NVIDIA GeForce 945M vs NVIDIA GeForce GTX 1070 Comparison

# NVIDIA GeForce GTX 1070 vs NVIDIA GeForce 945M

The database places the NVIDIA GeForce GTX 1070 and the NVIDIA GeForce 945M in different tiers, and the benchmark data reflects that separation clearly. The GTX 1070, built on the GP104 chip with a Pascal architecture, records an average benchmark score of 9780 across all tested workloads. The 945M, a Maxwell-generation part based on the GM107 chip, scores an average of 8099. The delta between the two is a substantial 20.8% in favor of the desktop part. Across the full suite of recorded tests, the GTX 1070 wins in 1 out of 1 head-to-head comparisons. The percentile data reinforces the gap: the GTX 1070 sits at the 47th percentile of all GPUs in the database, while the 945M rests at the 42nd percentile. In short, the data indicates a clear performance hierarchy, but the 945M remains a capable mobile part in its own right.

Where Each One Wins

The benchmark results break down into distinct use cases where one card or the other takes the lead.

For raw compute throughput, the GTX 1070 is the definitive winner. In the Geekbench OpenCL test, the 1070 scores 44,700, which is 451.9% higher than the 945M's 8,099 in the same test. This is a massive, compute-heavy workload, and the desktop part simply has far more execution resources to throw at it. The 945M, however, has its own strengths in less demanding scenarios. The database shows the 945M scoring 1,305.6 GFLOPS in FP32 compute, a figure that is not matched by the GTX 1070 in the recorded data. For legacy API overhead and lightweight tasks, the 945M's Maxwell architecture provides a different balance of work.

Architecture Differences

The architectural split between these two NVIDIA parts is significant.

The GTX 1070 uses the GP104 chip, built on a 16 nm process at TSMC. This part integrates 7,200 million transistors on a 314 mm² die, for a transistor density of 22.9 million per square millimeter. The 945M uses the GM107 chip, a 28nm TSMC part with 1,870 million transistors on a 148 mm² die, yielding a density of 12.6 million per square millimeter. The newer, denser process is a major reason the 1070 can sustain its much larger compute load.

Clock behavior differs sharply as well. The GTX 1070 runs a base clock of 1506 MHz and a boost of 1683 MHz. Its memory operates at 2002 MHz effective, delivering 256 GB/s of bandwidth over a 256 bit bus with 8 GB of GDDR5. The 945M, by contrast, is an MXM module with a 928 MHz base and 1020 MHz boost. Its DDR3 memory runs at 900 MHz effective, with a 128 bit bus, 2 GB capacity, and 28.80 GB/s of bandwidth.

The shading and geometry processing rates tell the same story. The 1070 has 1920 shading units, 120 texture mapping units, and 64 ROPs. Its pixel rate is 107.7 GPixel/s and its texture rate is 202.0 GTexel/s. The 945M has 640 shading units, 40 TMUs, and 16 ROPs, with a pixel rate of 16.32 GPixel/s and a texture rate of 40.80 GTexel/s. The 1070 outperforms the 945M in every geometry throughput category.

The feature set also diverges. The GTX 1070 supports DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.4. It has 1x DVI, 1x HDMI 2.0, and 3x DisplayPort 1.4a outputs. The 945M supports DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.4, with its display outputs being portable device dependent. The 1070 is a dual-slot card with a 450 W suggested PSU and a single 8-pin connector. The 945M is an MXM module with no additional power connectors.

The Verdict

The data is unambiguous about which product wins in raw performance: the NVIDIA GeForce GTX 1070 is the choice for anyone who needs maximum compute throughput. Its 451.9% lead over the 945M in the OpenCL test is not a marginal edge, it is a generation-class gap. Anyone whose workloads are dominated by compute, large framebuffers, or high resolution should pick the 1070.

The NVIDIA GeForce 945M still has a role for constrained form factors or low-power tasks. Its FP32 throughput of 1,305.6 GFLOPS exceeds the recorded FP32 of the 1070, and its MXM form factor and 75W TDP are useful where the GTX 1070's dual-slot, 8-pin desktop requirements are a poor fit. It also holds its own in the 42nd percentile of all GPUs, which is respectable for a mobile part.

The data, however, is unambiguous: for overall average benchmark score, the 1070 is 19.5% ahead of the 945M, and it wins the sole recorded head-to-head test.

FAQ

Q: How much faster is the GTX 1070 in the OpenCL test?

A: The GTX 1070 scores 44,700 in geekbench_opencl, while the 945M scores 8,099. That is a 451.9% difference in favor of the GTX 1070.

Q: What is the memory bandwidth difference?

A: The GTX 1070 has 256 GB/s of bandwidth with a 256 bit GDDR5 memory bus. The 945M has 28.80 GB/s over a 128 bit DDR3 bus. The 1070's bandwidth is roughly 8.9 times higher.

Q: Does the 945M ever win in compute?

A: Yes, in FP32 (FP32) compute the 945M records 1,305.6 GFLOPS, a figure the GTX 1070 does not match in the recorded data.

Q: Which card has a better percentile ranking?

A: The 945M is at the 50th percentile of all GPUs, while the GTX 1070 is at the 47th percentile. Lower percentile is better in this database. The 1070 is the stronger part despite the lower percentile ranking.

Q: What process node do these use?

A: The GTX 1070 is on a 16nm TSMC node, and the 945M is on a 28nm TSMC node. The 1070's 7.2 billion transistors on a 314 mm² die, compared to 1.87 billion on 148 mm².

Head-to-Head Benchmarks

The recorded head-to-head benchmark is a single test: geekbench_opencl. In this test the GTX 1070 scores 44,700, which is 451.9% higher than the 945M's 8,099.

The wider database shows the GTX 1070 with a 9780 average score across all tests, which is 20.1% above the 945M's 8,099 average. The highest win margin for the 1070 comes in geekbench_opencl at 451.9%. The 945M's best relative result is in fp32 compute, where it posts 1,305.6 GFLOPS.

The 945M's performance is closer to these rivals: it is 0.2% above the NVIDIA GRID K2, 0.4% below the AMD Radeon R9 M360, 0.4% above the NVIDIA GeForce GTX 650 Ti Boost, and 0.4% below the NVIDIA GeForce GTX 950M. The GTX 1070's nearest rivals are the AMD FirePro W5000 (0.2% behind), NVIDIA Quadro M2000M (0.5% above), NVIDIA Tesla M10 (0.6% behind), and NVIDIA Tesla C2070 (0.7% behind).