NVIDIA GeForce GTX 780M vs NVIDIA GeForce GTX 870M Comparison

The NVIDIA GeForce GTX 780M and GTX 870M are both end-of-life Kepler mobile GPUs, but the data reveals a clear split between raw compute capability and architectural tuning. The GTX 780M wins both head-to-head benchmarks, yet the GTX 870M is the newer part with a more efficient power profile. The verdict from the data is straightforward: the GTX 780M is the stronger performer for compute-heavy workloads, while the GTX 870M is the more balanced choice for thermally constrained systems where the 22-watt lower TDP matters more than a 1.1% OpenCL advantage. The GTX 780M’s average benchmark score of 11,261 places it at the 50th percentile among all GPUs, while the GTX 870M sits at 9,959 and the 48th percentile. For users who prioritize maximum frame rates and compute throughput, the GTX 780M is the pick; for those who need lower power draw with slightly fewer shading units, the GTX 870M warrants consideration.

The Verdict

The GTX 780M is the decisive winner in this comparison, taking 2 wins out of 2 head-to-head benchmarks. Its most significant victory comes in Geekbench Metal, where it scores 8,319 against the GTX 870M’s 7,288 — a 14.1% advantage. This is not a marginal difference; it is a substantial lead that indicates the GTX 780M’s higher memory bandwidth and wider bus width translate directly into better API-specific performance. In Geekbench OpenCL, the margin narrows dramatically to just 1.1%, with scores of 12,769 versus 12,630, showing that in compute-heavy, memory-access-light workloads, the two GPUs are nearly interchangeable. The average benchmark score reinforces this hierarchy: the GTX 780M averages 11,261, which is 13.1% higher than the GTX 870M’s 9,959. When placed against their respective nearest rivals, the GTX 780M leads the AMD Radeon Pro WX 3200 by 0.3% (11,261 vs 11,228), while the GTX 870M trails the AMD Radeon Pro 5300M by 0.5% (9,959 vs 10,013). The verdict is clear: the GTX 780M is the superior performer, and the GTX 870M’s lower power draw does not compensate for its performance deficit in this data set.

Architecture Differences

Both GPUs share the same fundamental architecture — Kepler on a 28 nm TSMC process, with the GK104 chip containing 3,540 million transistors on a 294 mm² die. The transistor density is identical at 12.0M per mm². However, the two parts diverge significantly in how they utilize that silicon. The GTX 780M deploys 1,536 shading units, 128 texture mapping units, and 32 raster output units, while the GTX 870M reduces these counts to 1,344 shading units, 112 TMUs, and 24 ROPs. This means the GTX 780M has 14.3% more shaders, 14.3% more TMUs, and 33.3% more ROPs than the GTX 870M. The memory subsystem is also starkly different: the GTX 780M uses a 256-bit bus with 4 GB of GDDR5, delivering 160.0 GB/s of bandwidth, whereas the GTX 870M is limited to a 192-bit bus with 3 GB of GDDR5 and 120.0 GB/s of bandwidth. Despite these reductions, the GTX 870M compensates with higher clocks: its base clock is 941 MHz versus 771 MHz, and its boost clock is 967 MHz versus 797 MHz — increases of 22.0% and 21.3%, respectively. This clock advantage explains why the GTX 870M’s texture rate (108.3 GTexel/s) and pixel rate (27.08 GPixel/s) exceed the GTX 780M’s (102.0 GTexel/s and 25.50 GPixel/s), even though the latter has more hardware units. The GTX 870M also achieves a higher FP32 throughput at 2.599 TFLOPS versus 2.448 TFLOPS, a 6.2% advantage.

Head-to-Head Benchmarks

The Geekbench Metal test is where the GTX 780M establishes its dominance. With a score of 8,319 versus 7,288, the GTX 780M leads by 14.1%. This is the largest delta in the entire comparison and reflects the impact of the GTX 780M’s superior memory bandwidth (160.0 GB/s vs 120.0 GB/s) and wider 256-bit bus, which are critical for Metal’s resource-intensive workloads. The GTX 780M’s higher ROP count (32 vs 24) also contributes to better pixel throughput in this test. In contrast, the Geekbench OpenCL benchmark shows a much tighter race: the GTX 780M scores 12,769 against the GTX 870M’s 12,630, a mere 1.1% difference. This suggests that OpenCL’s compute kernels are less sensitive to memory bandwidth and more dependent on raw shader throughput. Here, the GTX 870M’s higher clock speeds (941 MHz base vs 771 MHz) and greater FP32 output (2.599 TFLOPS vs 2.448 TFLOPS) nearly close the gap, but the GTX 780M’s extra 192 shading units (1,536 vs 1,344) provide just enough edge to secure the win. The data shows that the GTX 780M is not universally faster, but it wins where it matters most for gaming and graphics — the Metal API. The GTX 870M’s only consolation is that its OpenCL deficit is small enough to be considered noise in real-world compute tasks.

Specification Differences

The two GPUs diverge on several key specification fields. The GTX 780M has a base clock of 771 MHz and a boost clock of 797 MHz, while the GTX 870M runs at 941 MHz and 967 MHz respectively. Memory capacity differs: the GTX 780M offers 4 GB, the GTX 870M offers 3 GB. The memory bus is 256-bit for the GTX 780M and 192-bit for the GTX 870M, resulting in bandwidth of 160.0 GB/s versus 120.0 GB/s. Shading units are 1,536 on the GTX 780M and 1,344 on the GTX 870M; TMUs are 128 versus 112; ROPs are 32 versus 24. The pixel rate is 25.50 GPixel/s on the GTX 780M and 27.08 GPixel/s on the GTX 870M. The texture rate is 102.0 GTexel/s on the GTX 780M and 108.3 GTexel/s on the GTX 870M. FP32 compute is 2.448 TFLOPS for the GTX 780M and 2.599 TFLOPS for the GTX 870M. The TDP is 122 W for the GTX 780M and 100 W for the GTX 870M — a 22 W difference that favors the newer part. The release dates also differ: the GTX 780M launched on 2013-05-10, while the GTX 870M launched on 2014-03-11. The GTX 780M’s predecessor is the GeForce 600M and successor is the GeForce 800M, while the GTX 870M’s predecessor is the GeForce 700M and successor is the GeForce 900M. Both use the MXM-B (3.0) bus interface and have no power connectors.

FAQ

Q: Which GPU has a higher average benchmark score?

A: The NVIDIA GeForce GTX 780M has an average benchmark score of 11,261, which is 13.1% higher than the GTX 870M’s 9,959. This places the GTX 780M at the 50th percentile among all GPUs, compared to the GTX 870M’s 48th percentile.

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

A: The GTX 780M scores 8,319 in Geekbench Metal, which is 14.1% higher than the GTX 870M’s 7,288. This is the largest performance gap between the two in any test.

Q: Is the Geekbench OpenCL performance difference significant?

A: No, the difference is minimal. The GTX 780M scores 12,769 while the GTX 870M scores 12,630 — a margin of just 1.1%. This indicates near-parity in general compute workloads.

Q: Which GPU has more memory bandwidth?

A: The GTX 780M has 160.0 GB/s of memory bandwidth, which is 33.3% higher than the GTX 870M’s 120.0 GB/s. This is due to the GTX 780M’s wider 256-bit bus compared to the GTX 870M’s 192-bit bus.

Q: What is the TDP difference between the two?

A: The GTX 870M has a TDP of 100 W, which is 22 W lower than the GTX 780M’s 122 W. This makes the GTX 870M a more power-efficient option for thin-and-light laptops.

Q: Which GPU has more shading units?

A: The GTX 780M has 1,536 shading units, which is 192 more than the GTX 870M’s 1,344 shading units. This 14.3% advantage is partially offset by the GTX 870M’s higher clock speeds.

Where Each One Wins

The GTX 780M wins decisively in memory-bound and graphics-API-specific workloads. Its 160.0 GB/s bandwidth and 256-bit bus give it a 33.3% bandwidth advantage, which directly contributes to its 14.1% lead in Geekbench Metal. The higher ROP count (32 vs 24) also makes it the better choice for games that are fill-rate limited or heavily use pixel shaders. For users running Metal-based applications or games on macOS or iOS, the GTX 780M is the clear winner. Its 4 GB of VRAM also provides more headroom for high-resolution textures and larger frame buffers. The GTX 780M’s nearest rivals, the AMD Radeon Pro WX 3200 and FirePro W4300, both sit within 0.3% of its average score, indicating it competes well even against newer workstation parts.

The GTX 870M wins on power efficiency and raw shader throughput per watt. With a TDP of 100 W versus 122 W, it draws 18.0% less power while achieving a 6.2% higher FP32 compute rate (2.599 TFLOPS vs 2.448 TFLOPS). Its higher clocks also give it a superior texture rate (108.3 GTexel/s vs 102.0 GTexel/s) and pixel rate (27.08 GPixel/s vs 25.50 GPixel/s). This makes the GTX 870M the better fit for laptops with smaller cooling solutions or smaller batteries, where the 22 W power saving can extend battery life or allow for a thinner chassis. In OpenCL workloads, the GTX 870M is effectively tied with the GTX 780M (1.1% difference), so users who prioritize compute tasks over gaming will not sacrifice much performance. The GTX 870M’s nearest rival, the AMD Radeon Pro 5300M, is only 0.5% faster, showing that the GTX 870M holds its own in its performance class. For users who value portability and lower heat output over the GTX 780M’s Metal advantage, the GTX 870M is the sensible choice.