NVIDIA GeForce GTX 870M vs NVIDIA GeForce GTX 950A Comparison

The GeForce GTX 950A and the GeForce GTX 870M are two end-of-life mobile GPUs that, despite being close in overall market position, offer distinctly different performance profiles. The data shows that while both sit at the 48th percentile among all GPUs, the GTX 870M holds a clear performance advantage in the one shared benchmark, though its average score tells a more nuanced story. This analysis breaks down the benchmark results, architectural differences, and specification gaps to help you decide which mobile graphics solution fits your needs.

Head-to-Head Benchmarks

The only direct comparison available is the Geekbench OpenCL test, and the results are decisive. The GTX 870M scores 12,630 points, while the GTX 950A manages 10,273 points. That is an 18.7% advantage for the 870M, a substantial margin in real-world terms. This is not a close race; the older Kepler-based chip simply outruns the newer Maxwell part in this compute workload. For perspective, the 870M’s score puts it 0.5% behind the AMD Radeon Pro 5300M (10,013) and 1.1% ahead of the NVIDIA Quadro 6000 (9,846). The 950A, by contrast, sits just 0.9% behind the AMD Radeon RX 6500M (10,362) and 2% ahead of the AMD Radeon R9 M375 (10,070). So while the 870M is the clear winner here, both cards are competing in the same general performance tier.

However, the average benchmark scores complicate the picture. The 950A’s average score is 10,273, identical to its OpenCL result, since that is the only test recorded. The 870M’s average score is 9,959, which is significantly lower than its OpenCL score of 12,630. This discrepancy is due to an additional Geekbench Metal test where the 870M scores 7,288 points. That Metal result drags down the 870M’s average, making its overall average score 3.1% lower than the 950A’s. In practical terms, if you are using OpenCL-heavy applications, the 870M is the superior choice. If you are looking at average performance across all workloads, the 950A appears more consistent, though the 870M’s compute potential is higher.

The win count in the head-to-head data is 1-0 in favor of the 870M, reflecting that single OpenCL result. There is no scenario where the 950A wins a directly compared benchmark. This is a straightforward case where the 870M’s higher shading unit count and memory bandwidth translate into a measurable lead in compute performance. The deltaPct of -18.7% for the 950A is the kind of gap that will be noticeable in any GPU-accelerated task.

FAQ

Q: Which GPU is faster in OpenCL compute workloads?

A: The GeForce GTX 870M is significantly faster, scoring 12,630 versus the GTX 950A’s 10,273 in the Geekbench OpenCL test. This represents an 18.7% lead for the 870M.

Q: Why does the GTX 870M have a lower average benchmark score than the GTX 950A?

A: The 870M’s average score of 9,959 is pulled down by its Geekbench Metal score of 7,288, which is much lower than its OpenCL score. The 950A’s average is 10,273, based solely on its OpenCL result. The 870M’s average is 3.1% lower than the 950A’s, despite winning the direct comparison.

Q: How close are these GPUs to their nearest rivals?

A: The 950A is within 2% of several competitors: it is 0.9% behind the AMD Radeon RX 6500M (10,362), 1.2% behind the NVIDIA Tesla C2075 (10,400), and 2% ahead of the AMD Radeon R9 M375 (10,070). The 870M is similarly close, sitting 0.5% behind the AMD Radeon Pro 5300M (10,013) and 1.1% ahead of the NVIDIA Quadro 6000 (9,846).

Q: Do both cards support the same DirectX version?

A: Yes, both the GTX 950A and the GTX 870M support DirectX 12 (11_0) and OpenGL 4.6. The key API difference is in Vulkan support, where the 950A supports version 1.4, while the 870M supports version 1.2.175.

Q: What is the memory configuration difference?

A: The GTX 950A uses 2 GB of DDR3 memory on a 128-bit bus with 32.03 GB/s bandwidth. The GTX 870M uses 3 GB of GDDR5 memory on a 192-bit bus with 120.0 GB/s bandwidth. The 870M has both more memory and nearly four times the bandwidth.

Q: Which GPU has more shading units?

A: The GTX 870M has 1,344 shading units, more than double the GTX 950A’s 640. This is a major factor in the 870M’s higher compute performance.

Architecture Differences

The two GPUs come from different NVIDIA architectures, and that distinction drives most of their performance characteristics. The GTX 950A is built on the Maxwell architecture using the GM107 chip, while the GTX 870M is based on the older Kepler architecture with the GK104 chip. Both are manufactured on a 28 nm process at TSMC, but the similarities end there.

The GM107 chip in the 950A is a compact die, measuring 148 mm² and containing 1,870 million transistors. This gives it a transistor density of 12.6M per mm². The GK104 chip in the 870M is much larger, at 294 mm², and packs 3,540 million transistors, though its density is slightly lower at 12.0M per mm². The 870M’s chip is nearly double the size, which explains its higher power draw and greater compute capability.

The core configurations are starkly different. The 950A has 640 shading units, 40 texture mapping units (TMUs), and 16 raster operation units (ROPs). The 870M has 1,344 shading units, 112 TMUs, and 24 ROPs. The 870M has more than double the shading units and TMUs, and 50% more ROPs. This translates directly into higher pixel and texture rates: the 870M achieves 27.08 GPixel/s and 108.3 GTexel/s, compared to the 950A’s 17.98 GPixel/s and 44.96 GTexel/s.

The memory subsystems are also architecturally different. The 950A uses DDR3 memory, while the 870M uses GDDR5. This is not just a spec sheet difference; it affects effective bandwidth dramatically. The 870M’s 120.0 GB/s bandwidth is 3.75 times higher than the 950A’s 32.03 GB/s. The 870M also runs its memory at 1250 MHz (5 Gbps effective) versus the 950A’s 1001 MHz (2 Gbps effective). For memory-intensive workloads, this is a decisive advantage for the 870M.

Both GPUs lack dedicated ray tracing and tensor cores, which is expected for their generation. They also share the same MXM-B (3.0) bus interface and are MXM modules with no power connectors. The display outputs are listed as "Portable Device Dependent" for both, meaning they rely on the laptop manufacturer’s design.

Specification Differences

The specification table reveals clear distinctions between the two cards. The GTX 870M has a higher base clock of 941 MHz and a boost clock of 967 MHz, while the GTX 950A runs at 993 MHz base and 1124 MHz boost. The 950A actually has higher clock speeds, but this does not compensate for the 870M’s massive core count advantage.

Memory is a major differentiator. The 870M offers 3 GB of GDDR5 on a 192-bit bus with 120.0 GB/s bandwidth, while the 950A offers 2 GB of DDR3 on a 128-bit bus with 32.03 GB/s bandwidth. The 870M has 50% more memory and 3.75 times the bandwidth. The 870M also has a higher transistor count (3,540 million vs 1,870 million) and a larger die size (294 mm² vs 148 mm²).

Compute performance is where the 870M pulls ahead decisively. Its FP32 performance is 2.599 TFLOPS, compared to the 950A’s 1,438.7 GFLOPS (1.44 TFLOPS). The 870M’s pixel rate is 27.08 GPixel/s versus 17.98 GPixel/s, and its texture rate is 108.3 GTexel/s versus 44.96 GTexel/s. The 870M has a higher TDP of 100 W, while the 950A draws 75 W.

The generation and release timing differ as well. The 950A is from the GeForce 900A generation and was released on March 12, 2015, with its predecessor being the GeForce 800A. The 870M is from the GeForce 800M generation, released on March 11, 2014, with its predecessor being the GeForce 700M and its successor being the GeForce 900M. The 870M is essentially a year older but still outperforms the newer part. The 950A has no listed successor, while the 870M’s successor is the GeForce 900M. Both are end-of-life products.

The Verdict

The data is unambiguous: the GeForce GTX 870M is the faster GPU. In the only direct benchmark comparison, it beats the GTX 950A by 18.7% in OpenCL performance. Its average benchmark score is slightly lower, but that is due to a single weak Metal result, not a reflection of its compute capability. The 870M offers more than double the shading units, 3.75 times the memory bandwidth, and 80.7% higher FP32 throughput. For any compute-heavy task, the 870M is the clear choice.

The GTX 950A does have some advantages, but they are not performance-related. It has a lower TDP of 75 W versus 100 W, which could be beneficial in a thin-and-light laptop. It also supports a newer Vulkan version (1.4 vs 1.2.175), which may matter for specific applications. Its higher clock speeds (993 MHz base, 1124 MHz boost) are notable, but they do not overcome the core count deficit. The 950A’s average benchmark score of 10,273 is higher than the 870M’s 9,959, but this is a statistical artifact of the 870M’s Metal score, not a sign of real-world superiority.

For the buyer choosing between these two, the GTX 870M is the safer bet for performance. The GTX 950A is only preferable if power consumption is a critical constraint or if you specifically need Vulkan 1.4 support. Otherwise, the 870M’s superior memory subsystem and compute throughput make it the stronger GPU. The 870M’s 3 GB of GDDR5 memory also provides more headroom for modern workloads than the 950A’s 2 GB of DDR3.

Where Each One Wins

The GeForce GTX 870M wins in every performance-critical category. In OpenCL compute, it leads by 18.7%. In raw specifications, it dominates with 1,344 shading units versus 640, 112 TMUs versus 40, and 24 ROPs versus 16. Its memory bandwidth of 120.0 GB/s is in a completely different class than the 950A’s 32.03 GB/s. The 870M’s FP32 performance of 2.599 TFLOPS is 80.7% higher than the 950A’s 1,438.7 GFLOPS. If you need pixel throughput, texture throughput, or memory bandwidth, the 870M is the only choice.

The GeForce GTX 950A wins in efficiency and API support. Its 75 W TDP is 25% lower than the 870M’s 100 W, which could translate to better battery life or less heat in a laptop chassis. It supports Vulkan 1.4, a newer version than the 870M’s 1.2.175, which could be relevant for Vulkan-based applications. Its higher clock speeds (993 MHz base, 1124 MHz boost) suggest it may scale efficiently at lower power levels. The 950A also has a higher average benchmark score (10,273 vs 9,959), though this is not a reliable indicator given the 870M’s superior OpenCL result.

In practical terms, the 870M is for users who prioritize performance in GPU-accelerated tasks like rendering, compute, or high-end gaming. The 950A is for users who need a more power-efficient mobile GPU and are willing to sacrifice significant compute performance. The data does not support choosing the 950A for speed, but it does support choosing it for efficiency. If you have the choice, take the 870M for performance and the 950A only if power constraints are paramount.