NVIDIA GeForce GTX 680M vs NVIDIA Quadro K620M Comparison

Head-to-Head Benchmarks

The recorded database contains a single shared benchmark between the NVIDIA GeForce GTX 680M and the NVIDIA Quadro K620M: Geekbench OpenCL. The results are decisive. The GTX 680M scores 9230, while the Quadro K620M scores 5957. This yields a delta of 54.9%, meaning the GTX 680M is more than half again as fast as the K620M in this compute workload. The magnitude of that lead is substantial, and it aligns with the broader positioning of the two parts in the database's rankings.

The GTX 680M's average benchmark score across all recorded tests is 7023. The Quadro K620M's average is 5957, which is exactly its single OpenCL result. The difference in average scores, 1066 points, is roughly 17.9% in favor of the GTX 680M, though the head-to-head delta is far larger because the two parts share only one test. In that shared test, the GTX 680M's 9230 is not just a win; it is a dominant one.

Looking at the percentile standings, the GTX 680M sits at the 39th percentile of all GPUs in the database, while the Quadro K620M sits at the 34th percentile. A five-percentile gap is meaningful, but it understates the chasm seen in the direct comparison. The GTX 680M's nearest rivals include the NVIDIA T600 (average score 7035, 0.2% behind the GTX 680M) and the NVIDIA GeForce GTX 970 (average score 7157, 1.9% ahead). The Quadro K620M's nearest rivals are clustered much lower: the AMD Radeon HD 8730M (5955, 0.0% delta), the AMD Radeon HD 8750M (5970, 0.2% ahead), and the NVIDIA Quadro K4000 (5982, 0.4% ahead). This places the K620M in a performance tier occupied by entry-level mobile and older workstation parts, while the GTX 680M competes with far more capable desktop and mobile GPUs.

In the OpenCL test specifically, the GTX 680M's 9230 is 54.9% higher than the K620M's 5957. For context, the GTX 680M's nearest rivals in overall average score, such as the NVIDIA T600 at 7035, are only 0.2% behind it, meaning the GTX 680M is competitive with that class of GPU. The K620M, by contrast, is essentially tied with the AMD Radeon HD 8730M, which scores 5955, a 0.0% delta. The K620M cannot reach the GTX 680M's performance envelope; it is not even close to the GTX 680M's nearest rival tier.

Where Each One Wins

The wins distribution is straightforward: the GTX 680M wins the only head-to-head benchmark, giving it one win, while the Quadro K620M has zero wins. There is no benchmark in the database where the K620M outperforms the GTX 680M.

For compute-heavy workloads, such as OpenCL tasks, the GTX 680M is the clear choice. Its 54.9% advantage in the shared Geekbench OpenCL test indicates that any application leveraging general-purpose GPU compute will see a large performance uplift with the GTX 680M. The data also shows that the GTX 680M has a second recorded benchmark, Geekbench Metal, with a score of 4815, further demonstrating its versatility across different compute APIs. The K620M has no Metal result recorded, so its compute capabilities are only quantified through OpenCL.

For professional or workstation use cases, the K620M's advantage is not in raw performance but in efficiency and feature positioning. It is a Maxwell-architecture part with a 30 W TDP, versus the GTX 680M's 100 W TDP. This makes the K620M a more power-efficient option for sustained mobile workloads where battery life and thermal constraints are paramount. However, the benchmark data does not show any performance win for the K620M; it only shows a power draw that is one-third of the GTX 680M's. In scenarios where compute throughput is the metric, the GTX 680M wins outright.

The GTX 680M also benefits from a larger memory configuration: 4 GB of GDDR5 on a 256-bit bus, yielding 115.2 GB/s of bandwidth. The K620M has 2 GB of DDR3 on a 64-bit bus, with 16.02 GB/s of bandwidth. That is a 7.2x difference in memory bandwidth, which directly impacts texture-heavy and data-intensive workloads. The GTX 680M's pixel rate is 21.22 GPixel/s versus 8.992 GPixel/s for the K620M, and its texture rate is 84.90 GTexel/s versus 17.98 GTexel/s. Every measurable throughput metric in the database favors the GTX 680M.

Architecture Differences

The two GPUs come from different architectural generations despite sharing the same 28 nm TSMC process node. The GTX 680M is built on the Kepler architecture with the GK104 chip, while the Quadro K620M is built on the Maxwell architecture with the GM108S chip. The process node is identical at 28 nm, and both are fabricated by TSMC, but the transistor budgets differ enormously. The GK104 packs 3,540 million transistors on a 294 mm² die, while the GM108S contains 1,020 million transistors on a 77 mm² die. This yields a transistor density of 12.0M per mm² for the GTX 680M and 13.2M per mm² for the K620M, indicating that Maxwell is a denser design, but the Kepler chip is far larger and more complex.

The GTX 680M has 1344 shading units, 112 texture mapping units (TMUs), and 32 render output units (ROPs). The K620M has 384 shading units, 16 TMUs, and 8 ROPs. These are not close comparisons: the GTX 680M has 3.5x the shading units, 7x the TMUs, and 4x the ROPs. The compute throughput reflects this: the GTX 680M delivers 2.038 TFLOPS of FP32 compute, while the K620M delivers 863.2 GFLOPS, which is less than half.

Clock speeds tell a different story. The K620M runs at a base clock of 1029 MHz and a boost clock of 1124 MHz, while the GTX 680M runs at a base of 719 MHz and a boost of 758 MHz. The Maxwell part is clocked roughly 43% higher at base and 48% higher at boost, but the massive core-count advantage of the Kepler part overwhelms this frequency lead. Memory clocks also differ: the GTX 680M's memory runs at 900 MHz with 3.6 Gbps effective data rate, while the K620M's memory runs at 1001 MHz with 2 Gbps effective. The effective rate is higher on the GTX 680M, and the bus width difference is decisive.

Both parts support DirectX 12 (11_0) and OpenGL 4.6, but the Vulkan support differs. The GTX 680M supports Vulkan 1.2.175, while the K620M supports Vulkan 1.4. The K620M is a newer design with more modern API support, but this does not translate into a performance advantage in the recorded benchmarks. The GTX 680M also carries a GeForce 600M generation label with a predecessor of GeForce 500M and a successor of GeForce 700M. The K620M's generation is listed as Quadro Kepler-M (Kx200M), with a predecessor of Quadro Fermi-M and a successor of Quadro Maxwell-M. The K620M's generation name is somewhat contradictory, as it is a Maxwell chip but is grouped in a Kepler-M generation label.

Power consumption is a major architectural differentiator. The GTX 680M has a TDP of 100 W, while the K620M has a TDP of 30 W. This is a 3.33x difference in favor of the K620M's efficiency. The GTX 680M uses an MXM-B (3.0) bus interface, while the K620M uses MXM-A (3.0). Both are MXM modules, and both have no power connectors, relying on the MXM slot for power delivery. Display outputs are listed as portable device dependent for both, meaning the outputs vary by laptop implementation.

FAQ

Q: Which GPU has the higher average benchmark score?

A: The NVIDIA GeForce GTX 680M has an average benchmark score of 7023, while the NVIDIA Quadro K620M has an average of 5957. The GTX 680M leads by 1066 points.

Q: What is the performance difference in the shared OpenCL benchmark?

A: In Geekbench OpenCL, the GTX 680M scores 9230 and the K620M scores 5957. The GTX 680M wins by 54.9%.

Q: How do their memory subsystems compare?

A: The GTX 680M has 4 GB of GDDR5 on a 256-bit bus with 115.2 GB/s bandwidth. The K620M has 2 GB of DDR3 on a 64-bit bus with 16.02 GB/s bandwidth. The GTX 680M offers over 7 times the memory bandwidth.

Q: Which GPU has higher clock speeds?

A: The Quadro K620M runs at a base clock of 1029 MHz and a boost of 1124 MHz. The GTX 680M runs at a base of 719 MHz and a boost of 758 MHz. The K620M is clocked significantly higher, but the GTX 680M still wins in compute throughput.

Q: What is the TDP difference between the two?

A: The GTX 680M has a TDP of 100 W, while the K620M has a TDP of 30 W. The K620M draws one-third the power of the GTX 680M.

Q: Do both GPUs support the same APIs?

A: Both support DirectX 12 (11_0) and OpenGL 4.6. The GTX 680M supports Vulkan 1.2.175, while the K620M supports Vulkan 1.4, giving the K620M a newer Vulkan version.

Specification Differences

The two GPUs differ across nearly every specification in the database. The GTX 680M uses the GK104 chip with Kepler architecture, while the K620M uses the GM108S chip with Maxwell architecture. The process node is the same (28 nm, TSMC), but the transistor count differs: 3,540 million for the GTX 680M versus 1,020 million for the K620M. Die size is 294 mm² for the GTX 680M and 77 mm² for the K620M, with transistor densities of 12.0M / mm² and 13.2M / mm² respectively.

Clock speeds favor the K620M: base 1029 MHz versus 719 MHz, boost 1124 MHz versus 758 MHz. Memory clocks also differ, with the GTX 680M at 900 MHz (3.6 Gbps effective) and the K620M at 1001 MHz (2 Gbps effective). Memory size is 4 GB GDDR5 for the GTX 680M versus 2 GB DDR3 for the K620M. Bus width is 256 bit versus 64 bit, and bandwidth is 115.2 GB/s versus 16.02 GB/s.

The core configuration is heavily lopsided. The GTX 680M has 1344 shading units, 112 TMUs, and 32 ROPs. The K620M has 384 shading units, 16 TMUs, and 8 ROPs. Pixel rate is 21.22 GPixel/s versus 8.992 GPixel/s, and texture rate is 84.90 GTexel/s versus 17.98 GTexel/s. FP32 compute is 2.038 TFLOPS versus 863.2 GFLOPS.

TDP is 100 W for the GTX 680M and 30 W for the K620M. Both are MXM modules, but the bus interface differs: MXM-B (3.0) for the GTX 680M and MXM-A (3.0) for the K620M. Both have no power connectors and portable device dependent display outputs. In terms of API support, both share DirectX 12 (11_0) and OpenGL 4.6, but Vulkan support differs: 1.2.175 for the GTX 680M versus 1.4 for the K620M. The GTX 680M belongs to the GeForce 600M generation, while the K620M belongs to the Quadro Kepler-M (Kx200M) generation. Both are end-of-life in production status, with the GTX 680M released earlier and the K620M later, but no launch MSRP is recorded for either.