NVIDIA GeForce GTX 675MX vs NVIDIA GeForce GTX 870M Comparison

NVIDIA GeForce GTX 870M and NVIDIA GeForce GTX 675MX are both Kepler-generation mobile GPUs built by NVIDIA, but they serve different eras of the laptop market. The GTX 870M, from the GeForce 800M generation, consistently outperforms the older GTX 675MX from the GeForce 600M generation across recorded benchmark data. The database shows the GTX 870M leading in both tested workloads, with a significant margin that makes it the stronger choice for compute-heavy tasks, while the GTX 675MX remains a capable option for its time.

FAQ

Q: Which GPU is faster in the recorded benchmarks?

A: The NVIDIA GeForce GTX 870M wins both head-to-head tests. It scores 7288 in Geekbench Metal versus 6131 for the GTX 675MX, an 18.9% advantage. In Geekbench OpenCL, the GTX 870M scores 12630 against 10723, a 17.8% lead.

Q: How do their average benchmark scores compare?

A: The GTX 870M has an average benchmark score of 9959, placing it in the 48th percentile of all GPUs. The GTX 675MX averages 8427, sitting in the 43rd percentile. This means the GTX 870M outperforms roughly 48% of all GPUs, while the GTX 675MX outperforms about 43%.

Q: What are the closest rivals to each card according to the database?

A: The GTX 870M's nearest rival is the AMD Radeon Pro 5300M with an average score of 10013, which is 0.5% higher. The GTX 675MX's closest competitor is the AMD Radeon 880M at 8436, just 0.1% higher. Both cards sit in a tight competitive band around their respective scores.

Q: Do both cards use the same graphics architecture?

A: Yes, both are based on NVIDIA's Kepler architecture. They share the same GK104 chip, manufactured by TSMC on a 28 nm process. Both have 3,540 million transistors and a die size of 294 mm².

Q: Is there a difference in memory capacity and bus width?

A: Yes. The GTX 870M comes with 3 GB of GDDR5 memory on a 192-bit bus, while the GTX 675MX has 2 GB of GDDR5 on a wider 256-bit bus. Despite the narrower bus, the GTX 870M achieves higher bandwidth at 120.0 GB/s versus 115.2 GB/s.

Q: Which card has more shading units and texture units?

A: The GTX 870M has 1344 shading units and 112 texture mapping units. The GTX 675MX has 960 shading units and 80 TMUs. However, the GTX 675MX has more render output units: 32 versus 24 on the GTX 870M.

Architecture Differences

Both GPUs are built on the same Kepler architecture, using the identical GK104 chip. The foundry is TSMC with a 28 nm process node, and the transistor count is equal at 3,540 million with a die size of 294 mm². The transistor density works out to 12.0 million per mm² on both. This means the fundamental silicon design is shared, but the configuration differs significantly.

The GTX 870M is a higher-tier implementation of the GK104. It packs 1344 shading units, 112 TMUs, and 24 ROPs. The GTX 675MX is a cut-down version with 960 shading units, 80 TMUs, and 32 ROPs. The GTX 870M trades fewer ROPs for substantially more shader and texture hardware, which directly impacts compute throughput and texture filtering. The GTX 675MX's higher ROP count suggests a design geared toward fill-rate-bound scenarios, but the raw shader deficit limits its overall capability.

Memory subsystems also diverge. The GTX 870M uses 3 GB of GDDR5 on a 192-bit interface with a memory clock of 1250 MHz, translating to 5 Gbps effective and 120.0 GB/s bandwidth. The GTX 675MX has 2 GB of GDDR5 on a 256-bit bus with a 900 MHz clock (3.6 Gbps effective), yielding 115.2 GB/s. Despite the narrower bus, the GTX 870M's faster memory clock gives it a bandwidth advantage of about 4.2%.

The compute rates reflect the core differences. The GTX 870M delivers a pixel rate of 27.08 GPixel/s and a texture rate of 108.3 GTexel/s, with FP32 performance at 2.599 TFLOPS. The GTX 675MX produces 13.08 GPixel/s and 52.32 GTexel/s, with FP32 at 1,255.7 GFLOPS. The GTX 870M is roughly twice as fast in pixel and texture throughput and more than double in FP32 compute. Both cards share the same API support: DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.2.175. They also share the same MXM-B (3.0) bus interface and 100 W TDP.

Head-to-Head Benchmarks

The database records two head-to-head tests: Geekbench Metal and Geekbench OpenCL. The GTX 870M wins both, giving it a 2-0 record against the GTX 675MX.

In Geekbench Metal, the GTX 870M scores 7288 against the GTX 675MX's 6131. That is a delta of 18.9%, meaning the newer card is nearly a fifth faster in this compute API. For context, this is not a marginal edge; it represents a clear generational step in usable performance for Metal-accelerated workloads.

In Geekbench OpenCL, the gap is slightly narrower but still decisive. The GTX 870M posts 12630 versus 10723 for the GTX 675MX, a 17.8% lead. OpenCL is a more general-purpose compute benchmark, and the result aligns with the raw FP32 and texture rate advantages seen in the specifications.

Looking at absolute positioning, the GTX 870M's average score of 9959 puts it within 0.5% of the AMD Radeon Pro 5300M (10013) and 0.8% below the NVIDIA Quadro K5100M (10043). It also edges out the NVIDIA Quadro 6000 by 1.1% (9846). The GTX 675MX, by contrast, sits at 8427, which is 0.1% behind the AMD Radeon 880M (8436), 0.4% behind the NVIDIA GeForce MX330 (8458), and 0.4% behind the AMD Radeon HD 8870M (8462). It leads the AMD Radeon R9 M375X by 1.2% (8325). The GTX 870M's nearest rivals cluster around the 10,000 mark, while the GTX 675MX's rivals hover near 8,400, confirming the performance tier separation.

The wins are consistent across both test types. There is no test in the database where the GTX 675MX pulls ahead. The margin is larger in Metal than in OpenCL, but both favor the GTX 870M by a substantial double-digit percentage.

Specification Differences

The two cards differ in several key specification fields. The GTX 870M belongs to the GeForce 800M generation, while the GTX 675MX comes from the GeForce 600M generation. The GTX 870M was released on March 11, 2014, whereas the GTX 675MX launched on September 30, 2012.

Clocks are another difference. The GTX 870M has a base clock of 941 MHz and a boost clock of 967 MHz. The GTX 675MX has no recorded base or boost clock values in the database. Its memory runs at 900 MHz (3.6 Gbps effective), while the GTX 870M's memory runs at 1250 MHz (5 Gbps effective).

Memory configuration is a major split. The GTX 870M has 3 GB of GDDR5 on a 192-bit bus with 120.0 GB/s bandwidth. The GTX 675MX has 2 GB of GDDR5 on a 256-bit bus with 115.2 GB/s bandwidth. The GTX 870M's larger capacity and faster effective memory speed outweigh the narrower bus.

Compute resources differ sharply. The GTX 870M has 1344 shading units, 112 TMUs, and 24 ROPs. The GTX 675MX has 960 shading units, 80 TMUs, and 32 ROPs. The GTX 870M leads in shaders and TMUs, but the GTX 675MX has more ROPs.

Rates and throughput metrics are all higher on the GTX 870M. Pixel rate is 27.08 GPixel/s versus 13.08 GPixel/s, texture rate is 108.3 GTexel/s versus 52.32 GTexel/s, and FP32 is 2.599 TFLOPS versus 1,255.7 GFLOPS. The TDP is identical at 100 W for both, and both use the MXM-B (3.0) interface with no power connectors.

The Verdict

The data is unambiguous: the NVIDIA GeForce GTX 870M is the superior GPU in every recorded benchmark. It wins both head-to-head tests with margins of 18.9% and 17.8%, and its average benchmark score is 9959 versus 8427 for the GTX 675MX. In percentile terms, the GTX 870M sits at 48 versus 43 for the older card, a meaningful gap in overall GPU hierarchy.

The GTX 870M's advantage stems from its fuller GK104 implementation. More shading units (1344 versus 960), more TMUs (112 versus 80), and higher clocks (941 MHz base versus no recorded base clock) translate directly to higher FP32 throughput at 2.599 TFLOPS versus 1,255.7 GFLOPS. The GTX 675MX's only spec win is its 32 ROPs against 24, and that does not translate to a benchmark victory in the recorded tests.

For a user choosing between these two in a laptop, the GTX 870M is the pick if the workload involves general compute, OpenCL, or Metal acceleration. It is also the better choice for texture-heavy rendering given its 108.3 GTexel/s rate. The GTX 675MX should only be considered if the system already has it and the workload is light, or if the wider 256-bit bus is somehow more relevant than the actual bandwidth deficit, which the data does not support.

Where Each One Wins

The GTX 870M wins in all recorded scenarios. In Geekbench Metal, its 18.9% lead indicates a strong advantage for applications leveraging Apple's Metal API, such as certain creative suites or compute tasks on compatible systems. In Geekbench OpenCL, the 17.8% edge covers a broader range of cross-platform compute workloads, including physics simulation, image processing, and general GPGPU tasks.

The GTX 870M's texture rate of 108.3 GTexel/s versus 52.32 GTexel/s makes it the clear choice for games or applications that stress texture filtering and shader complexity. Its 2.599 TFLOPS FP32 performance is more than double the GTX 675MX's 1,255.7 GFLOPS, so any floating-point-heavy task will favor the newer card.

The GTX 675MX has no benchmark wins recorded. Its only notable specification advantage is the 32 ROPs versus 24, which could theoretically benefit fill-rate-bound scenarios at low resolutions. However, the recorded data shows no test where this translates to a victory. The GTX 675MX also has a wider 256-bit memory bus, but the GTX 870M's faster memory clock delivers higher real bandwidth at 120.0 GB/s versus 115.2 GB/s.

In practical terms, the GTX 870M is the default recommendation for anyone needing mobile GPU compute performance, whether for gaming, content creation, or general acceleration. The GTX 675MX is an older part that has been superseded, and the database confirms it trails by roughly 18% in both tested APIs. Users with the GTX 675MX should expect lower scores across the board, with no workload category where it overtakes the GTX 870M.