NVIDIA GeForce GTX 680M vs NVIDIA GRID K2 Comparison

The NVIDIA GRID K2 and the NVIDIA GeForce GTX 680M both rely on the same GK104 chip and the Kepler architecture, but they serve entirely different purposes. The recorded benchmark data shows the GRID K2 ahead in every tested workload, yet the GTX 680M holds its own in efficiency and mobile integration. This analysis breaks down where each card wins, what separates them internally, and who should choose which based strictly on the numbers.

Where Each One Wins

The GRID K2 wins outright in raw compute performance across both benchmark suites. In the Geekbench Metal test, it scores 5,557 against 4,815 for the GTX 680M, a lead of 15.4%. In the Geekbench OpenCL test, the GRID K2 reaches 10,602 while the GTX 680M manages 9,230, a 14.9% advantage. These are not marginal differences; they represent a consistent tier gap in execution throughput. The GRID K2 also holds a higher average benchmark score of 8,080 versus 7,023 for the GTX 680M, and it sits at the 42nd percentile of all GPUs in the database, compared to the 39th percentile for the GTX 680M.

The GTX 680M, however, wins in power efficiency and physical footprint. Its thermal design power is 100 W, while the GRID K2 draws 225 W. The GTX 680M is an MXM Module with no power connectors required, whereas the GRID K2 is a dual-slot card needing a 1x 6-pin plus 1x 8-pin connector and a 550 W suggested power supply. For portable systems, the GTX 680M's display outputs are described as "Portable Device Dependent," meaning it can drive laptop screens or external panels as configured. The GRID K2 has no display outputs at all, making it unsuitable for any direct video connection. Thus, the GRID K2 is the clear winner for compute-heavy server or virtualized workloads, while the GTX 680M wins for mobile graphics where power draw and physical size are critical constraints.

Architecture Differences

Both GPUs share the same fundamental silicon: a 28 nm process from TSMC, 3,540 million transistors, and a 294 mm² die size. The transistor density is identical at 12.0 million per square millimeter. The architecture is Kepler for both, and the chip is GK104. However, the implementation differs significantly.

The GRID K2 has more execution resources. It packs 1,536 shading units, 128 texture mapping units, and 32 raster operations pipelines. The GTX 680M has 1,344 shading units, 112 texture units, and the same 32 ROPs. That difference in shading and texture hardware directly explains the GRID K2's higher theoretical rates: its pixel rate is 23.84 GPixel/s versus 21.22 GPixel/s, and its texture rate is 95.36 GTexel/s versus 84.90 GTexel/s. The GRID K2 also delivers 2.289 TFLOPS of FP32 compute, while the GTX 680M provides 2.038 TFLOPS.

Memory clocks differ as well. The GRID K2 runs its GDDR5 memory at 1250 MHz, translating to 5 Gbps effective, while the GTX 680M runs at 900 MHz, or 3.6 Gbps effective. Both have 4 GB of GDDR5 on a 256-bit bus, but the GRID K2 achieves 160.0 GB/s of bandwidth versus 115.2 GB/s for the GTX 680M. The GRID K2 also has a PCIe 3.0 x16 interface, while the GTX 680M uses an MXM-B (3.0) connection, which is a laptop-oriented slot.

The clock behavior is telling: the GTX 680M has a base clock of 719 MHz and a boost clock of 758 MHz, while the GRID K2 lists no base or boost clocks in the database, likely because its clocks are managed differently for sustained server operation. The GRID K2 belongs to the GRID generation, while the GTX 680M is part of the GeForce 600M generation. The GTX 680M has a predecessor in the GeForce 500M series and a successor in the GeForce 700M series, while the GRID K2 has no predecessor or successor listed.

Head-to-Head Benchmarks

The two recorded benchmark results show a clear pattern: the GRID K2 wins both by nearly identical margins. In Geekbench Metal, the GRID K2 scores 5,557 against 4,815, a 15.4% advantage. In Geekbench OpenCL, the GRID K2 scores 10,602 against 9,230, a 14.9% advantage. The win count is 2 for the GRID K2 and 0 for the GTX 680M.

Looking at the nearest rivals in the database puts these scores in context. The GRID K2's average score of 8,080 is just 0.2% above the NVIDIA GeForce GTX 650 Ti Boost (8,067) and 0.3% above the GTX 650 Ti (8,053). It is 0.2% below the GeForce 945M (8,099) and 0.5% above the GeForce GTX 880M (8,040). This places the GRID K2 in a tight cluster of mid-range GPUs from several generations. The GTX 680M, with an average of 7,023, sits 0.2% below the NVIDIA T600 (7,035), 0.7% above the AMD Radeon R5 M240 (6,975), and 1.1% above the GeForce GTX 675M (6,946). It is 1.9% below the GeForce GTX 970 (7,157). The GRID K2, therefore, is not just faster than the GTX 680M; it occupies a different performance bracket entirely, roughly one tier higher.

The bandwidth gap is a key differentiator in the head-to-head. The GRID K2's 160.0 GB/s versus 115.2 GB/s means it can feed its extra 192 shading units and 16 additional texture units more effectively. In OpenCL, which often stresses memory throughput and arithmetic intensity, the GRID K2's 14.9% lead matches its 12.3% advantage in shading units and 14.3% advantage in texture units. In Metal, the 15.4% lead is slightly larger, suggesting that the GRID K2's higher pixel rate (23.84 versus 21.22 GPixel/s) also contributes to geometry and rasterization tasks. The GTX 680M's boost clock of 758 MHz is higher than the GRID K2's unspecified clocks, but that does not compensate for the resource deficit.

FAQ

Q: Which GPU has a higher average benchmark score?

A: The NVIDIA GRID K2 has an average benchmark score of 8,080, while the NVIDIA GeForce GTX 680M scores 7,023. That is a 15% difference in favor of the GRID K2.

Q: Are these GPUs based on the same chip?

A: Yes, both use the GK104 chip on the Kepler architecture, built on a 28 nm process at TSMC. They share the same 3,540 million transistors and 294 mm² die size.

Q: What is the memory bandwidth difference?

A: The GRID K2 has 160.0 GB/s of bandwidth from its 1250 MHz memory clock, while the GTX 680M has 115.2 GB/s from a 900 MHz memory clock. Both use 4 GB of GDDR5 on a 256-bit bus.

Q: Can the GRID K2 output video to a display?

A: No, the GRID K2 has no display outputs. The GTX 680M, by contrast, has outputs described as "Portable Device Dependent," meaning it can drive screens in a laptop configuration.

Q: How does the power draw compare?

A: The GRID K2 has a thermal design power of 225 W and requires a dual-slot cooler with a 1x 6-pin and 1x 8-pin power connector. The GTX 680M has a 100 W TDP, uses an MXM Module form factor, and requires no power connectors.

Q: Which GPU is closer to the NVIDIA GeForce GTX 970 in performance?

A: The GTX 680M is 1.9% below the GeForce GTX 970 (7,157 average score), while the GRID K2 is not listed as a rival to that card. The GRID K2 sits near the GTX 650 Ti Boost, GTX 650 Ti, and GeForce 945M.

The Verdict

The data points to a clear split: the NVIDIA GRID K2 is the choice for anyone prioritizing raw compute throughput, whether for virtualized graphics or general-purpose processing. It wins both benchmark tests by roughly 15%, has a higher pixel rate, texture rate, FP32 throughput, and memory bandwidth. Its average score places it in a performance tier above the GTX 680M, and its percentile rank of 42 versus 39 confirms that it outpaces a majority of the database entries that the GTX 680M does not. The GRID K2's lack of display outputs is irrelevant for its intended server role.

The NVIDIA GeForce GTX 680M, on the other hand, is the only viable option for mobile systems. Its 100 W TDP, MXM Module form factor, and no external power connectors make it suitable for laptops where the GRID K2's 225 W draw and dual-slot design are physically impossible. The GTX 680M also has a boost clock of 758 MHz, which is a listed clock speed, whereas the GRID K2 has no base or boost clocks recorded, suggesting a different clock management strategy. For a user who needs a discrete GPU in a portable chassis, the GTX 680M's lower power draw and connectivity options outweigh its 15% performance deficit.

There is no scenario in the recorded data where the GTX 680M outperforms the GRID K2 in compute benchmarks. But the GTX 680M wins on integration and efficiency, which are not captured in raw scores. The GRID K2 also has a launch MSRP of 5,199 USD, a figure that reflects its enterprise positioning, while the GTX 680M has no launch MSRP listed in the database. In short: take the GRID K2 for performance, take the GTX 680M for portability.

Specification Differences

The following fields differ between the two GPUs, based solely on the database records:

• Generation: GRID (K2) for the GRID K2; GeForce 600M for the GTX 680M.
• Base clock: Not listed for the GRID K2; 719 MHz for the GTX 680M.
• Boost clock: Not listed for the GRID K2; 758 MHz for the GTX 680M.
• Memory clock: 1250 MHz (5 Gbps effective) for the GRID K2; 900 MHz (3.6 Gbps effective) for the GTX 680M.
• Memory bandwidth: 160.0 GB/s for the GRID K2; 115.2 GB/s for the GTX 680M.
• Shading units: 1,536 for the GRID K2; 1,344 for the GTX 680M.
• Texture mapping units: 128 for the GRID K2; 112 for the GTX 680M.
• Pixel rate: 23.84 GPixel/s for the GRID K2; 21.22 GPixel/s for the GTX 680M.
• Texture rate: 95.36 GTexel/s for the GRID K2; 84.90 GTexel/s for the GTX 680M.
• FP32 performance: 2.289 TFLOPS for the GRID K2; 2.038 TFLOPS for the GTX 680M.
• Thermal design power: 225 W for the GRID K2; 100 W for the GTX 680M.
• Slot width: Dual-slot for the GRID K2; MXM Module for the GTX 680M.
• Power connectors: 1x 6-pin + 1x 8-pin for the GRID K2; None for the GTX 680M.
• Suggested PSU: 550 W for the GRID K2; not listed for the GTX 680M.
• Bus interface: PCIe 3.0 x16 for the GRID K2; MXM-B (3.0) for the GTX 680M.
• Display outputs: No outputs for the GRID K2; Portable Device Dependent for the GTX 680M.
• Dimensions: The GRID K2 is 267 mm (10.5 inches) long; the GTX 680M has no length listed.
• Release date: 2013-05-10 for the GRID K2; 2012-06-03 for the GTX 680M.
• Predecessor: Not listed for the GRID K2; GeForce 500M for the GTX 680M.
• Successor: Not listed for the GRID K2; GeForce 700M for the GTX 680M.
• Launch MSRP: 5,199 USD for the GRID K2; not listed for the GTX 680M.
• Geekbench Metal score: 5,557 for the GRID K2; 4,815 for the GTX 680M.
• Geekbench OpenCL score: 10,602 for the GRID K2; 9,230 for the GTX 680M.
• Average benchmark score: 8,080 for the GRID K2; 7,023 for the GTX 680M.
• Percentile vs all GPUs: 42 for the GRID K2; 39 for the GTX 680M.

Identical specifications include the 28 nm process node, TSMC foundry, 3,540 million transistors, 294 mm² die size, 12.0 million transistors per mm², 4 GB GDDR5 memory, 256-bit memory bus, 32 ROPs, DirectX 12 (11_0), OpenGL 4.6, Vulkan 1.2.175, and end-of-life production status. Both have no ray tracing cores and no tensor cores. The GTX 680M's predecessor and successor are listed, while the GRID K2's are not, reflecting their distinct product lifecycles. The data shows two GPUs from the same silicon family that diverged sharply in purpose, and the benchmark results confirm that divergence.