NVIDIA GeForce 840M vs NVIDIA GeForce GTX 970M Comparison

Head-to-Head Benchmarks

The benchmark data presents a stark contrast between these two mobile graphics processors. Across the two common tests recorded in the database, the NVIDIA GeForce GTX 970M dominates the NVIDIA GeForce 840M without exception. The head-to-head comparison shows the GTX 970M winning both recorded tests, with the 840M failing to secure a single victory.

In the Geekbench OpenCL test, the GTX 970M scores 18,946 points, while the 840M manages only 5,764 points. This represents a delta of -69.6%, meaning the 840M trails the GTX 970M by nearly 70% in this compute-oriented workload. The magnitude of this gap is substantial, indicating a fundamental difference in raw compute capability between the two parts.

The Vulkan test reveals an even wider chasm. The GTX 970M records 18,292 points, while the 840M scores 4,880 points, a delta of -73.3%. This suggests that the GTX 970M's advantage becomes more pronounced in modern graphics API workloads, likely due to its more robust architecture and higher resource counts.

Looking at the average benchmark score, the 840M sits at 5,322, which places it in the 31st percentile of all GPUs in the database. The GTX 970M, despite its decisive head-to-head wins, has an average score of 4,628 and sits in the 27th percentile. This apparent contradiction is explained by the different test suites available for each card: the 840M's average is drawn from only two Geekbench tests, while the GTX 970M's average includes several PassMark tests with lower scores, such as 28 in DirectX 10 and 24 in DirectX 12, which drag down its overall average.

The nearest rivals for the 840M include the NVIDIA GeForce 930A with an average score of 5,317 and a delta of 0.1%, the NVIDIA GeForce GTX 980M at 5,308 with a delta of 0.3%, and the AMD Radeon R7 M445 at 5,358 with a delta of -0.7%. These figures indicate that the 840M performs essentially on par with its closest competitors, with differences well under 1%. For the GTX 970M, the nearest rivals include the NVIDIA Quadro M3000M at 4,621 with a delta of 0.2%, and the AMD Radeon R5 M320 at 4,657 with a delta of -0.6%, again showing near-identical performance within its own peer group.

Where Each One Wins

The use-case split between these two GPUs is unambiguous according to the recorded data. The GTX 970M wins every benchmark where both cards have results, making it the clear choice for any workload that relies on compute performance, graphics rendering, or modern API support.

For the 840M, the data shows no benchmark victories against the GTX 970M. However, its performance profile aligns with lighter tasks. Its 2 GB DDR3 memory on a 64-bit bus yields 16.02 GB/s of bandwidth, which is sufficient for basic display output and modest 2D workloads. Its 33 W TDP and IGP slot width suggest it is designed for thin-and-light laptops where power efficiency takes priority over performance. The 840M's average benchmark score of 5,322, which surpasses the GTX 970M's 4,628 due to the different test distributions, indicates that in certain Geekbench workloads it produces competitive numbers, but those results do not translate to the head-to-head tests.

The GTX 970M, by contrast, is built for demanding applications. Its 6 GB GDDR5 memory on a 192-bit bus provides 120.3 GB/s of bandwidth, roughly 7.5 times that of the 840M. This bandwidth advantage directly supports higher-resolution textures and more complex scenes. In the PassMark suite, the GTX 970M records a G3D score of 5,704, a G2D score of 381, and a GPU compute score of 2,289, demonstrating strength across both traditional rasterization and compute tasks. Its DirectX 9 score of 99 and DirectX 11 score of 42 further confirm its capability in legacy and modern graphics APIs, while its DirectX 12 score of 24 and 3DMark Steel Nomad DX12 score of 472 show continued competence in current-generation workloads.

FAQ

Q: Which GPU has the higher Geekbench OpenCL score?

A: The NVIDIA GeForce GTX 970M scores 18,946 points, while the NVIDIA GeForce 840M scores 5,764 points, giving the GTX 970M a 69.6% lead.

Q: How much faster is the GTX 970M in the Vulkan benchmark?

A: The GTX 970M records 18,292 points versus 4,880 points for the 840M, a delta of 73.3% in favor of the GTX 970M.

Q: What memory configurations do these GPUs use?

A: The 840M has 2 GB of DDR3 memory on a 64-bit bus with 16.02 GB/s bandwidth. The GTX 970M has 6 GB of GDDR5 memory on a 192-bit bus with 120.3 GB/s bandwidth.

Q: Are these GPUs from the same architecture generation?

A: No. The 840M uses the GM108S chip based on Maxwell architecture, while the GTX 970M uses the GM204 chip based on Maxwell 2.0 architecture. The 840M belongs to the GeForce 800M generation, while the GTX 970M belongs to the GeForce 900M generation.

Q: What is the average benchmark score for each GPU?

A: The 840M has an average benchmark score of 5,322, placing it in the 31st percentile. The GTX 970M has an average benchmark score of 4,628, placing it in the 27th percentile.

Q: Which GPU has more shading units?

A: The GTX 970M has 1,280 shading units, compared to 384 for the 840M. The GTX 970M also has 80 texture mapping units and 48 raster output pipelines, versus 16 TMUs and 8 ROPs for the 840M.

Specification Differences

The two GPUs differ substantially across nearly every specification category. The 840M uses the GM108S chip fabricated on a 28 nm process at TSMC, containing 1,020 million transistors on a 77 mm² die. The GTX 970M uses the GM204 chip, also on 28 nm at TSMC, but packs 5,200 million transistors on a 398 mm² die. Transistor density is nearly identical at 13.2 million per mm² for the 840M and 13.1 million per mm² for the GTX 970M.

Clock speeds show an interesting inversion: the 840M has a higher base clock of 1,029 MHz and boost clock of 1,124 MHz, while the GTX 970M operates at 924 MHz base and 1,038 MHz boost. The GTX 970M compensates with far more execution resources. Memory clocks also differ, with the 840M running at 1,001 MHz (2 Gbps effective) and the GTX 970M at 1,253 MHz (5 Gbps effective).

Memory capacity and type represent major differences. The 840M offers 2 GB of DDR3 on a 64-bit bus, yielding 16.02 GB/s bandwidth. The GTX 970M offers 6 GB of GDDR5 on a 192-bit bus, yielding 120.3 GB/s bandwidth. The 840M has 384 shading units, 16 TMUs, and 8 ROPs, while the GTX 970M has 1,280 shading units, 80 TMUs, and 48 ROPs.

Pixel and texture fill rates scale accordingly: the 840M achieves 8.992 GPixel/s and 17.98 GTexel/s, while the GTX 970M achieves 49.82 GPixel/s and 83.04 GTexel/s. FP32 compute is 863.2 GFLOPS for the 840M versus 2.657 TFLOPS for the GTX 970M. The 840M has a 33 W TDP and uses an IGP slot form factor with no power connectors, while the GTX 970M uses an MXM Module form factor with no power connectors and no recorded TDP. Bus interfaces differ as well: the 840M uses PCIe 3.0 x8, while the GTX 970M uses MXM-B (3.0).

Architecture Differences

Architecturally, these GPUs represent two distinct steps in NVIDIA's Maxwell evolution. The 840M is built on the original Maxwell architecture, while the GTX 970M is built on Maxwell 2.0. Both use a 28 nm process at TSMC, but the chip designs diverge significantly. The GM108S chip in the 840M is a small, power-efficient design with 1,020 million transistors, while the GM204 chip in the GTX 970M is a large, high-performance design with 5,200 million transistors, more than five times the transistor count.

The Maxwell 2.0 architecture in the GTX 970M introduces support for DirectX 12 (12_1), whereas the 840M's Maxwell architecture supports DirectX 12 (11_0) only. Both GPUs support OpenGL 4.6 and Vulkan 1.4, so API compatibility is identical for those standards. The higher DirectX feature level on the GTX 970M enables more advanced rendering features in modern games and applications.

The shading unit count difference (384 versus 1,280) is the most direct measure of compute capability, and it explains the benchmark results. The GTX 970M has 3.3 times as many shading units, 5 times as many TMUs, and 6 times as many ROPs as the 840M. These resource ratios align with the observed performance deltas in the Geekbench tests, where the GTX 970M leads by roughly 3 to 4 times.

Memory architecture also differs fundamentally. The 840M uses DDR3 memory, which is lower bandwidth and higher latency than the GDDR5 used in the GTX 970M. The memory bus width difference (64-bit versus 192-bit) compounds this, resulting in a 7.5-fold bandwidth advantage for the GTX 970M. This memory bandwidth is critical for texture-heavy workloads and high-resolution rendering.

The 840M's release date of March 2014 places it in the GeForce 800M generation, with the GeForce 700M as its predecessor and the GeForce 900M as its successor. The GTX 970M was released in October 2014 as part of the GeForce 900M generation, with the GeForce 800M as its predecessor and GeForce 10 Mobile as its successor. Both are now marked as end-of-life products.

The Verdict

The data directs a clear conclusion: the NVIDIA GeForce GTX 970M is the superior performer for any compute-intensive or graphics-heavy workload. Its 69.6% lead in OpenCL and 73.3% lead in Vulkan, combined with its larger memory pool, wider memory bus, and significantly higher resource counts, make it the appropriate choice for gaming laptops or mobile workstations that require substantial GPU capability.

Users should select the GTX 970M if their primary applications involve 3D rendering, modern game titles, GPU-accelerated compute, or any task that benefits from high memory bandwidth. The 6 GB GDDR5 memory and 120.3 GB/s bandwidth provide ample headroom for large textures and complex scenes, while the 2.657 TFLOPS FP32 performance supports demanding computational workloads.

The 840M, conversely, suits users whose needs are limited to basic graphics output, office productivity, and light multimedia. Its 33 W TDP and IGP form factor indicate it is intended for power-constrained ultraportable laptops where battery life and thermal management take precedence over performance. The 2 GB DDR3 memory and 863.2 GFLOPS compute are adequate for 2D applications and casual use, but the benchmark results show it cannot compete with the GTX 970M in any recorded test.

The percentile rankings provide additional context: the 840M sits at the 31st percentile of all GPUs, while the GTX 970M sits at the 27th percentile. These rankings reflect the full database population, not just the head-to-head comparison, and both cards occupy similar overall positions. However, their nearest rivals tell different stories. The 840M competes with the GeForce 930A, GeForce GTX 980M, Radeon R7 M445, and GeForce 940M, all within 0.7% of its average score. The GTX 970M competes with the Quadro M3000M, Radeon R5 M320, Radeon RX 9060 XT 16 GB, and Radeon R5 M230, all within 1.1% of its average.

For the user making a choice between these two specific cards, the decision hinges on workload requirements. If the system must handle modern 3D applications, the GTX 970M is the only viable option based on the recorded data. If the system only needs display output and light compute, the 840M's lower power draw and smaller footprint may be preferable, but the performance gap remains decisive in every measured category.