Intel Arc Pro B60 Dual vs Intel Arc Pro B65 Comparison

Head-to-Head Benchmarks

The recorded database contains no head-to-head benchmark results for the Intel Arc Pro B60 Dual and the Intel Arc Pro B65. Both cards share the same chip, BMG-G21, and the same Xe2-HPG architecture, which means their raw compute ceilings are identical. The FP32 throughput for both is 12.29 TFLOPS, and the FP16 figure is 24.58 TFLOPS (2:1) for each. Pixel rate is 192.0 GPixel/s on both, and texture rate is 384.0 GTexel/s on both. Shading units, TMUs, ROPs, and RT cores are all equal: 2560, 160, 80, and 20 respectively.

With no benchmark deltas recorded, the primary differentiators fall entirely on memory configuration, power delivery, and interface. The B60 Dual runs a 24 GB GDDR6 frame buffer on a 192 bit bus, yielding 456.0 GB/s of bandwidth. The B65 carries 32 GB GDDR6 on a 256 bit bus, which raises bandwidth to 608.0 GB/s. That is a 33.3% bandwidth advantage for the B65, and a 33.3% capacity advantage as well. In any workload where memory throughput or capacity is the limiting factor, the B65 should pull ahead proportionally. Conversely, the B60 Dual has a base clock of 2000 MHz versus the B65's 2400 MHz base, though both boost to 2400 MHz. Since the boost clock matches, sustained peak compute under load should converge, but the B65 reaches that peak from a higher idle-to-load baseline.

The TDP figures diverge sharply: the B60 Dual is rated at 400 W, while the B65 is rated at 200 W. This implies the B60 Dual is designed for a higher sustained power envelope, possibly for longer render passes or continuous compute, whereas the B65 achieves the same peak clocks at half the rated power. The B60 Dual requires a 16-pin power connector and an 800 W suggested PSU, while the B65 uses a single 8-pin and a 550 W suggested PSU. The B60 Dual also runs a PCIe 5.0 x8 interface, while the B65 runs PCIe 5.0 x16. For bandwidth-hungry data transfers, the x16 link offers double the lanes.

FAQ

Q: Which card has more memory?

A: The Intel Arc Pro B65 has 32 GB of GDDR6, while the Intel Arc Pro B60 Dual has 24 GB of GDDR6. The B65 also uses a 256 bit memory bus versus the B60 Dual's 192 bit bus.

Q: Is there any difference in raw compute performance?

A: No. Both cards report identical FP32 throughput at 12.29 TFLOPS, identical FP16 at 24.58 TFLOPS (2:1), identical pixel rate at 192.0 GPixel/s, and identical texture rate at 384.0 GTexel/s. Shader count, TMUs, ROPs, and RT cores are all the same.

Q: What about memory bandwidth?

A: The B65 delivers 608.0 GB/s, which is 33.3% higher than the B60 Dual's 456.0 GB/s. This comes from the wider 256 bit bus combined with the same 2375 MHz memory clock (19 Gbps effective).

Q: How do their power requirements differ?

A: The B60 Dual is rated at 400 W TDP with a 1x 16-pin power connector and an 800 W suggested PSU. The B65 is rated at 200 W TDP with a 1x 8-pin connector and a 550 W suggested PSU. The B60 Dual consumes twice the rated power of the B65.

Q: What about the PCIe interface?

A: The B60 Dual uses PCIe 5.0 x8, while the B65 uses PCIe 5.0 x16. The B65 has double the lane count, which can matter for data transfer-heavy workloads.

Q: Do their display outputs differ?

A: Yes. The B60 Dual has 4x mini-DisplayPort 2.1 outputs, while the B65 has 4x DisplayPort 2.1 outputs (full-size). Both support the same API set: DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4.

The Verdict

The data indicates two cards with identical compute cores but distinct memory and power profiles. For workloads that scale with memory capacity or bandwidth, such as large dataset rendering, high-resolution texture streaming, or AI inference with big model weights, the B65 is the stronger choice. Its 608.0 GB/s bandwidth and 32 GB capacity give it a clear edge in memory-bound scenarios, while its 200 W TDP means it fits into systems with a 550 W PSU.

The B60 Dual, despite lower memory bandwidth and capacity, carries a 400 W TDP and an 800 W suggested PSU. With the same boost clock of 2400 MHz and the same FP32 throughput, the B60 Dual does not offer any compute advantage. Its only distinguishing traits are the dual-slot cooling design with specific dimensions (300 mm length, 110 mm height, 40 mm width) and the mini-DisplayPort outputs. The B65 has no recorded dimensions, so physical size comparisons are not possible from the database.

Given that the B65 reaches the same 2400 MHz boost clock at half the TDP, doubles the memory bus width, and adds 8 GB of capacity, the recorded data favors the B65 for nearly all use cases. The B60 Dual's higher power envelope does not translate into higher clock speeds or higher fill rates in the specifications. Therefore, the B65 appears to be the more efficient and capable card on paper.

Specification Differences

The two cards share the same chip, architecture, process node, foundry, transistor count, and die size. Differences appear in the following fields:

• Base clock: B60 Dual at 2000 MHz, B65 at 2400 MHz. Boost clock is identical at 2400 MHz.
• Memory size: B60 Dual 24 GB, B65 32 GB.
• Memory bus width: B60 Dual 192 bit, B65 256 bit.
• Memory bandwidth: B60 Dual 456.0 GB/s, B65 608.0 GB/s.
• TDP: B60 Dual 400 W, B65 200 W.
• Power connector: B60 Dual 1x 16-pin, B65 1x 8-pin.
• Suggested PSU: B60 Dual 800 W, B65 550 W.
• Bus interface: B60 Dual PCIe 5.0 x8, B65 PCIe 5.0 x16.
• Display outputs: B60 Dual 4x mini-DisplayPort 2.1, B65 4x DisplayPort 2.1.
• Dimensions: B60 Dual has recorded length 300 mm, height 110 mm, width 40 mm; B65 has no recorded dimensions.
• Launch MSRP: B60 Dual has a launch MSRP of 1,199 USD; B65 has no recorded launch MSRP.
• Release date: B60 Dual released 2025-09-04, B65 released 2026-03-31.

Architecture Differences

Both cards are built on the BMG-G21 chip using the Xe2-HPG architecture, which belongs to the Battlemage (Pro Series) generation. The process node is 5 nm at TSMC, with 19,600 million transistors on a 272 mm² die. Transistor density is 72.1M / mm² for both. There are no differences in the core layout: 2560 shading units, 160 TMUs, 80 ROPs, and 20 RT cores on each. The memory type is GDDR6 for both, running at 2375 MHz (19 Gbps effective). The API support is identical: DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. Both are dual-slot cards and both are marked as Active in production status.

Since the architecture, chip, and core counts are identical, the only architectural-level differences are the memory subsystem (capacity, bus width, bandwidth), the power delivery design (400 W vs 200 W), the PCIe lane allocation (x8 vs x16), and the physical output configuration (mini-DisplayPort vs full-size DisplayPort). The B65's higher base clock of 2400 MHz versus the B60 Dual's 2000 MHz suggests a different binning or power management strategy, but the boost clock being equal at 2400 MHz means peak performance is the same.

Where Each One Wins

The B65 wins in memory capacity, memory bandwidth, and power efficiency. Its 32 GB frame buffer and 608.0 GB/s bandwidth are both 33.3% higher than the B60 Dual's 24 GB and 456.0 GB/s. For workloads that frequently exceed 24 GB of video memory, such as large 3D scenes, high-resolution video editing, or multi-model AI inference, the B65 is the only card that fits. The 200 W TDP and 550 W suggested PSU also make it compatible with a wider range of systems. The x16 PCIe interface doubles the lane count over the B60 Dual's x8 link, which can reduce transfer bottlenecks for large datasets.

The B60 Dual wins in none of the recorded compute metrics, as all shader, texture, pixel, and ray tracing figures are identical. Its only advantages are the recorded physical dimensions (300 mm length, 110 mm height, 40 mm width), which may matter for chassis fit, and the mini-DisplayPort outputs, which some multi-display setups may prefer for cable management. The 400 W TDP and 800 W suggested PSU do not confer any performance benefit in the specifications. The higher power draw without a higher boost clock suggests the B60 Dual may be designed for sustained loads, but the database does not include benchmark data to confirm that behavior.

The B65's later release date (2026-03-31 versus 2025-09-04) and lack of a recorded launch MSRP make direct value comparisons impossible. However, based purely on the recorded specifications, the B65 delivers the same compute throughput, double the memory bus width, 33.3% more memory capacity and bandwidth, half the TDP, and double the PCIe lanes. The B60 Dual's only distinguishing features are its form factor dimensions and mini-DisplayPort configuration. For any workload where memory matters, the B65 is the clear pick. For workloads that are purely compute-bound and fit within 24 GB, the two cards should perform identically, since their FP32, FP16, pixel, and texture rates are all equal.