Intel Arc A310 vs NVIDIA Quadro K620 Comparison

The Intel Arc A310 and NVIDIA Quadro K620 occupy the same low-power, single-slot corner of the database, but the recorded data shows they are not close competitors. Both cards are end-of-life, both slot into a single expansion slot, both draw power without auxiliary connectors, and both carry a suggested PSU rating of 200 W. Beyond those shared physical traits, the A310 leads in every measured category, sometimes by staggering margins. The Quadro K620's case rests on its pedigree as a workstation card from NVIDIA's Quadro line, but the benchmark record does not support choosing it over Intel's newer entry-level design.

The Verdict

The data points to the Arc A310 for essentially every workload where the two overlap. It wins both head-to-head benchmarks in the database, Geekbench OpenCL and Geekbench Vulkan, with margins of 357.3 percent and 393.4 percent respectively. Its average benchmark score of 7550 sits above the K620's 6282, and its percentile placement against all GPUs in the database is 40 versus 36 for the K620.

The K620 has no recorded win in any test. It is the older platform by a wide margin, launching in July 2014 versus October 2022 for the A310, and its architecture trails in every measurable specification the database tracks. Someone choosing between the two for compute, graphics, or display output should pick the A310 according to the recorded numbers. The only scenario that favors the K620 is one requiring its specific legacy output configuration: it offers one DVI and one DisplayPort 1.2 connector, while the A310 provides four mini-DisplayPort 2.0 outputs. A system still driving a DVI-only monitor without an adapter is the one niche the data can justify.

Architecture Differences

The generational gap between these cards is the dominant story. The Arc A310 is built on Intel's Xe-HPG architecture, the DG2-128 chip, fabricated on TSMC's 6 nm process. The Quadro K620 uses NVIDIA's Maxwell architecture, the GM107 chip, on a 28 nm TSMC process. That process advantage translates directly into density: the A310 packs 7,200 million transistors into a 157 mm² die, for 45.9 million transistors per square millimeter. The K620 manages 1,870 million transistors across a similar-sized 148 mm² die, yielding only 12.6 million per square millimeter.

The shader counts reflect that gap. The A310 has 768 shading units, 32 texture mapping units, 16 render output units, and 6 ray tracing cores. The K620 has 384 shading units, 24 TMUs, and 16 ROPs, with no ray tracing hardware recorded. The A310 doubles the K620's shading resources and adds a ray tracing capability the older card simply lacks.

Clock behavior differs in kind, not just degree. The A310 runs a flat 1750 MHz at both base and boost, indicating a single sustained operating frequency. The K620 clocks between 1058 MHz base and 1124 MHz boost. Memory is another stark divide: the A310 pairs 4 GB of GDDR6 on a 64-bit bus, running at 15.5 Gbps effective for 124.0 GB/s of bandwidth. The K620 carries 2 GB of DDR3 on a 128-bit bus at 1800 Mbps effective, delivering just 28.80 GB/s. Despite the wider bus, the K620's memory subsystem provides less than a quarter of the A310's bandwidth.

Theoretical throughput figures confirm the hierarchy. The A310 produces 2.688 TFLOPS of FP32 compute and 5.376 TFLOPS of FP16 at a 2:1 ratio, with a pixel rate of 28.00 GPixel/s and a texture rate of 56.00 GTexel/s. The K620 delivers 863.2 GFLOPS FP32, 17.98 GPixel/s, and 26.98 GTexel/s. No FP16 figure is recorded for the Maxwell card.

Platform interfaces also differ. The A310 uses PCIe 4.0 x8, while the K620 uses PCIe 2.0 x16. Software support is where the A310's DirectX 12 Ultimate (12_2) feature level stands out against the K620's DirectX 12 (11_0) support, meaning the Intel card exposes modern rendering features the K620 cannot address. Both report OpenGL 4.6 and Vulkan 1.4. Power draw is one category where the older card loses too: the K620 has a 45 W TDP against 30 W for the A310, so the newer card delivers far more performance while drawing less.

Where Each One Wins

The Arc A310 wins every performance category in the database. Its Geekbench OpenCL score of 30607 against 6693 makes it the clear choice for OpenCL compute workloads. Its Geekbench Vulkan score of 28964 against 5870 does the same for Vulkan-driven graphics and compute. Its broader benchmark set reinforces the picture: a Passmark G3D score of 5433, a GPU compute score of 2157, and a 2D graphics score of 625, alongside DirectX 9 through 12 Passmark results of 69, 31, 33, and 29 respectively. The K620 has no comparable recorded results in those tests.

Context from each card's nearest rivals clarifies the positioning. The A310's average score of 7550 lands within a percent of the AMD Radeon R7 250 (7557, a delta of -0.1 percent), the AMD Radeon Pro WX 3100 (7580, -0.4 percent), and the NVIDIA GeForce GTX 1650 (7472, 1 percent ahead), plus the AMD Radeon HD 8850M (7447, 1.4 percent ahead). That is entry-level discrete territory, but current-generation entry-level. The K620's average of 6282 places it 0.2 percent behind the NVIDIA GeForce RTX 5070 Ti SUPER and RTX 4070 Ti SUPER AD102 averages recorded in the database, and 0.7 to 0.8 percent behind the AMD Radeon R7 M350 and Radeon Pro WX 4100, a quirk of averaging across very different test coverage rather than a claim of parity in capability.

The K620's only wins are qualitative. It is the card with a DVI output, useful for legacy displays. Its 160 mm length and 69 mm height are recorded, making it a known quantity for tight enclosures. And its Quadro workstation lineage, sitting between the Quadro Fermi predecessor and the Quadro Maxwell successor in NVIDIA's lineup, carries certified-driver heritage that the raw database scores cannot capture. None of that translates into a single benchmark victory.

FAQ

Q: Is the Intel Arc A310 faster than the Quadro K620?

A: Yes, in every recorded test. The A310 wins Geekbench OpenCL by 357.3 percent and Geekbench Vulkan by 393.4 percent, and holds a higher average benchmark score, 7550 versus 6282.

Q: Which card has more memory bandwidth?

A: The A310, at 124.0 GB/s from 4 GB of GDDR6 on a 64-bit bus. The K620 delivers 28.80 GB/s from 2 GB of DDR3 on a 128-bit bus, so its wider bus does not offset its slower memory type.

Q: Do either of these cards support ray tracing?

A: The A310 does, with 6 ray tracing cores. The K620 has no ray tracing hardware recorded.

Q: Which card uses less power?

A: The A310, with a 30 W TDP versus 45 W for the K620. Neither requires a power connector, and both carry a suggested PSU rating of 200 W.

Q: Which DirectX feature level does each support?

A: The A310 supports DirectX 12 Ultimate at feature level 12_2. The K620 supports DirectX 12 at feature level 11_0, which excludes it from modern rendering features.

Q: Are these cards still in production?

A: No. Both are listed as end-of-life. The A310's successor line is Battlemage; the K620's is Quadro Maxwell.

Head-to-Head Benchmarks

Only two tests appear in both cards' records, and the A310 sweeps them.

Geekbench OpenCL is the compute-focused contest, and it is a rout: 30607 for the Arc A310 against 6693 for the Quadro K620. That 357.3 percent margin means the A310 delivers roughly four and a half times the OpenCL throughput of the K620, a gap consistent with the raw specification differences. The A310's 2.688 TFLOPS of FP32 compute is more than triple the K620's 863.2 GFLOPS, and its memory bandwidth advantage of 124.0 GB/s over 28.80 GB/s removes any bottleneck that might narrow the result. For OpenCL-accelerated workloads, media processing, scientific compute, or GPU-assisted encoding, the data shows no scenario where the K620 is competitive.

Geekbench Vulkan widens the gap further: 28964 to 5870, a 393.4 percent victory for the A310. The Vulkan result exercises the modern graphics API path both cards nominally support, both list Vulkan 1.4, but the outcome shows how different that support is in practice. The A310's 768 shading units at 1750 MHz, its 56.00 GTexel/s texture rate, and its 28.00 GPixel/s pixel rate give it roughly double the rasterization and texturing throughput of the K620, whose corresponding figures are 26.98 GTexel/s and 17.98 GPixel/s from 384 shading units at up to 1124 MHz.

The aggregate picture matches the test-by-test one. The A310's average benchmark score of 7550 places it in the 40th percentile of all GPUs in the database, ahead of the K620's 6282 average and 36th percentile. The A310 sits in a cluster with the Radeon R7 250, Radeon Pro WX 3100, GeForce GTX 1650, and Radeon HD 8850, all within roughly one percent of its average. The K620's nearest recorded rivals sit marginally above it in average score, with the RTX 5070 Ti SUPER and RTX 4070 Ti SUPER AD102 at 0.2 percent deltas and the R7 M350 and Pro WX 4100 at -0.7 and -0.8 percent.

The verdict from the data is unambiguous. The Arc A310's 6 nm Xe-HPG design, quadruple-class memory bandwidth, doubled shader count, ray tracing hardware, and modern DirectX 12 Ultimate support make it the stronger card in every measured dimension. The Quadro K620's remaining relevance is physical and legacy: DVI output, compact recorded dimensions, and workstation lineage. For anything measured by the database, the A310 is the pick.