Intel Arc B770 vs Intel Arc Pro B65 Comparison

Head-to-Head Benchmarks

The recorded data for the Intel Arc B770 and Intel Arc Pro B65 contains no direct benchmark scores, so the comparison relies entirely on the theoretical specifications and calculated rates provided in the database. The Arc B770 delivers a higher FP32 compute rating of 19.66 TFLOPS against the Arc Pro B65's 12.29 TFLOPS, a difference of roughly 60%. This places the B770 clearly ahead in raw shading throughput, which typically translates to faster fill-rate-bound workloads and higher frame rates in rasterized gaming scenarios.

The texture rate follows the same pattern. The Arc B770 reaches 614.4 GTexel/s while the Arc Pro B65 manages 384.0 GTexel/s, a 60% advantage for the B770. Pixel rate also favors the B770 at 307.2 GPixel/s versus 192.0 GPixel/s, again a 60% gap. These figures are direct outputs of the shading unit, TMU, and ROP counts listed in the database: the B770 has 4096 shading units, 256 TMUs, and 128 ROPs, while the B65 has 2560 shading units, 160 TMUs, and 80 ROPs.

Memory bandwidth is one area where the Arc Pro B65 strikes back. The B65's 32 GB GDDR6 memory runs at 2375 MHz (19 Gbps effective) across a 256-bit bus, yielding 608.0 GB/s of bandwidth. The Arc B770's 16 GB GDDR6 runs at 2000 MHz (16 Gbps effective) on the same 256-bit bus, yielding 512.0 GB/s. That gives the B65 roughly 19% more bandwidth, which can help in memory-heavy compute workloads and high-resolution texture streaming.

Clock speeds also differentiate the two. The Arc Pro B65 has a base clock of 2400 MHz and a boost clock of 2400 MHz, meaning it runs at a fixed frequency. The Arc B770 has a base clock of 2100 MHz and a boost clock of 2400 MHz, so its boost matches the B65's sustained clock, but its base is lower. The B65's higher base clock partially compensates for its lower core count in workloads that hold clocks near base, though the B770's larger core array still gives it the overall compute lead.

In terms of wins, the database records 0 wins for each card because no head-to-head benchmark entries exist. The only measurable victories come from specification-derived rates, where the B770 wins on FP32, texture, and pixel throughput, while the B65 wins on memory capacity and bandwidth.

Where Each One Wins

The Arc B770 is the stronger choice for scenarios that depend on shader throughput. Its 4096 shading units and 19.66 TFLOPS FP32 rating indicate a clear advantage in general-purpose GPU compute, particularly for tasks like real-time rendering, physics simulation, and gaming at high detail settings. The 60% lead in texture rate and pixel rate reinforces this: the B770 can push more textured polygons and fill more pixels per second, which matters for 1440p and 4K gaming with complex materials.

The Arc Pro B65, by contrast, wins in memory capacity and bandwidth. Its 32 GB VRAM is double the B770's 16 GB, which is significant for large dataset processing, AI inference batches, and professional 3D scenes that exceed 16 GB. The 608.0 GB/s bandwidth gives it a 19% edge over the B770's 512.0 GB/s, helping in bandwidth-bound operations such as ray tracing acceleration structures, large framebuffer reads, and multi-display output with high resolutions.

The B65 also has a higher base clock (2400 MHz vs 2100 MHz), which means sustained performance in workloads that do not boost well, such as long-running compute kernels or thermal-limited environments, may hold up better on the B65. However, the B770's boost clock matches at 2400 MHz, so peak single-threaded shader speed is equal.

For professional workloads, the B65's 32 GB capacity and fixed clock make it suited for rendering scenes that fit entirely in VRAM and for compute tasks that run for hours. The B770's higher FP32 throughput suits it for interactive applications and real-time workloads where latency and frame pacing matter more than capacity.

Architecture Differences

Both cards use the Xe2-HPG architecture from Intel's Battlemage generation, built on a 5 nm process at TSMC. The Arc B770 carries the BMG-G31 chip, while the Arc Pro B65 uses the BMG-G21 chip. The database lists the B65's transistor count as 19,600 million and die size as 272 mm², giving a transistor density of 72.1M per mm². The B770's transistor count is listed as unknown, but its die size is 368 mm², which is 35% larger than the B65's die.

The core configurations diverge significantly. The B770 has 4096 shading units, 256 TMUs, 128 ROPs, and 32 ray tracing cores. The B65 has 2560 shading units, 160 TMUs, 80 ROPs, and 20 ray tracing cores. The B770 therefore has 60% more shading units, TMUs, and ROPs, and 60% more RT cores. This consistent scaling suggests the B770 is a fully enabled or higher-tier implementation of the same architecture, while the B65 is a cut-down variant with fewer execution resources.

Memory subsystems differ despite the same 256-bit bus width. The B770 uses 16 GB GDDR6 at 2000 MHz (16 Gbps effective), while the B65 uses 32 GB GDDR6 at 2375 MHz (19 Gbps effective). The B65 not only has double the capacity but also faster memory clocks, explaining its higher bandwidth. The power delivery also differs: the B770 uses a 1x 6-pin plus 1x 8-pin connector setup with a 225 W TDP, while the B65 uses a single 8-pin connector and a 200 W TDP. Both recommend a 550 W PSU.

The bus interface differs as well. The B770 uses PCIe 4.0 x16, while the B65 uses PCIe 5.0 x16. This gives the B65 double the theoretical link bandwidth to the host system, which can benefit data transfer-heavy workloads that move large datasets between CPU and GPU.

Display outputs also differ. The B770 provides 1x HDMI 2.1a and 3x DisplayPort 2.1, while the B65 provides 4x DisplayPort 2.1 with no HDMI. This makes the B65 more oriented toward multi-monitor professional setups, while the B770 retains HDMI for consumer displays and VR headsets.

Both cards support DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4, so API feature sets are identical. The B65 is listed as Active in production status, while the B770 has no production status listed. The B770's predecessor is listed as Alchemist, while the B65 has no predecessor. The B770 is dated 2025-12-31 in the database, and the B65 is dated 2026-03-31, so the B65 is newer by the recorded release dates.

FAQ

Q: Which card has higher raw compute performance?

A: The Intel Arc B770. Its FP32 rating is 19.66 TFLOPS versus the Arc Pro B65's 12.29 TFLOPS, a 60% advantage.

Q: Which card has more memory and bandwidth?

A: The Intel Arc Pro B65. It has 32 GB GDDR6 with 608.0 GB/s bandwidth, while the Arc B770 has 16 GB GDDR6 with 512.0 GB/s bandwidth.

Q: Do both cards support the same graphics APIs?

A: Yes. Both support DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4.

Q: What is the difference in ray tracing core counts?

A: The Arc B770 has 32 RT cores, while the Arc Pro B65 has 20 RT cores. The B770 has 60% more RT cores.

Q: Which card has a faster memory clock?

A: The Arc Pro B65. Its memory runs at 2375 MHz (19 Gbps effective), while the Arc B770 runs at 2000 MHz (16 Gbps effective).

Q: What are the power connector requirements for each card?

A: The Arc B770 requires 1x 6-pin plus 1x 8-pin connectors with a 225 W TDP. The Arc Pro B65 requires a single 8-pin connector with a 200 W TDP. Both recommend a 550 W PSU.

Specification Differences

The following table lists only the fields where the two cards differ according to the database. Fields that match, such as architecture, process node, foundry, memory type, bus width, API support, slot width, and suggested PSU, are excluded.

| Specification | Intel Arc B770 | Intel Arc Pro B65 |

|---|---|---|

| Chip | BMG-G31 | BMG-G21 |

| Generation | Battlemage (Arc 7) | Battlemage (Pro Series) |

| Die Size | 368 mm² | 272 mm² |

| Transistors | unknown | 19,600 million |

| Transistor Density | null | 72.1M / mm² |

| Base Clock | 2100 MHz | 2400 MHz |

| Boost Clock | 2400 MHz | 2400 MHz |

| Memory Clock | 2000 MHz (16 Gbps effective) | 2375 MHz (19 Gbps effective) |

| Memory Size | 16 GB | 32 GB |

| Memory Bandwidth | 512.0 GB/s | 608.0 GB/s |

| Shading Units | 4096 | 2560 |

| TMUs | 256 | 160 |

| ROPs | 128 | 80 |

| RT Cores | 32 | 20 |

| Pixel Rate | 307.2 GPixel/s | 192.0 GPixel/s |

| Texture Rate | 614.4 GTexel/s | 384.0 GTexel/s |

| FP32 | 19.66 TFLOPS | 12.29 TFLOPS |

| FP16 | 39.32 TFLOPS (2:1) | 24.58 TFLOPS (2:1) |

| TDP | 225 W | 200 W |

| Power Connectors | 1x 6-pin + 1x 8-pin | 1x 8-pin |

| Bus Interface | PCIe 4.0 x16 | PCIe 5.0 x16 |

| Display Outputs | 1x HDMI 2.1a, 3x DisplayPort 2.1 | 4x DisplayPort 2.1 |

| Production Status | null | Active |

| Release Date | 2025-12-31 | 2026-03-31 |

| Predecessor | Alchemist | null |