NVIDIA GeForce 945M vs NVIDIA GeForce GTX 680M Comparison

Head-to-Head Benchmarks

The only direct benchmark comparison available in the database is the Geekbench OpenCL test, and it delivers a clear verdict in favor of the GeForce GTX 680M. The GTX 680M scores 9,230 points, while the GeForce 945M records 8,099 points. That is a 12.3% deficit for the 945M, a substantial margin in compute workloads that stresses parallel throughput. The GTX 680M wins the sole head-to-head contest, giving it a 1-0 record over the 945M in recorded comparisons.

Context from the nearest rivals makes the result more telling. The 945M sits at the 42nd percentile of all GPUs in the database, with an average benchmark score of 8,099. Its closest competitors include the NVIDIA GRID K2 at 8,080 (0.2% behind the 945M), the GTX 650 Ti Boost at 8,067 (0.4% behind), the AMD Radeon R9 M360 at 8,129 (0.4% ahead), and the GTX 950M at 8,135 (0.4% ahead). These deltas are tiny, meaning the 945M is clustered with a group of mid-range parts where performance differences are within noise. The GTX 680M, by contrast, sits at the 39th percentile with an average score of 7,023 across all its recorded tests, though that figure includes a separate Geekbench Metal result of 4,815, which drags its average down. In the OpenCL test specifically, the 680M is 12.3% faster than the 945M, and that gap is far larger than any of the sub-1% differences seen among the 945M's immediate rivals.

Looking at the 680M's nearest rivals, its OpenCL strength becomes clearer. The T600 averages 7,035 (0.2% behind the 680M), the Radeon R5 M240 averages 6,975 (0.7% behind), the GTX 675M averages 6,946 (1.1% behind), and the GTX 970 averages 7,157 (1.9% ahead). The 680M's OpenCL score of 9,230 is well above all of these averages, indicating that in the specific compute test, it punches far above its percentile ranking. The 945M, meanwhile, sits in a tight pack where no rival is more than 0.4% away, suggesting its OpenCL performance is squarely mainstream rather than exceptional.

Where Each One Wins

The GTX 680M wins the compute benchmark outright, and the data implies it should also lead in raw throughput-centric tasks. Its architecture packs 1,344 shading units, 112 texture mapping units, and 32 raster operation units, compared to the 945M's 640 shading units, 40 TMUs, and 16 ROPs. In texture fill rate, the 680M achieves 84.90 GTexel/s versus 40.80 GTexel/s for the 945M, more than double. Pixel rate also favors the 680M at 21.22 GPixel/s versus 16.32 GPixel/s, a 30% advantage. Single-precision floating point performance is 2.038 TFLOPS for the 680M against 1,305.6 GFLOPS for the 945M, a 56% lead. For any workload that scales with shader count or memory bandwidth, the 680M is the clear choice. Its memory bus is 256-bit wide with 115.2 GB/s of bandwidth, versus the 945M's 128-bit bus and 28.80 GB/s. That bandwidth gap is nearly 4x, which will matter in texture-heavy scenes and large data transfers.

The 945M does have its own territory. It is built on the same 28 nm process but uses the Maxwell architecture, which is newer than the Kepler design in the 680M. The 945M's boost clock reaches 1,020 MHz versus 758 MHz for the 680M, and its base clock is 928 MHz versus 719 MHz. Higher clocks per shader can help in latency-sensitive or poorly parallelized workloads, even if raw throughput is lower. The 945M also draws 75 W compared to 100 W for the 680M, meaning it generates less heat and may sustain boost clocks longer in thermally constrained laptop chassis. The 945M's transistor density is slightly higher at 12.6M per mm² versus 12.0M per mm², and its die is much smaller at 148 mm² versus 294 mm², which typically translates to lower production cost and better yield, though the database does not confirm pricing. For users prioritizing power efficiency, the 945M is the more sensible option.

Architecture Differences

The fundamental split is Maxwell versus Kepler. The 945M uses the GM107 chip, a 28 nm TSMC design with 1,870 million transistors on a 148 mm² die. The 680M uses the GK104 chip, also 28 nm TSMC, but with 3,540 million transistors on a 294 mm² die. The 680M has nearly twice the transistor count and exactly double the die area, which explains its higher shading unit count: 1,344 versus 640. The 680M also has 112 TMUs versus 40, and 32 ROPs versus 16. Memory configurations diverge sharply. The 945M ships with 2 GB of DDR3 on a 128-bit bus, achieving 28.80 GB/s. The 680M ships with 4 GB of GDDR5 on a 256-bit bus, achieving 115.2 GB/s. The memory clock is the same at 900 MHz, but the effective data rate differs: 1,800 Mbps for the 945M versus 3.6 Gbps for the 680M, due to the GDDR5 architecture.

Clock speeds favor the 945M. Its base is 928 MHz and boost is 1,020 MHz, while the 680M operates at 719 MHz base and 758 MHz boost. This means the 945M's Maxwell cores run at a higher frequency, partially compensating for having fewer of them. The 945M's FP32 output is 1,305.6 GFLOPS, while the 680M reaches 2.038 TFLOPS, so the clock advantage does not close the gap. API support is nearly identical: both support DirectX 12 (11_0) and OpenGL 4.6. The 945M supports Vulkan 1.4, while the 680M supports Vulkan 1.2.175, a minor version difference that may affect future compatibility with newer Vulkan features. Both use the MXM-B (3.0) bus interface and are MXM modules with no power connectors listed. Neither has dedicated RT or tensor cores, as both predate those features. The 680M's release date is June 2012, while the 945M arrived in October 2015, a gap of over three years that shows in the architectural maturity.

The Verdict

The data points to a split decision based on workload. For pure compute performance, the GTX 680M is the winner. It leads by 12.3% in OpenCL, has more than double the texture fill rate, a 30% pixel rate advantage, and 56% higher FP32 throughput. Its 115.2 GB/s memory bandwidth is a decisive factor for any data-intensive task, and its 4 GB frame buffer offers double the capacity of the 945M, which matters for high-resolution textures or larger datasets. The 680M also sits close to the GTX 970 in its rival cluster, averaging 7,157, and beats the T600, R5 M240, and GTX 675M in its own comparisons, indicating it holds its own against newer parts despite its 2012 origin.

The 945M is not without merit. Its Maxwell architecture brings higher clocks, lower power draw at 75 W versus 100 W, and a smaller die that suggests better thermal behavior. For users in thin-and-light laptops where heat is a constraint, the 945M may sustain performance better over long sessions. Its Vulkan 1.4 support is newer, which could matter for games that rely on the latest API extensions. However, the benchmark data does not show a single test where the 945M beats the 680M. The 945M's nearest rivals are all within 0.4%, placing it in a crowded mid-range segment, while the 680M's OpenCL score of 9,230 stands well above its own average and its rivals' averages. The verdict is straightforward: choose the GTX 680M for maximum compute throughput, choose the 945M for power efficiency and newer API support, but accept that in raw performance, the 680M is the stronger part.

FAQ

Q: Which GPU wins in the OpenCL benchmark?

A: The NVIDIA GeForce GTX 680M scores 9,230 in Geekbench OpenCL, while the GeForce 945M scores 8,099. The 680M wins by 12.3%.

Q: How does the 945M compare to its closest rivals?

A: The 945M's nearest rivals are all within 0.4% of its score. The GRID K2 is 0.2% behind, the GTX 650 Ti Boost is 0.4% behind, the R9 M360 is 0.4% ahead, and the GTX 950M is 0.4% ahead.

Q: What is the memory bandwidth difference?

A: The GTX 680M has 115.2 GB/s of bandwidth from a 256-bit bus with GDDR5 memory. The 945M has 28.80 GB/s from a 128-bit bus with DDR3 memory. The 680M has exactly four times the bandwidth.

Q: Does the 945M have any advantage in clock speed?

A: Yes. The 945M runs at a 928 MHz base clock and 1,020 MHz boost, while the 680M runs at 719 MHz base and 758 MHz boost. The 945M's clocks are 29% higher at base and 35% higher at boost.

Q: What are the TDP ratings for each GPU?

A: The 945M is rated at 75 W, while the 680M is rated at 100 W. The 945M consumes 25% less power.

Q: Which GPU supports a newer Vulkan version?

A: The 945M supports Vulkan 1.4, while the 680M supports Vulkan 1.2.175. Both support DirectX 12 (11_0) and OpenGL 4.6.

Specification Differences

| Specification | GeForce 945M | GeForce GTX 680M |

|---------------|--------------|------------------|

| Architecture | Maxwell | Kepler |

| Chip | GM107 | GK104 |

| Transistors | 1,870 million | 3,540 million |

| Die Size | 148 mm² | 294 mm² |

| Transistor Density | 12.6M / mm² | 12.0M / mm² |

| Base Clock | 928 MHz | 719 MHz |

| Boost Clock | 1,020 MHz | 758 MHz |

| Memory Size | 2 GB | 4 GB |

| Memory Type | DDR3 | GDDR5 |

| Memory Bus Width | 128 bit | 256 bit |

| Memory Bandwidth | 28.80 GB/s | 115.2 GB/s |

| Effective Memory Speed | 1,800 Mbps | 3.6 Gbps |

| Shading Units | 640 | 1,344 |

| TMUs | 40 | 112 |

| ROPs | 16 | 32 |

| Pixel Rate | 16.32 GPixel/s | 21.22 GPixel/s |

| Texture Rate | 40.80 GTexel/s | 84.90 GTexel/s |

| FP32 Performance | 1,305.6 GFLOPS | 2.038 TFLOPS |

| TDP | 75 W | 100 W |

| Vulkan Version | 1.4 | 1.2.175 |

| Release Date | October 2015 | June 2012 |

| Generation | GeForce 900M | GeForce 600M |

| Predecessor | GeForce 800M | GeForce 500M |

| Successor | GeForce 10 Mobile | GeForce 700M |

| Percentile | 42nd | 39th |

| Average Benchmark Score | 8,099 | 7,023 |