Intel Arc A770 vs NVIDIA GB10 Comparison

Head-to-Head Benchmarks

The recorded data shows a decisive sweep for the NVIDIA GB10 in the two common benchmark workloads. In Geekbench OpenCL, the GB10 scores 120,137 against 109,175 for the Intel Arc A770, a 10% advantage. The gap widens significantly in Geekbench Vulkan, where the GB10 posts 114,648 versus the Arc A770's 94,284, a 21.6% lead.

These results place the GB10 in the 95th percentile of all GPUs in the database, while the Arc A770 sits in the 90th percentile. The GB10's average benchmark score of 117,393 puts it in a competitive bracket with professional and workstation cards: it trails the AMD Radeon PRO W7700 by 1.3%, edges out the NVIDIA RTX 4000 SFF Ada Generation by 0.3%, and leads the NVIDIA Tesla V100 SXM2 16 GB by 2.6% and the NVIDIA RTX A5500 Mobile by 3%. The Arc A770's average of 68,809 places it near the NVIDIA CMP 90HX, which it trails by 0.3%, and the AMD Radeon Instinct MI25, which it leads by 0.4%. It sits 1.5% behind the AMD Radeon Pro WX 8200 and 1.7% behind the NVIDIA Quadro P6000.

The OpenCL result is notable because both chips scale differently in compute-heavy workloads. The GB10's 29.71 TFLOPS FP32 throughput and 384 tensor cores deliver a substantial raw compute advantage. The Arc A770's 19.66 TFLOPS FP32 is lower, and its FP16 rate of 39.32 TFLOPS (2:1 ratio) does not translate into a win here. The Vulkan gap of 21.6% is the largest recorded delta between the two, and it reflects architectural efficiency rather than just peak numbers: the GB10 achieves this with a 140 W TDP against the Arc A770's 225 W TDP.

Pixel and texture rates tell a different story, though they are not part of the head-to-head benchmark set. The Arc A770 reaches 307.2 GPixel/s and 614.4 GTexel/s, while the GB10 manages 116.1 GPixel/s and 928.5 GTexel/s. The Arc wins decisively on pixel throughput, but the GB10's texture rate is higher by roughly 51%. These are specification-level differences; the benchmark results still favor the GB10 in both recorded tests.

The GB10 also holds a memory capacity and bandwidth profile that is unusual for a chip in its power class. Its 128 GB of LPDDR5X on a 256-bit bus delivers 273.2 GB/s, whereas the Arc A770's 16 GB of GDDR6 on the same 256-bit bus width achieves 512.0 GB/s. The Arc has nearly double the bandwidth, but the GB10's far larger capacity and its Blackwell architecture appear to compensate in the recorded workloads.

The Verdict

From the data, the NVIDIA GB10 is the stronger compute performer. It wins both head-to-head benchmarks, holds a 5-percentile advantage over the Arc A770, and delivers over 70% higher average benchmark scores. The GB10's nearest rivals in the database are all professional or data-center class cards, which frames its performance tier clearly.

The Arc A770 should be chosen when the workload favors its specific strengths: higher pixel fill rate (307.2 GPixel/s), greater memory bandwidth (512.0 GB/s), and a broader API support set including DirectX 12 Ultimate, OpenGL 4.6, and Vulkan 1.4. The GB10 lists its API support as N/A across DirectX, OpenGL, and Vulkan, which suggests its software ecosystem is not oriented toward traditional graphics APIs.

Buyers who need a current-generation compute device with active production status and a 95th-percentile standing should pick the GB10. It is an integrated graphics processor (IGP) with no power connectors, a 140 W TDP, and a suggested PSU of 300 W. Its successor is listed as Server Rubin, and its predecessor is Server Hopper. The Arc A770 is end-of-life, uses dual-slot cooling, requires one 6-pin and one 8-pin power connector, and suggests a 550 W PSU. Its successor is Battlemage.

The GB10 also carries a launch MSRP of 3,999 USD. The Arc A770 carries a launch MSRP of 329 USD.

FAQ

Q: Which GPU wins in Geekbench OpenCL?

A: The NVIDIA GB10 scores 120,137 versus 109,175 for the Intel Arc A770, a 10% advantage.

Q: How large is the Vulkan performance gap?

A: The GB10 scores 114,648 against the Arc A770's 94,284, a 21.6% lead for the GB10.

Q: What are the average benchmark scores for each GPU?

A: The GB10 has an average benchmark score of 117,393, while the Arc A770 averages 68,809.

Q: How do these GPUs rank against all other GPUs in the database?

A: The GB10 is in the 95th percentile, and the Arc A770 is in the 90th percentile.

Q: Which GPU has more memory bandwidth?

A: The Arc A770 has 512.0 GB/s of bandwidth from 16 GB of GDDR6, while the GB10 has 273.2 GB/s from 128 GB of LPDDR5X.

Q: What API support does each GPU offer?

A: The Arc A770 supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The GB10 lists N/A for DirectX, OpenGL, and Vulkan.

Specification Differences

The two chips differ in nearly every major specification category. The GB10 uses a 5 nm process node from TSMC, while the Arc A770 uses a 6 nm node, also from TSMC. The GB10's die size is 382 mm², the Arc's is 406 mm². Transistor counts are unknown for the GB10, while the Arc A770 has 21,700 million transistors at a density of 53.4M per mm².

Clock speeds differ at the base level: the GB10 runs at 1665 MHz base and 2418 MHz boost, the Arc A770 at 2100 MHz base and 2400 MHz boost. Memory clocks also differ: the GB10's LPDDR5X runs at 1067 MHz (8.5 Gbps effective), while the Arc's GDDR6 runs at 2000 MHz (16 Gbps effective).

Memory capacity is a major divider: 128 GB for the GB10 versus 16 GB for the Arc. Both use a 256-bit bus, but bandwidth favors the Arc at 512.0 GB/s versus 273.2 GB/s.

Compute unit counts diverge sharply. The GB10 has 6144 shading units, 384 TMUs, 48 ROPs, 48 RT cores, and 384 tensor cores. The Arc A770 has 4096 shading units, 256 TMUs, 128 ROPs, and 32 RT cores, with no tensor core count listed. Pixel rate favors the Arc at 307.2 GPixel/s versus 116.1 GPixel/s. Texture rate favors the GB10 at 928.5 GTexel/s versus 614.4 GTexel/s. FP32 throughput favors the GB10 at 29.71 TFLOPS versus 19.66 TFLOPS. FP16 throughput favors the Arc at 39.32 TFLOPS (2:1) versus 29.71 TFLOPS (1:1) for the GB10.

Power and physical design differ substantially. The GB10 is rated at 140 W TDP, is an IGP with no power connectors, and suggests a 300 W PSU. The Arc is 225 W, dual-slot, requires one 6-pin and one 8-pin connector, and suggests a 550 W PSU. The GB10 uses PCIe 5.0 x16; the Arc uses PCIe 4.0 x16. Display outputs are 1x HDMI for the GB10 versus 1x HDMI 2.1 and 3x DisplayPort 2.0 for the Arc.

Dimensions are only listed for the GB10: 150 mm length, 51 mm height, and 150 mm width. No dimensions are recorded for the Arc.

Release timing and status differ: the GB10 released on 2025-10-14 and is Active; the Arc released on 2022-10-11 and is End-of-life.

Architecture Differences

The NVIDIA GB10 is built on the Blackwell 2.0 architecture, part of the Server Blackwell (Bxx) generation, using the GB20B chip. The Intel Arc A770 uses the Xe-HPG architecture, part of the Alchemist (Arc 7) generation, with the DG2-512 chip. The GB10's predecessor is Server Hopper and its successor is Server Rubin. The Arc's predecessor is Xe Graphics and its successor is Battlemage.

The GB10 integrates tensor cores (384 of them), which the Arc A770 does not list. The GB10's FP16 compute runs at a 1:1 ratio with FP32, meaning 29.71 TFLOPS for both. The Arc's FP16 runs at a 2:1 ratio, yielding 39.32 TFLOPS. This indicates different compute priorities: the GB10 is designed for AI and data-center workloads, while the Arc's higher FP16 rate suggests a graphics-oriented design.

Memory architecture differs in type and capacity. The GB10 uses LPDDR5X with 128 GB capacity, a configuration suited to large in-memory datasets. The Arc uses GDDR6 with 16 GB, a conventional graphics memory setup. Both use a 256-bit bus, but the GDDR6 clocks higher and delivers nearly double the bandwidth.

Process technology is close but not identical: 5 nm for the GB10 versus 6 nm for the Arc, both from TSMC. The Arc's die is slightly larger at 406 mm² versus 382 mm², and it carries 21,700 million transistors. The GB10's transistor count is not recorded.

The GB10's API support is listed as N/A for DirectX, OpenGL, and Vulkan, which reflects a compute-first design rather than a graphics-card role. The Arc A770 fully supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4, making it the choice for traditional graphics API workloads where those interfaces are required.

The GB10's display output is a single HDMI port, while the Arc offers 1x HDMI 2.1 and 3x DisplayPort 2.0. The GB10's form factor is IGP with no power connectors, whereas the Arc is a dual-slot add-in card requiring external power. These architectural choices align with the performance data: the GB10 dominates compute benchmarks, while the Arc's feature set targets conventional GPU use cases.