NVIDIA GeForce GTX 1070 vs NVIDIA GeForce GTX 870M Comparison

The NVIDIA GeForce GTX 870M and the NVIDIA GeForce GTX 1070 represent two distinct generations of mobile graphics, separated by over two years of architectural evolution. The GTX 870M is a Kepler-based part from the GeForce 800M generation, while the GTX 1070 is a Pascal-based part from the GeForce 10-series. The data shows a decisive performance gap, with the GTX 1070 winning both head-to-head benchmarks, yet the GTX 870M holds its own in specific legacy and compute contexts. This analysis breaks down the benchmark results, architectural shifts, and specification differences between the two GPUs.

FAQ

Q: Which GPU has the higher average benchmark score?

A: The NVIDIA GeForce GTX 1070 has an average benchmark score of 9780, while the NVIDIA GeForce GTX 870M scores 9959. Despite the GTX 1070 winning both head-to-head tests, the GTX 870M holds a higher average score by 179 points, a 1.8% difference.

Q: How do the two GPUs compare in the Geekbench Metal benchmark?

A: The GTX 1070 achieves a Geekbench Metal score of 18801, which is 61.2% higher than the GTX 870M's score of 7288. This is a substantial margin, indicating the GTX 1070 is significantly faster in this API.

Q: What is the difference in OpenCL performance between the two?

A: In the Geekbench OpenCL test, the GTX 1070 scores 44700, while the GTX 870M scores 12630. The GTX 1070 leads by 71.7%, an even larger gap than in the Metal test.

Q: Which GPU has a higher percentile ranking among all GPUs?

A: The GTX 870M has a percentile rank of 48, while the GTX 1070 has a percentile rank of 47. This places both cards in the mid-range of the database, with the GTX 870M holding a slight edge in overall standing.

Q: What are the nearest rivals for the GTX 870M?

A: The GTX 870M's nearest rivals include the AMD Radeon Pro 5300M (avg score 10013, -0.5%), the NVIDIA Quadro K5100M (10043, -0.8%), and the AMD Radeon R9 M375 (10070, -1.1%). Its closest competitor below is the NVIDIA Quadro 6000 (9846, +1.1%).

Q: What are the nearest rivals for the GTX 1070?

A: The GTX 1070's nearest rivals include the AMD FirePro W5000 (avg score 9803, -0.2%), the NVIDIA Quadro M2000M (9832, -0.5%), and the NVIDIA Tesla M10 (9724, +0.6%). The NVIDIA Tesla C2070 (9716, +0.7%) is its closest competitor below.

Where Each One Wins

The GTX 1070 is the clear winner in raw compute performance. It dominates both head-to-head benchmark tests, with a 61.2% advantage in Geekbench Metal and a 71.7% advantage in Geekbench OpenCL. This makes it the superior choice for any workload that leverages modern compute APIs, such as gaming, content creation, or GPU-accelerated applications. The GTX 1070 also offers a wider feature set with support for DirectX 12 (12_1) and Vulkan 1.4, which are essential for contemporary titles and applications.

The GTX 870M, despite losing both head-to-head tests, has a higher average benchmark score (9959 vs 9780) and a higher percentile rank (48 vs 47). This suggests that in a broader set of benchmarks, the GTX 870M can outperform the GTX 1070 in certain legacy or specific workloads. Its OpenGL 4.6 and Vulkan 1.2.175 support, while less current, still provide compatibility with older software. The GTX 870M also has a lower TDP of 100 W, making it a more power-efficient option for thin and light laptops, whereas the GTX 1070's 150 W TDP requires a more robust cooling solution.

The GTX 1070 wins on memory capacity and bandwidth. With 8 GB of GDDR5 memory on a 256-bit bus, it offers 256.3 GB/s of bandwidth. The GTX 870M has 3 GB of GDDR5 on a 192-bit bus, providing only 120.0 GB/s. This makes the GTX 1070 far better suited for high-resolution textures and memory-intensive applications. The GTX 870M, however, remains a capable option for older games and less demanding tasks, where its lower power draw and smaller footprint (MXM Module) can be an advantage.

Architecture Differences

The two GPUs are built on fundamentally different architectures. The GTX 870M uses the Kepler architecture, fabricated on a 28 nm process at TSMC. It features the GK104 chip, which contains 3,540 million transistors on a die size of 294 mm². This results in a transistor density of 12.0M per mm². The GTX 1070, in contrast, uses the Pascal architecture, fabricated on a 16 nm process at TSMC. It features the GP104 chip, containing 7,200 million transistors on a die size of 314 mm². This gives it a transistor density of 22.9M per mm², nearly double that of the GTX 870M.

The architectural shift from Kepler to Pascal brings significant changes in compute capabilities. The GTX 870M has 1344 shading units, 112 TMUs, and 24 ROPs. The GTX 1070 increases these to 1920 shading units, 120 TMUs, and 64 ROPs. This translates to a massive difference in pixel and texture throughput: the GTX 1070 has a pixel rate of 107.7 GPixel/s and a texture rate of 202.0 GTexel/s, compared to the GTX 870M's 27.08 GPixel/s and 108.3 GTexel/s.

The FP32 performance tells a similar story. The GTX 870M delivers 2.599 TFLOPS, while the GTX 1070 delivers 6.463 TFLOPS, a 2.5x increase. The GTX 1070 also has FP16 capability at 101.0 GFLOPS (1:64), whereas the GTX 870M has no listed FP16 performance. The GTX 1070 supports DirectX 12 (12_1) and Vulkan 1.4, while the GTX 870M only supports DirectX 12 (11_0) and Vulkan 1.2.175. This means the GTX 1070 is ready for modern graphics features and APIs.

Specification Differences

The most obvious specification difference is the process node: the GTX 870M uses 28 nm, while the GTX 1070 uses 16 nm. This is a major factor in the performance and efficiency gap. The GTX 870M has a base clock of 941 MHz and a boost clock of 967 MHz, while the GTX 1070 has a base clock of 1506 MHz and a boost clock of 1683 MHz. The memory clocks also differ: the GTX 870M runs at 1250 MHz (5 Gbps effective), while the GTX 1070 runs at 2002 MHz (8 Gbps effective).

Memory configuration is another key differentiator. The GTX 870M has 3 GB of GDDR5 on a 192-bit bus, yielding 120.0 GB/s of bandwidth. The GTX 1070 has 8 GB of GDDR5 on a 256-bit bus, yielding 256.3 GB/s of bandwidth. This is a 2.1x increase in bandwidth, which is critical for high-resolution gaming and data-intensive workloads.

The physical and power characteristics also differ significantly. The GTX 870M has a TDP of 100 W and uses an MXM-B (3.0) interface, with slot width listed as "MXM Module" and no power connectors. The GTX 1070 has a TDP of 150 W, uses a PCIe 3.0 x16 interface, is a dual-slot card, and requires a 1x 8-pin power connector. The GTX 1070 also has a suggested PSU of 450 W, while the GTX 870M has no suggested PSU listed. The GTX 1070 measures 267 mm in length, 112 mm in height, and 40 mm in width, while the GTX 870M has no dimensions listed.

Display outputs also differ: the GTX 870M is "Portable Device Dependent," while the GTX 1070 offers 1x DVI, 1x HDMI 2.0, and 3x DisplayPort 1.4a. The GTX 1070 has a launch MSRP of 379 USD. The GTX 870M was released on 2014-03-11, while the GTX 1070 was released on 2016-06-09.

Head-to-Head Benchmarks

The GTX 1070 wins both head-to-head benchmarks by substantial margins. In the Geekbench Metal test, the GTX 1070 scores 18801, while the GTX 870M scores 7288. This represents a delta of -61.2% for the GTX 870M, meaning the GTX 1070 is roughly 2.6x faster. This is a clear indicator of the GTX 1070's superiority in modern compute environments.

The Geekbench OpenCL test shows an even larger gap. The GTX 1070 scores 44700, while the GTX 870M scores 12630. The delta here is -71.7%, meaning the GTX 1070 is approximately 3.5x faster. This is the biggest win for the GTX 1070 in the head-to-head data.

The GTX 870M has no wins in the head-to-head benchmarks, but its average benchmark score of 9959 is higher than the GTX 1070's 9780. This is a curious anomaly, explained by the fact that the GTX 870M's average is based on only two benchmark results (Geekbench Metal and OpenCL), while the GTX 1070's average includes a wider range of tests, such as 3DMark Steel Nomad, Passmark DirectX 9/10/11/12, and Passmark G2D/G3D. The GTX 1070's Passmark G3D score of 13498 and Passmark GPU Compute score of 6102 are strong, but its lower scores in Passmark DirectX 10 (82) and DirectX 12 (48) drag down the average.

In contrast, the GTX 870M's nearest rivals are all within a narrow band of 9846 to 10070, showing it sits in a competitive cluster. The GTX 1070's nearest rivals range from 9716 to 9832, also a tight grouping. The data suggests that while the GTX 1070 is the superior performer in the two head-to-head tests, the GTX 870M's overall standing in the database is marginally higher, likely due to the different mix of benchmarks used to calculate the averages.