clustered within a fraction of a percent. Yet the head-to-head results tell a more nuanced story, revealing that the 960’s advantage is real but heavily dependent on the workload.

Head-to-Head Benchmarks

The most striking divergence appears in the Geekbench OpenCL test. Here, the GTX 960 scores 18,925, while the GTX 850M manages only 9,821. That is a 48.1% deficit for the mobile part, meaning the 960 delivers nearly double the raw compute throughput in this API. The gap is so large that it completely overshadows the near-tie in average scores. This result aligns with the silicon differences: the 960 features 1,024 shading units versus 640 on the 850M, and its FP32 output is 2.413 TFLOPS compared to 1,154.6 GFLOPS. The OpenCL workload appears to scale almost linearly with these resources, punishing the smaller chip.

The second head-to-head test, Geekbench Vulkan, is far closer. The 960 wins again, but by only 4.9%, scoring 9,231 against the 850M’s 8,782. This suggests that Vulkan’s lower-level API is less sensitive to raw shading-unit count and more dependent on other factors, such as memory bandwidth or driver scheduling. The 850M’s DDR3 memory provides just 32.03 GB/s of bandwidth, while the 960’s GDDR5 delivers 112.2 GB/s — a 3.5x advantage for the desktop card. Yet the Vulkan result compresses that gap to under 5%, implying that the mobile chip’s architecture is surprisingly efficient in this scenario, or that the test is not bandwidth-bound.

In terms of overall wins, the 960 takes both head-to-head comparisons. However, the average benchmark scores flip the narrative: the 850M actually holds a 0.3% edge over the 960 in aggregate (9,302 vs. 9,273). This is a statistical wash, well within the noise of the nearest rivals. The 850M also sits 0.9% above the AMD Radeon Vega 8 (9,221), while the 960 trails that same iGPU by 0.6%. The practical takeaway is that these two GPUs deliver comparable average performance, but the 960’s peak compute in OpenCL is categorically superior.

Architecture Differences

Both GPUs are built by TSMC on a 28 nm process, but they are not the same silicon. The GTX 850M uses the GM107 chip, which is the entry-level Maxwell design. It packs 1,870 million transistors into a 148 mm² die, yielding a transistor density of 12.6M per mm². The GTX 960, in contrast, uses the GM206 chip, a Maxwell 2.0 part with 2,940 million transistors on a 228 mm² die, for a density of 12.9M per mm². The 960’s die is 54% larger and carries 57% more transistors, which directly explains its higher core counts: 1,024 shading units, 64 TMUs, and 32 ROPs, versus the 850M’s 640 shading units, 40 TMUs, and 16 ROPs.

Memory is another major differentiator. The 850M ships with 2 GB of DDR3 on a 128-bit bus, running at 1001 MHz (2 Gbps effective), which results in a bandwidth of just 32.03 GB/s. The 960 also has 2 GB, but it is GDDR5 on the same 128-bit bus, clocked at 1753 MHz (7 Gbps effective), yielding 112.2 GB/s. This 3.5x bandwidth advantage is critical for texture-heavy games and compute tasks that stream large datasets. The pixel rate tells a similar story: the 960 outputs 37.70 GPixel/s versus the 850M’s 14.43 GPixel/s, and the texture rate is 75.39 GTexel/s versus 36.08 GTexel/s.

Feature support also diverges. The 850M is rated for DirectX 12 (11_0), while the 960 supports DirectX 12 (12_1). Both offer OpenGL 4.6 and Vulkan 1.4, and neither has dedicated ray tracing or tensor cores. The 960 adds a 6-pin power connector and a suggested PSU of 300 W, whereas the 850M is an IGP with no external power connectors. The 960’s display outputs include 1x DVI, 1x HDMI 2.0, and 3x DisplayPort 1.2; the 850M’s outputs are listed as portable-device dependent. Finally, the 960 has a base clock of 1127 MHz and a boost of 1178 MHz, while the 850M has no specified base or boost clocks in the data.

FAQ

Q: Which GPU has a higher average benchmark score?

A: The GTX 850M edges out the GTX 960, with an average score of 9,302 versus 9,273. This is a 0.3% difference, placing them statistically even.

Q: How much faster is the GTX 960 in OpenCL?

A: The 960 scores 18,925 in Geekbench OpenCL, which is 48.1% higher than the 850M’s 9,821. This is the largest performance gap between the two.

Q: Does the GTX 960 win in Vulkan too?

A: Yes, but narrowly. The 960 scores 9,231 versus the 850M’s 8,782, a margin of just 4.9%.

Q: What are the transistor counts for each chip?

A: The 850M’s GM107 has 1,870 million transistors, while the 960’s GM206 has 2,940 million. Both are fabricated by TSMC on a 28 nm node.

Q: Do these GPUs support the same DirectX version?

A: No. The 850M supports DirectX 12 (11_0), while the 960 supports DirectX 12 (12_1). Both support OpenGL 4.6 and Vulkan 1.4.

Q: Which card has more memory bandwidth?

A: The GTX 960 has 112.2 GB/s of bandwidth from its GDDR5 memory, versus the 850M’s 32.03 GB/s from DDR3. Both use a 128-bit bus.

The Verdict

The data presents a clear trade-off between form factor and compute ceiling. For anyone constrained to a laptop or a low-power IGP setup, the GTX 850M is the only choice here, and its average score of 9,302 shows it holds its own against desktop parts. Its 45 W TDP and lack of power connectors make it drop-in friendly for portable systems, and its Vulkan score of 8,782 is within 4.9% of the 960, proving it is not a slouch in modern APIs. The 0.1% edge over the GTX 465 and 0.9% over the Vega 8 in average score further cements its position as a competent entry-level mobile GPU.

For a desktop builder, the GTX 960 is the superior part despite the near-identical average scores. Its 48.1% lead in OpenCL is decisive for any compute-heavy application, and its 2.413 TFLOPS FP32 throughput is more than double the 850M’s 1,154.6 GFLOPS. The 960 also offers DirectX 12 (12_1) support, which the 850M lacks, and its GDDR5 memory provides 3.5x the bandwidth, making it better suited for high-resolution textures. The 960’s 120 W TDP and 6-pin connector are acceptable for a dual-slot card, and its 241 mm length fits most mid-tower cases.

The final call depends on the use case. If the requirement is a mobile GPU that can handle light gaming and general compute, the 850M is adequate, but its performance ceiling is capped by memory bandwidth and core count. If the requirement is a desktop GPU for serious gaming or OpenCL workloads, the 960 is the obvious winner, even though its average score is technically lower by 0.3%. The 960’s wins in both head-to-head tests, including the 4.9% Vulkan margin, indicate it is the more capable architecture overall. The 850M’s only advantage is efficiency and portability, which the benchmark data does not directly measure but which are implicit in its 45 W TDP versus the 960’s 120 W.

Specification Differences

| Specification | GTX 850M | GTX 960 |

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

| Chip | GM107 | GM206 |

| Architecture | Maxwell | Maxwell 2.0 |

| Transistors | 1,870 million | 2,940 million |

| Die Size | 148 mm² | 228 mm² |

| Transistor Density | 12.6M / mm² | 12.9M / mm² |

| Base Clock | Not specified | 1127 MHz |

| Boost Clock | Not specified | 1178 MHz |

| Memory Clock | 1001 MHz (2 Gbps effective) | 1753 MHz (7 Gbps effective) |

| Memory Type | DDR3 | GDDR5 |

| Memory Bus Width | 128 bit | 128 bit |

| Memory Bandwidth | 32.03 GB/s | 112.2 GB/s |

| Shading Units | 640 | 1024 |

| TMUs | 40 | 64 |

| ROPs | 16 | 32 |

| Pixel Rate | 14.43 GPixel/s | 37.70 GPixel/s |

| Texture Rate | 36.08 GTexel/s | 75.39 GTexel/s |

| FP32 Performance | 1,154.6 GFLOPS | 2.413 TFLOPS |

| TDP | 45 W | 120 W |

| Slot Width | IGP | Dual-slot |

| Power Connectors | None | 1x 6-pin |

| Suggested PSU | Not specified | 300 W |

| Display Outputs | Portable Device Dependent | 1x DVI, 1x HDMI 2.0, 3x DisplayPort 1.2 |

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

| Length | Not specified | 241 mm (9.5 inches) |

| Release Date | 2014-03-11 | 2015-01-21 |

| Predecessor | GeForce 700M | GeForce 700 |

| Successor | GeForce 900M | GeForce 10 |

| Launch MSRP | Not specified | 199 USD |

| Average Benchmark Score | 9,302 | 9,273 |

| Percentile (vs all GPUs) | 46 | 45 |