Intel Arc B770 vs Intel Arc Pro B370 Comparison

Head-to-Head Benchmarks

The recorded data contains no direct head-to-head benchmark results between the Intel Arc B770 and the Intel Arc Pro B370. Both entries show empty benchmark arrays, zero win counts for either side, and average benchmark scores of zero. The percentile fields place both at the 50th percentile against all GPUs in the database, though this figure is identical and likely reflects the absence of measured performance data rather than actual equivalence.

What the data does provide is a set of theoretical compute and rendering figures derived from each chip's fixed hardware configuration. The Arc B770 delivers 19.66 TFLOPS of FP32 throughput, while the Arc Pro B370 delivers 6.144 TFLOPS. This translates to a difference of roughly 13.5 TFLOPS, meaning the B770 offers approximately 3.2 times the raw single-precision compute of the B370. In FP16 with a 2:1 ratio, the B770 reaches 39.32 TFLOPS versus 12.29 TFLOPS for the B370, preserving the same 3.2x gap.

Pixel throughput shows an even wider separation. The B770 records 307.2 GPixel/s against 48.00 GPixel/s for the B370, a multiple of 6.4. Texture rate tells a similar story: 614.4 GTexel/s versus 96.00 GTexel/s, again a 6.4x difference. These ratios indicate that the B770's advantage grows in fill-rate-bound workloads, which depend heavily on ROP and TMU counts rather than raw shader throughput alone.

The shading unit counts support these figures. The B770 uses 4096 shading units, 256 texture mapping units, and 128 ROPs. The B370 uses 1280 shading units, 40 TMUs, and 20 ROPs. The B770 therefore has 3.2x the shading units, 6.4x the TMUs, and 6.4x the ROPs. The parallel structure of these ratios is exact, suggesting the B370's render backend is scaled down proportionally relative to its shader array.

Ray tracing hardware follows a different pattern. The B770 has 32 RT cores, while the B370 has 10 RT cores, a 3.2x difference. This matches the shading unit ratio, so ray tracing resource scaling is consistent with compute scaling rather than fill-rate scaling. For workloads that stress the RT pipeline, the B770 should maintain its 3.2x compute-level advantage, assuming identical per-core efficiency.

Memory configuration diverges completely. The B770 uses 16 GB of GDDR6 on a 256-bit bus, yielding 512.0 GB/s of bandwidth. The B370 uses system shared memory, with bandwidth described as system dependent. No fixed bandwidth figure exists for the B370 in the database, so a direct memory throughput comparison is impossible. The B770's dedicated 512.0 GB/s allocation is a fixed resource, while the B370's bandwidth varies with the host platform's memory subsystem.

Clock behavior differs in an interesting way. Both chips share a 2400 MHz boost clock. The B770's base clock sits at 2100 MHz, while the B370's base clock is only 300 MHz. That 300 MHz figure is unusually low for a discrete graphics part, but the B370 is classified as an IGP, an integrated graphics processor. Its base clock likely reflects aggressive power management for mobile or low-power integration, with the boost clock representing the peak under load. The B770's base clock is 7x higher, implying sustained operation at near-boost frequencies is expected for the discrete card.

Where Each One Wins

The B770 wins decisively in every quantified compute and rendering category available in the database. Its 19.66 TFLOPS FP32 output, 307.2 GPixel/s pixel rate, 614.4 GTexel/s texture rate, 512.0 GB/s memory bandwidth, and 32 RT cores position it as the stronger part for any workload that can utilize dedicated graphics resources. The 225 W TDP and dual-slot form factor indicate a card designed for sustained high-throughput operation, with a 1x 6-pin plus 1x 8-pin power connector arrangement and a 550 W suggested PSU.

The B370 wins in power efficiency and integration. Its 25 W TDP is 200 W lower than the B770's 225 W TDP, a 9x reduction in power draw. The B370 uses no power connectors, occupies no expansion slot (classified as IGP), and relies on system shared memory. This makes it suitable for compact or low-power platforms where a dedicated card cannot be installed. The B370's 3 nm process node, built by Intel, contrasts with the B770's 5 nm TSMC node, suggesting the B370 uses a more advanced manufacturing process despite its smaller hardware configuration.

The B370 also carries the newer architecture generation. It uses Xe3-LPG on the Panther Lake chip, while the B770 uses Xe2-HPG on the BMG-G31 chip. The B370's release date is 2026-01-26, roughly one month after the B770's 2025-12-31 release. The B370's production status is listed as Active, while the B770's production status is null, which may indicate the B770 is not currently in active production according to the database.

For use cases, the B770 is the clear choice for rendering, gaming, or compute tasks where dedicated VRAM and high bandwidth matter. The B370's system shared memory means performance depends entirely on the host system's memory speed and capacity, which can vary widely. The B770's fixed 512.0 GB/s bandwidth removes that variable.

Architecture Differences

The two GPUs come from different architectural families within Intel's lineup. The B770 uses Xe2-HPG, the high-performance graphics architecture from the Battlemage generation, specifically the Arc 7 tier. Its chip is designated BMG-G31. The B370 uses Xe3-LPG, a low-power graphics architecture from the Panther Lake generation, classified under Arc Graphics-WM. The B370's chip is named Panther Lake, which is notable because Panther Lake is typically associated with Intel's processor platforms, and the B370's IGP classification confirms it is an integrated solution.

Process technology differs significantly. The B770 is fabricated on a 5 nm process at TSMC. The B370 is fabricated on a 3 nm process at Intel. The B370's process node is two steps smaller, using Intel's own foundry rather than TSMC. The B770's die size is 368 mm², while the B370's die size is unknown in the database. Transistor counts are unknown for both, preventing density comparisons.

The B770's memory architecture is a traditional discrete design: 16 GB of GDDR6 over a 256-bit bus. The B370's memory is entirely system shared, meaning it borrows from the host's main memory. This is a fundamental architectural split. Discrete memory offers predictable bandwidth and capacity, while shared memory scales with the platform and is subject to contention with CPU workloads.

Both support the same API set: DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. No difference exists in API feature support between the two. Display outputs differ in presentation: the B770 lists 1x HDMI 2.1a and 3x DisplayPort 2.1, while the B370 lists "Portable Device Dependent," indicating its outputs vary by the host device.

The B370's predecessor is HD Graphics-WM, a prior integrated graphics line. The B770's predecessor is Alchemist, Intel's first discrete Arc generation. This lineage places the B770 as a direct evolution of Intel's discrete GPU efforts, while the B370 continues the integrated graphics lineage with a newer architecture.

Power delivery reflects the architectural split. The B770 requires external power via 1x 6-pin and 1x 8-pin connectors, with a 550 W suggested PSU. The B370 requires no connectors and has no suggested PSU, drawing from the host platform's power delivery. The B770 is dual-slot, the B370 is IGP, meaning it occupies no slot.

FAQ

Q: What is the raw FP32 compute difference between the two GPUs?

A: The Intel Arc B770 delivers 19.66 TFLOPS, while the Intel Arc Pro B370 delivers 6.144 TFLOPS. The B770 has approximately 3.2 times the FP32 throughput.

Q: Do both GPUs boost to the same clock speed?

A: Yes, both list a 2400 MHz boost clock. The base clocks differ: the B770 runs at 2100 MHz, while the B370 runs at 300 MHz.

Q: How much memory does each GPU have?

A: The B770 has 16 GB of GDDR6 on a 256-bit bus with 512.0 GB/s bandwidth. The B370 uses system shared memory, with bandwidth described as system dependent.

Q: Which GPU uses a smaller manufacturing process?

A: The B370 uses a 3 nm process at Intel. The B770 uses a 5 nm process at TSMC.

Q: Are the ray tracing resources scaled differently than the fill-rate resources?

A: Yes. The B770 has 32 RT cores versus 10 for the B370, a 3.2x ratio matching the shader unit ratio. The B770's pixel rate (307.2 GPixel/s versus 48.00 GPixel/s) and texture rate (614.4 GTexel/s versus 96.00 GTexel/s) show a 6.4x ratio, which is larger.

Q: What is the power draw difference?

A: The B770 has a 225 W TDP. The B370 has a 25 W TDP. The B370 draws 200 W less.

Specification Differences

| Specification | Intel Arc B770 | Intel Arc Pro B370 |

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

| Chip | BMG-G31 | Panther Lake |

| Architecture | Xe2-HPG | Xe3-LPG |

| Generation | Battlemage (Arc 7) | Arc Graphics-WM (Panther Lake) |

| Process Node | 5 nm (TSMC) | 3 nm (Intel) |

| Die Size | 368 mm² | Unknown |

| Base Clock | 2100 MHz | 300 MHz |

| Boost Clock | 2400 MHz | 2400 MHz |

| Memory | 16 GB GDDR6 | System Shared |

| Memory Bus | 256 bit | System Shared |

| Memory Bandwidth | 512.0 GB/s | System Dependent |

| Shading Units | 4096 | 1280 |

| TMUs | 256 | 40 |

| ROPs | 128 | 20 |

| RT Cores | 32 | 10 |

| Pixel Rate | 307.2 GPixel/s | 48.00 GPixel/s |

| Texture Rate | 614.4 GTexel/s | 96.00 GTexel/s |

| FP32 | 19.66 TFLOPS | 6.144 TFLOPS |

| FP16 | 39.32 TFLOPS | 12.29 TFLOPS |

| TDP | 225 W | 25 W |

| Slot Width | Dual-slot | IGP |

| Power Connectors | 1x 6-pin + 1x 8-pin | None |

| Suggested PSU | 550 W | None |

| Bus Interface | PCIe 4.0 x16 | IGP |

| Display Outputs | 1x HDMI 2.1a, 3x DisplayPort 2.1 | Portable Device Dependent |

| Production Status | Not listed | Active |

| Release Date | 2025-12-31 | 2026-01-26 |

| Predecessor | Alchemist | HD Graphics-WM |

The Verdict

The Intel Arc B770 is the dominant part in every measured performance category. Its 19.66 TFLOPS FP32, 307.2 GPixel/s, 614.4 GTexel/s, 32 RT cores, and 512.0 GB/s memory bandwidth make it the appropriate choice for workloads requiring dedicated graphics compute, high-resolution rendering, or substantial fixed memory capacity. The B770's dual-slot design, 225 W TDP, and 550 W suggested PSU indicate it targets full-size desktop systems with adequate power delivery.

The Intel Arc Pro B370 is the appropriate choice for platforms where power draw and physical size are constrained. Its 25 W TDP, IGP form factor, and lack of power connectors allow it to function in devices that cannot host a discrete card. The 3 nm Intel process node suggests a modern manufacturing approach, and the Xe3-LPG architecture is newer than the B770's Xe2-HPG. However, its 6.144 TFLOPS FP32 and system shared memory mean its performance is inherently limited and platform-dependent.

The data does not include any measured benchmark scores, so real-world application performance cannot be verified from the database. What the specifications show is a clear tier separation: the B770 is a high-throughput discrete GPU, and the B370 is a low-power integrated GPU. Any workload that fits within the B370's 6.144 TFLOPS and shared memory constraints may run acceptably, but the B770 offers 3.2x compute, 6.4x fill rate, and dedicated memory. The choice between them depends on whether the host system can physically and electrically support a discrete card. If it can, the B770's specification sheet is superior in every quantified category. If it cannot, the B370 is the only one of the two that fits.