NVIDIA GeForce GTX 970 vs NVIDIA Quadro K620 Comparison

Head-to-Head Benchmarks

The direct comparison between these two cards is limited to two recorded tests, and the results are decisively one-sided. In Geekbench OpenCL, the NVIDIA GeForce GTX 970 scores 29,982 points, while the NVIDIA Quadro K620 manages 6,693. That is a 348% advantage for the GTX 970, a margin that dwarfs any architectural nuance. The gap is not a narrow lead; it is a full performance class separation.

The second shared test, Geekbench Vulkan, tells the same story but with a slightly narrower spread. The GTX 970 posts 20,005 points against the K620’s 5,870, a 240.8% difference. While the percentage lead shrinks compared to OpenCL, the absolute gap remains enormous. The data shows the GTX 970 winning both head-to-head tests, giving it a clean 2-0 record in this matchup.

What stands out is not just the raw delta, but what it implies about workload scaling. OpenCL often stresses memory bandwidth and raw compute throughput, while Vulkan leans on driver overhead and shader occupancy. The GTX 970 wins big in both, suggesting its advantage is not tied to one specific API quirk but to a fundamentally higher execution capacity. The K620’s smaller shader array and narrower memory path simply cannot keep pace when the same instruction stream is fed to both cards.

It is also worth noting where each card sits relative to its own peer group. The GTX 970’s average benchmark score of 7,157 places it at the 39th percentile of all GPUs in the database. Its nearest rivals, such as the Intel Iris Pro Graphics P580 (7,170, only 0.2% behind) and the AMD Radeon Vega 8 Mobile (7,203, 0.6% ahead), are separated by fractions of a percent. That clustering indicates the GTX 970 is positioned in a dense, competitive midrange band. The Quadro K620, by contrast, has an average score of 6,282, sitting at the 36th percentile. Its nearest listed rival, the AMD Radeon Pro WX 4100, is a mere 0.8% faster. The K620 is not alone at the bottom, but it is clearly not in the same league as the GTX 970.

Architecture Differences

The two cards share a common designer and a common manufacturing process, but diverge sharply in almost every other architectural dimension. Both are built on TSMC’s 28 nm node, yet the chips underneath are very different creatures. The GTX 970 uses the GM204 die, a Maxwell 2.0 design, while the K620 relies on the GM107, which is first-generation Maxwell. This generation gap within the same family explains part of the performance disparity.

Transistor counts reinforce the divide. The GM204 packs 5,200 million transistors onto a 398 mm² die, yielding a transistor density of 13.1 million per square millimeter. The GM107, in contrast, holds 1,870 million transistors on a 148 mm² die, a density of 12.6 million per square millimeter. The GTX 970’s die is nearly three times larger in area and holds close to three times the transistors. That physical headroom translates directly into more execution resources.

The shading unit counts are where the gap becomes most visible. The GTX 970 carries 1,664 shading units, 104 texture mapping units, and 56 ROPs. The K620 has 384 shading units, 24 TMUs, and just 16 ROPs. These are not incremental differences; the GTX 970 has over four times the shader count and over four times the TMU count. Pixel throughput follows suit: the GTX 970 renders at 65.97 GPixel/s, while the K620 manages 17.98 GPixel/s. Texture rate is similarly lopsided at 122.5 GTexel/s versus 26.98 GTexel/s. Floating-point performance, measured as FP32, is 3.920 TFLOPS for the GTX 970 and 863.2 GFLOPS for the K620. Every compute metric points the same way.

Memory architecture adds another layer of separation. The GTX 970 uses 4 GB of GDDR5 on a 256-bit bus, delivering 224.4 GB/s of bandwidth. The K620 uses 2 GB of DDR3 on a 128-bit bus, capping out at 28.80 GB/s. That is a roughly eight-fold difference in memory bandwidth, which heavily influences any bandwidth-sensitive workload. The GTX 970’s memory clock is listed as 1753 MHz (7 Gbps effective), while the K620’s is 900 MHz (1800 Mbps effective). Even the bus interface differs: the GTX 970 runs on PCIe 3.0 x16, whereas the K620 is limited to PCIe 2.0 x16.

Power and physical design also diverge. The GTX 970 has a TDP of 148 W, requires a dual-slot cooler, and draws power through two 6-pin connectors. The K620 is a 45 W card, fits in a single slot, and needs no external power connectors. The suggested PSU rating reflects this: 300 W for the GTX 970, 200 W for the K620. Dimensions follow the power envelope, with the GTX 970 at 267 mm in length and the K620 at 160 mm. The GTX 970 also supports DirectX 12 with feature level 12_1, while the K620 only reaches DirectX 12 (11_0). Both support OpenGL 4.6 and Vulkan 1.4.

FAQ

Q: Which card has the higher average benchmark score?

A: The NVIDIA GeForce GTX 970 records an average benchmark score of 7,157, compared to the NVIDIA Quadro K620’s 6,282. The GTX 970 also holds a higher percentile rank at 39 versus 36.

Q: How much faster is the GTX 970 in Geekbench Vulkan?

A: In the Geekbench Vulkan test, the GTX 970 scores 20,005 against the K620’s 5,870, a 240.8% advantage.

Q: Is the Quadro K620 more power-efficient?

A: Yes, the K620 has a TDP of 45 W and requires no power connectors, while the GTX 970 has a TDP of 148 W and needs two 6-pin connectors. The suggested PSU is also lower for the K620 at 200 W versus 300 W.

Q: Do both cards support the same DirectX version?

A: No. The GTX 970 supports DirectX 12 (12_1), while the K620 supports DirectX 12 (11_0). Both support OpenGL 4.6 and Vulkan 1.4.

Q: What is the memory bandwidth difference?

A: The GTX 970 provides 224.4 GB/s over a 256-bit GDDR5 bus, while the K620 provides 28.80 GB/s over a 128-bit DDR3 bus. The GTX 970’s bandwidth is roughly 7.8 times higher.

Q: Which card has more shading units?

A: The GTX 970 has 1,664 shading units, compared to the K620’s 384. The GTX 970 also has 104 TMUs and 56 ROPs, versus 24 TMUs and 16 ROPs on the K620.

Specification Differences

The two cards differ in nearly every measurable specification. The GTX 970 uses the GM204 chip on Maxwell 2.0, while the K620 uses the GM107 on first-generation Maxwell. Both are 28 nm TSMC parts, but the GTX 970 has 5,200 million transistors on a 398 mm² die, versus 1,870 million on 148 mm² for the K620. Transistor density is slightly higher on the GTX 970 at 13.1M / mm², compared to 12.6M / mm².

Clock speeds are close at the base level: 1050 MHz for the GTX 970 and 1058 MHz for the K620. Boost clocks are also similar, at 1178 MHz and 1124 MHz respectively. Memory clocks, however, are far apart: 1753 MHz (7 Gbps effective) for the GTX 970, versus 900 MHz (1800 Mbps effective) for the K620. Memory size, type, bus width, and bandwidth all favor the GTX 970, as detailed earlier.

Shading units, TMUs, and ROPs are all substantially higher on the GTX 970. Pixel rate is 65.97 GPixel/s versus 17.98 GPixel/s, and texture rate is 122.5 GTexel/s versus 26.98 GTexel/s. FP32 compute is 3.920 TFLOPS against 863.2 GFLOPS. TDP is 148 W versus 45 W, and slot width is dual-slot versus single-slot. The GTX 970 uses two 6-pin power connectors, while the K620 uses none. PCIe interface differs as well: PCIe 3.0 x16 on the GTX 970, PCIe 2.0 x16 on the K620. Display outputs are also different, with the GTX 970 offering 1x DVI, 1x HDMI 2.0, and 3x DisplayPort 1.2, while the K620 offers 1x DVI and 1x DisplayPort 1.2. Dimensions differ, with the GTX 970 at 267 mm length, 111 mm height, and 40 mm width, versus the K620 at 160 mm length and 69 mm height.

Where Each One Wins

The GTX 970 wins every direct benchmark comparison in the database. It is the clear choice for any workload that depends on raw compute throughput, memory bandwidth, or shader count. The 348% lead in OpenCL and 240.8% lead in Vulkan show that the GTX 970 is not just faster in one narrow test; it dominates across different API paradigms. For gaming, rendering, or any general-purpose GPU compute, the GTX 970 is the only sensible pick from this pair. Its 4 GB GDDR5 memory and 224.4 GB/s bandwidth also make it suitable for higher-resolution textures and larger data sets.

The Quadro K620, on the other hand, wins in areas that are not captured by benchmark scores. Its 45 W TDP means it can be installed in systems with smaller power supplies and limited cooling. It is a single-slot card requiring no external power connectors, which makes it viable for compact workstations or dense chassis where space is at a premium. Its shorter length (160 mm versus 267 mm) adds to its flexibility in small form factor builds. The K620 also carries the Quadro branding, which may imply certification for certain professional software stacks, though the benchmark data does not quantify any such advantage.

The K620’s nearest rivals in the database, such as the AMD Radeon Pro WX 4100 (only 0.8% faster), suggest it sits in a low-end professional tier. The GTX 970’s rivals, like the GeForce GTX 750 (0.9% slower), place it in a more capable midrange segment. If the task is purely about accelerating graphics or compute, the GTX 970 wins every scenario. If the task is about minimal power draw and physical footprint, the K620 is the only card that fits.

The Verdict

The data is unambiguous. The NVIDIA GeForce GTX 970 outperforms the NVIDIA Quadro K620 in every recorded benchmark by margins ranging from 240.8% to 348%. Its average benchmark score of 7,157 is 13.9% higher than the K620’s 6,282, and its percentile rank (39 vs 36) reflects a better standing across the entire GPU landscape. For any user whose priority is raw performance, the GTX 970 is the only choice.

The Quadro K620 is not without merit, but its advantages are purely physical. It consumes 103 W less power, fits in a single slot, requires no external power connectors, and is significantly shorter. These traits make it suitable for low-power or space-constrained systems. However, the benchmark results show no scenario where the K620 wins on speed. Even in the Vulkan test, where the gap narrows, the K620 still trails by 240.8%.

For a buyer who needs maximum compute performance and has the power budget and chassis space to accommodate a 148 W dual-slot card, the GTX 970 is the definitive pick. For a buyer who values minimal power draw and a compact footprint above all else, and who is willing to accept a vastly lower performance ceiling, the K620 can serve. The recorded data offers no third option: the GTX 970 is the performance winner, and the K620 is the efficiency-and-form-factor winner. There is no benchmark-based argument for choosing the K620 on speed.