NVIDIA GeForce GTX 675M vs NVIDIA Quadro K620 Comparison

Head-to-Head Benchmarks

The only direct comparison recorded in the database is the Geekbench OpenCL test, and it is a narrow affair. The NVIDIA GeForce GTX 675M posts a score of 6946, while the NVIDIA Quadro K620 trails at 6693. That is a 3.8% advantage for the older mobile gaming chip. In practical terms, this margin is small enough that it would be difficult to notice in most real-world workloads, but it does establish the GTX 675M as the faster card in raw compute throughput.

The GTX 675M’s nearest rivals in the database paint a more interesting picture. It sits just 0.4% behind the AMD Radeon R5 M240, which scores 6975, and 1.1% behind the NVIDIA GeForce GTX 680M, which hits 7023. Against the NVIDIA T600, the deficit is 1.3%, with that card scoring 7035. The only rival it clearly beats in its immediate cluster is the AMD FirePro M5100, which scores 6830 and is 1.7% slower. This places the GTX 675M in a tight pack of mid-range mobile GPUs, where a 2% swing either way can change the ranking.

The Quadro K620’s average benchmark score across all recorded tests is 6282, which is lower than its individual OpenCL result because the database also includes a Vulkan score of 5870. That Vulkan result drags the average down. The GTX 675M, by contrast, has only a single recorded OpenCL benchmark, so its average equals its peak score. The Quadro’s nearest rivals include the NVIDIA GeForce RTX 5070 Ti SUPER and the NVIDIA GeForce RTX 4070 Ti SUPER AD102, both scoring 6270, placing the K620 0.2% ahead of each. It also trails the AMD Radeon R7 M350 by 0.7% (6327) and the AMD Radeon Pro WX 4100 by 0.8% (6330). These are all very close scores, indicating that the K620 is firmly in the entry-level professional segment.

The head-to-head delta of 3.8% in favor of the GTX 675M is the only recorded win for either card. The Quadro K620 does not win any direct comparison in the database. However, the existence of the Vulkan benchmark for the Quadro, which the GTX 675M lacks entirely, means the K620 offers broader API coverage in the recorded data. The GTX 675M’s OpenCL score of 6946 is respectable for its generation, but the Quadro’s 6693 is not far behind, showing that the professional card can hold its own in compute despite having a much lower power envelope.

FAQ

Q: Which GPU has the higher OpenCL benchmark score?

A: The NVIDIA GeForce GTX 675M scores 6946 in Geekbench OpenCL, while the NVIDIA Quadro K620 scores 6693. The GTX 675M leads by 3.8%.

Q: How does the Quadro K620 compare to the GeForce GTX 675M in terms of average benchmark score?

A: The Quadro K620 has an average benchmark score of 6282, which includes both its OpenCL and Vulkan results. The GTX 675M has an average of 6946, based solely on its OpenCL test.

Q: What is the transistor density difference between the two cards?

A: The GTX 675M uses a 40 nm process and has a transistor density of 5.9 million transistors per square millimeter. The Quadro K620 uses a 28 nm process and achieves 12.6 million transistors per square millimeter, more than double the density.

Q: Which card has higher memory bandwidth?

A: The GTX 675M has a 256-bit memory bus with GDDR5 memory, delivering 96.00 GB/s. The Quadro K620 uses a 128-bit bus with DDR3 memory, providing 28.80 GB/s. The GTX 675M offers more than three times the bandwidth.

Q: Does the Quadro K620 support Vulkan?

A: Yes, the Quadro K620 records a Geekbench Vulkan score of 5870 and lists Vulkan 1.4 API support. The GTX 675M does not have any Vulkan benchmark recorded and lists no Vulkan support in its API table.

Q: Which card has a higher pixel fill rate?

A: The Quadro K620 achieves 17.98 GPixel/s, while the GTX 675M reaches 9.92 GPixel/s. Despite the GTX 675M’s higher overall compute score, the Quadro excels in pixel throughput.

Where Each One Wins

The GeForce GTX 675M takes the compute crown in the only direct benchmark comparison. Its 6946 OpenCL score is 3.8% higher than the Quadro K620’s 6693, indicating that for raw general-purpose compute tasks like OpenCL acceleration, the older gaming chip retains an edge. The GTX 675M also dominates in memory throughput: its 96.00 GB/s bandwidth is over three times the Quadro’s 28.80 GB/s, which matters for workloads that are bandwidth-sensitive, such as large data transfers or high-resolution texture streaming. With 64 texture mapping units and 32 render output units, the GTX 675M also offers more parallel hardware for texture-heavy operations, backed by a texture fill rate of 39.68 GTexel/s compared to the Quadro’s 26.98 GTexel/s.

The Quadro K620 wins in areas that are not captured by the single OpenCL benchmark. Its pixel rate of 17.98 GPixel/s is nearly double the GTX 675M’s 9.92 GPixel/s, making it the better choice for fragment-bound rendering tasks. The K620 also brings Vulkan support, with a recorded score of 5870, which opens up modern graphics APIs that the GTX 675M cannot access. For professional workflows that rely on certified drivers and stable viewport performance, the Quadro’s design goals align with precision over peak speed. Its 45 W power draw is less than half the GTX 675M’s 100 W, which means it fits in compact single-slot workstations and draws far less heat, a critical factor for dense multi-GPU setups or small form factor machines.

In terms of benchmark percentile standing, the GTX 675M places in the 39th percentile of all GPUs, while the Quadro K620 sits at the 36th percentile. The GTX 675M is therefore marginally better positioned globally, but both cards are firmly in the lower half of the performance distribution. The GTX 675M is the pick for users who need maximum compute throughput and memory bandwidth in a legacy mobile platform. The Quadro K620 is the pick for users who prioritize pixel throughput, modern API compatibility, and a drastically lower power footprint.

Specification Differences

The two cards diverge on nearly every measurable specification except memory capacity and shading unit count. Both have 2 GB of memory and 384 shading units, but the similarities end there. The GTX 675M uses GDDR5 memory with a 256-bit bus, while the Quadro K620 uses DDR3 with a 128-bit bus. This yields a bandwidth of 96.00 GB/s for the GTX 675M versus 28.80 GB/s for the Quadro, a 3.3x difference. The memory clock also differs: the GTX 675M runs at 750 MHz with 3 Gbps effective data rate, while the Quadro runs at 900 MHz with only 1800 Mbps effective.

Texture mapping units favor the GTX 675M at 64 versus 24 for the Quadro, and render output units favor it as well, 32 versus 16. The GTX 675M’s texture rate is 39.68 GTexel/s compared to 26.98 GTexel/s for the Quadro. However, the Quadro flips the script on pixel rate: 17.98 GPixel/s versus 9.92 GPixel/s. Floating point performance (FP32) is higher on the GTX 675M at 952.3 GFLOPS, while the Quadro delivers 863.2 GFLOPS.

Power consumption is a major separator. The GTX 675M draws 100 W and uses an MXM module form factor with no power connectors, designed for laptops. The Quadro K620 draws only 45 W, is a single-slot card with a 160 mm length (6.3 inches) and 69 mm height (2.7 inches), and suggests a 200 W power supply. The GTX 675M has a portable-device-dependent display output, while the Quadro offers 1x DVI and 1x DisplayPort 1.2. The bus interface also differs: MXM-B (3.0) for the GTX 675M versus PCIe 2.0 x16 for the Quadro. The Quadro has explicit base and boost clocks of 1058 MHz and 1124 MHz, while the GTX 675M lists no base or boost clock in the database.

Architecture Differences

The architectural gap is generational. The GeForce GTX 675M is built on NVIDIA’s Fermi 2.0 architecture, using the GF114 chip fabricated on a 40 nm process at TSMC. The Quadro K620 uses the Maxwell architecture with the GM107 chip on a 28 nm process, also at TSMC. The die sizes reflect the process shrink: the Fermi chip measures 332 mm², while the Maxwell chip is 148 mm². Transistor counts are surprisingly close, with the GF114 packing 1,950 million transistors and the GM107 carrying 1,870 million, but the density tells the story: 5.9 million transistors per square millimeter for Fermi versus 12.6 million for Maxwell.

The generation field in the database lists the GTX 675M as part of the GeForce 600M family, with a predecessor in GeForce 500M and a successor in GeForce 700M. The Quadro K620 is listed under the Quadro Kepler (Kx200) generation, despite using Maxwell silicon, with a predecessor in Quadro Fermi and a successor in Quadro Maxwell. The release dates are over two years apart: the GTX 675M launched on March 21, 2012, while the Quadro K620 launched on July 21, 2014.

API support shows the newer architecture’s advantages. Both cards list DirectX 12 (11_0) and OpenGL 4.6, but the Quadro adds Vulkan 1.4 support, while the GTX 675M lists no Vulkan capability. The Maxwell architecture also enables the Quadro’s much higher pixel rate despite fewer ROPs, a sign of architectural efficiency improvements. The Fermi chip’s higher FP32 output (952.3 GFLOPS) and texture throughput come from its wider 256-bit memory interface and larger number of TMUs, but it pays for that with a 100 W power draw versus 45 W. The GTX 675M’s production status is end-of-life, as is the Quadro’s, but the Maxwell part remains the more modern design in terms of process technology and feature set.