AMD Radeon HD 8870M vs NVIDIA GeForce GTX 960M Comparison

The GeForce GTX 960M and the Radeon HD 8870M are two mobile graphics solutions from different eras, one built on NVIDIA's Maxwell architecture and the other on AMD's GCN 1.0. The recorded data shows a single head-to-head benchmark result, with the GTX 960M securing the only win, but the underlying specifications reveal a more nuanced story about how each GPU approaches its workload.

Where Each One Wins

The database contains only one direct comparison test, Geekbench OpenCL, and the GTX 960M takes that win decisively. The NVIDIA part scores 11045 against the AMD part's 8462, a 30.5% advantage in this compute-oriented workload. This is not a marginal lead; it is a substantial gap that suggests the GTX 960M has significantly more raw computational throughput available for general-purpose GPU tasks.

The Radeon HD 8870M does not win any recorded benchmark, but that does not mean it lacks a purpose. Looking at its specifications, the AMD part operates at a much lower clock speed, 725 MHz base and 775 MHz boost, compared to the GTX 960M's 1097 MHz base and 1176 MHz boost. The AMD GPU compensates with an identical count of shading units, 640, and texture mapping units, 40, matching the GTX 960M in parallel resource allocation. In workloads that are latency-bound rather than throughput-bound, or those that scale better with memory bus width rather than sheer clock speed, the AMD part could theoretically hold its own, but the recorded data does not show a single test where it does.

The GTX 960M also has a clear advantage in memory capacity, offering 4 GB of GDDR5 versus the HD 8870M's 2 GB. For modern games or compute tasks that exceed 2 GB of video memory, the NVIDIA part simply has more room to work. The bandwidth figures tell a similar story: the GTX 960M delivers 80.19 GB/s while the HD 8870M reaches 72.00 GB/s, a difference of roughly 11% in favor of the NVIDIA chip.

Architecture Differences

Both GPUs are fabricated on the same 28 nm process at TSMC, but their internal designs diverge sharply. The GTX 960M uses the GM107 chip built on NVIDIA's Maxwell architecture, which is known for its high clock speeds and impressive performance per watt. The HD 8870M uses the Venus chip with AMD's GCN 1.0 architecture, the first iteration of a design that would go on to power many subsequent AMD products.

The transistor counts reflect this generational gap. The GTX 960M packs 1,870 million transistors into a 148 mm² die, giving it a transistor density of 12.6M per mm². The HD 8870M contains 1,500 million transistors on a smaller 123 mm² die, with a density of 12.2M per mm². The Maxwell chip has roughly 25% more transistors and a slightly higher density, which helps explain its superior compute performance.

The memory subsystems differ in capacity but share the same bus width. Both use a 128-bit bus with GDDR5 memory, but the GTX 960M runs its memory at 1253 MHz (5 Gbps effective) while the HD 8870M runs at 1125 MHz (4.5 Gbps effective). This clock difference accounts for the bandwidth gap, 80.19 GB/s versus 72.00 GB/s, and the GTX 960M doubles the capacity to 4 GB.

The API support also shows the age difference between the two parts. The GTX 960M supports DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.4, and the HD 8870M supports DirectX 12 (11_1), OpenGL 4.6, and Vulkan 1.2.170. Both GPUs are marked as end-of-life in the database, with release dates of 2015-03-03 for the NVIDIA part and 2013-03-31 for the AMD part, roughly two years earlier.

The bus interface differs as well, with the GTX 960M using an MXM-B (3.0) module form factor and the HD 8870M using PCIe 3.0 x16. The GTX 960M is a 75 W TDP part with no power connectors required, while the HD 8870M has no recorded TDP value.

Head-to-Head Benchmarks

The single recorded head-to-head benchmark is Geekbench OpenCL, and it is the only test where both GPUs have comparable data. The GTX 960M scores 11045, the HD 8870M scores 8462, and the result is a 30.5% win for NVIDIA. This test exercises the GPU's compute shaders, and the Maxwell architecture clearly benefits from its higher clocks and larger transistor budget.

The Geekbench Vulkan test is only available for the GTX 960M, where it scores 8245. The HD 8870M has no Vulkan result recorded in the database. This absence is notable because it means the NVIDIA part's Vulkan performance cannot be directly compared, but its 8245 Vulkan score is close to the AMD part's 8462 OpenCL score, suggesting the GTX 960M might be competitive even in API-specific workloads.

The average benchmark scores in the database place these GPUs in similar percentile territory. The GTX 960M sits at the 46th percentile against all GPUs with an average score of 9645, while the HD 8870M is at the 43rd percentile with an average score of 8462. The nearest rivals for the GTX 960M include the NVIDIA Quadro K5000 at 9637 (a 0.1% delta), the AMD Radeon Pro WX 2100 at 9653 (-0.1%), the NVIDIA Quadro P4000 at 9665 (-0.2%), and the NVIDIA Tesla C2070 at 9716 (-0.7%). The HD 8870M's nearest rivals are the NVIDIA GeForce MX330 at 8458 (0.1%), the AMD Radeon 880M at 8436 (0.3%), the NVIDIA GeForce GTX 675MX at 8427 (0.4%), and the Intel Arc A380 at 8558 (-1.1%).

The delta percentages show that both GPUs sit in a dense performance cluster. The GTX 960M is within 0.7% of its closest rivals, while the HD 8870M is within 1.1% of its own. The 30.5% gap between the two parts is therefore not an artifact of benchmark noise but a genuine performance tier difference.

FAQ

Q: Which GPU has a higher average benchmark score?

A: The NVIDIA GeForce GTX 960M has an average benchmark score of 9645, while the AMD Radeon HD 8870M scores 8462. The GTX 960M sits at the 46th percentile of all GPUs, the HD 8870M at the 43rd.

Q: How much faster is the GTX 960M in the only direct comparison?

A: In the Geekbench OpenCL test, the GTX 960M scores 11045 against the HD 8870M's 8462, a 30.5% improvement for the NVIDIA part.

Q: Do both GPUs have the same number of shading units and texture units?

A: Yes, both GPUs have 640 shading units and 40 texture mapping units. They also share a 128-bit memory bus and use GDDR5 memory.

Q: What is the difference in memory capacity?

A: The GTX 960M has 4 GB of GDDR5 memory, while the HD 8870M has 2 GB. The GTX 8870M also has a higher memory bandwidth at 80.19 GB/s versus 72.00 GB/s.

Q: Which GPU supports a newer version of Vulkan?

A: The GTX 960M supports Vulkan 1.4, while the HD 8870M supports Vulkan 1.2.170. The HD 8870M has no recorded Vulkan benchmark score, while the GTX 960M scores 8245 in that test.

Q: How do the two GPUs compare in transistor count?

A: The GTX 960M has 1,870 million transistors on a 148 mm² die, while the HD 8870M has 1,500 million transistors on a 123 mm² die. The GTX 960M's transistor density is 12.6M per mm² versus 12.2M per mm² for the AMD part.

Q: What are the clock speeds for each GPU?

A: The GTX 960M runs at 1097 MHz base and 1176 MHz boost, while the HD 8870M runs at 725 MHz base and 775 MHz boost. The GTX 960M also has a higher memory clock of 1253 MHz (5 Gbps effective) versus 1125 MHz (4.5 Gbps effective).

Specification Differences

The specifications where the two GPUs differ are numerous. The GTX 960M uses the GM107 chip with Maxwell architecture, while the HD 8870M uses the Venus chip with GCN 1.0 architecture. The transistor counts are 1,870 million versus 1,500 million, and die sizes are 148 mm² versus 123 mm². Transistor density is 12.6M per mm² versus 12.2M per mm².

Clock speeds differ substantially: the GTX 960M has a 1097 MHz base and 1176 MHz boost, the HD 8870M has a 725 MHz base and 775 MHz boost. Memory clocks are 1253 MHz (5 Gbps effective) versus 1125 MHz (4.5 Gbps effective). Memory capacity is 4 GB versus 2 GB, and bandwidth is 80.19 GB/s versus 72.00 GB/s.

The pixel rate is 18.82 GPixel/s for the GTX 960M versus 12.40 GPixel/s for the HD 8870M. The texture rate is 47.04 GTexel/s versus 31.00 GTexel/s. FP32 performance is 1.505 TFLOPS versus 992.0 GFLOPS. The GTX 960M has a 75 W TDP, while the HD 8870M has no recorded TDP. The bus interface is MXM-B (3.0) versus PCIe 3.0 x16. The GTX 960M has no power connectors, the HD 8870M has no recorded connector data.

DirectX support is 12 (11_0) for NVIDIA and 12 (11_1) for AMD, Vulkan support is 1.4 versus 1.2.170. Release dates are 2015-03-03 for the GTX 960M and 2013-03-31 for the HD 8870M. The GTX 960M's predecessor is GeForce 800M and its successor is GeForce 10 Mobile, while the HD 8870M's predecessor is London and successor is Gem System.

The GTX 960M has a recorded OpenCL score of 11045 and Vulkan score of 8245, while the HD 8870M has only an OpenCL score of 8462. The GTX 960M has one win in the head-to-head comparison, the HD 8870M has zero wins recorded.

The Verdict

The data points one way for almost every metric, pointing to the GTX 30.5% ahead in OpenCL and a higher percentile ranking at 46 versus 43. The GTX 960M is the faster GPU for compute workloads, with a 1.505 TFLOPS FP32 throughput versus 992.0 GFLOPS, and it offers double the memory capacity at 4 GB versus 2 GB. Its higher clock speeds, 1176 MHz boost versus 775 MHz boost, and larger transistor count of 1,870 million versus 1,500 million give it a structural advantage that the HD 8870M cannot overcome.

The HD 8870M is not without merit. It shares the same 640 shading units, 40 TMUs, 16 ROPs, and 128-bit bus as the GTX 960M, making it a capable part for its era. But its lower clocks and smaller memory pool place it a clear tier below, as confirmed by the 43rd percentile ranking and the 8462 average score. Both parts are end-of-life, so the choice is academic for new purchases, but for anyone comparing used or legacy systems, the GTX 960M is the stronger option in every recorded benchmark and most specification categories.

Users who need maximum compute performance, larger frame buffers, or newer API support should choose the GTX 960M. Users who are constrained to the HD 8870M platform should know that it will deliver roughly three-quarters of the OpenCL performance of the GTX 960M, with the recorded data showing the NVIDIA part scoring 11045 versus 8462. The verdict from the database is clear: the GTX 960M wins the only head-to-head test and offers superior specifications across the board, making it the recommended pick from the available data.