Intel Arc Pro B60 vs Intel Arc Pro B65 Comparison

Intel Arc Pro B60 and Intel Arc Pro B65 are two professional graphics cards built on the same Xe2-HPG architecture and the same BMG-G21 chip. Both target workstation and professional visualization workloads, but the data shows a clear split in capabilities, memory capacity, and interface specifications. The B60 has a recorded launch MSRP of 499 USD, while the B65 has no recorded launch MSRP in the database.

The Verdict

The Intel Arc Pro B60 and Intel Arc Pro B65 share identical compute resources: 2560 shading units, 160 texture mapping units, 80 raster operation units, and 20 ray tracing cores. Both cards deliver the same FP32 performance of 12.29 TFLOPS and the same FP16 performance of 24.58 TFLOPS (2:1). Pixel rate is identical at 192.0 GPixel/s, and texture rate is identical at 384.0 GTexel/s. The core architecture is the same, so raw shader throughput is not a differentiator.

The B65 distinguishes itself with a larger memory subsystem: 32 GB of GDDR6 on a 256 bit bus, delivering 608.0 GB/s of bandwidth. The B60 offers 24 GB of GDDR6 on a 192 bit bus, with 456.0 GB/s of bandwidth. That is a 33% advantage in memory capacity and a 33% advantage in bandwidth for the B65. The B65 also uses a PCIe 5.0 x16 interface, while the B60 uses PCIe 5.0 x8. This makes the B65 the better choice for workloads that are memory-bound or require large datasets that exceed 24 GB, such as high-resolution rendering, large model inference, or multi-display visualization.

The B60, however, is the only one of the two with recorded benchmark results in the database. It has a percentile rank of 20 against all GPUs, an average benchmark score of 3182, and a set of direct rivals with comparable average scores. The B65 has no benchmark entries, no average score, and no nearest rivals recorded; its percentile rank of 50 is a placeholder, not a measured result. For buyers who need a working, measured solution today, the B60 is the only card with verified performance data. The B65 is a higher-spec alternative that trades the B60's measured track record for more memory and a wider bus, but its actual performance in the database remains unquantified.

Architecture Differences

Both cards use the BMG-G21 chip built on a 5 nm process at TSMC. The die size is 272 mm², and the transistor count is 19,600 million, yielding a transistor density of 72.1M per mm². The architecture is Xe2-HPG, part of the Battlemage Pro Series generation. Both cards support DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4.

The fundamental architectural difference lies in the memory interface and the PCIe connection. The B60 has a 192 bit memory bus with 24 GB of GDDR6, while the B65 has a 256 bit bus with 32 GB of GDDR6. Memory clock is the same for both: 2375 MHz, operating at 19 Gbps effective. The wider bus on the B65 directly produces the higher bandwidth figure of 608.0 GB/s versus 456.0 GB/s on the B60. The B65 also uses a PCIe 5.0 x16 connection, double the lane width of the B60's PCIe 5.0 x8. This affects how quickly data can move between the GPU and the host system, particularly for workloads that stream data in and out of VRAM.

Clock behavior differs between the two cards. The B60 has a base clock of 2000 MHz and a boost clock of 2400 MHz, meaning it starts lower and boosts up. The B65 has a base clock of 2400 MHz and a boost clock of 2400 MHz, so it runs at its maximum clock continuously. Both cards have the same TDP of 200 W, the same dual-slot cooling design, and the same single 8-pin power connector, with a suggested PSU of 550 W.

Display outputs differ in form factor but not count. The B60 provides 4x mini-DisplayPort 2.1 connectors, while the B65 provides 4x full-size DisplayPort 2.1 connectors. The B60 has recorded physical dimensions of 167 mm in length, 69 mm in height, and 40 mm in width; the B65 has no recorded dimensions in the database.

Where Each One Wins

Based on the recorded data, the B60 wins in availability of measured performance and in physical compactness. It has a known length of 167 mm, a known height of 69 mm, and a known width of 40 mm, which makes it a dual-slot card that fits in smaller chassis. The B65 has no recorded dimensions, so its physical footprint is unknown from the database. The B60 also has benchmark scores that allow direct comparison with rivals, which is not possible for the B65.

The B65 wins on memory capacity and bandwidth. With 32 GB of GDDR6, it holds 8 GB more than the B60. With 608.0 GB/s of bandwidth, it outperforms the B60 by 152.0 GB/s. For workloads such as large texture sets, high-resolution framebuffers, or datasets that approach or exceed 24 GB, the B65 has a clear technical advantage. The B65 also has a wider PCIe 5.0 x16 interface, which can reduce transfer bottlenecks for data-heavy workflows.

The compute core is a tie. Both cards deliver the same FP32 and FP16 throughput, the same pixel and texture rates, and the same ray tracing core count. For shader-bound or RT-bound workloads, the data shows no difference between the two. The B60 and B65 also share the same TDP of 200 W, the same power connector, and the same suggested PSU, so power delivery requirements are identical.

The B60 has a recorded release date of 2025-09-04, while the B65 has a recorded release date of 2026-03-31. The B60 is the earlier product, and it has an active production status. The B65 also has an active production status, but its later release date means it is a newer addition to the lineup.

FAQ

Q: Which card has more memory?

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

Q: Do the two cards have the same compute performance?

A: Yes, both cards have identical shading units (2560), TMUs (160), ROPs (80), and ray tracing cores (20). Both deliver 12.29 TFLOPS FP32 and 24.58 TFLOPS FP16 (2:1), with the same pixel rate of 192.0 GPixel/s and texture rate of 384.0 GTexel/s.

Q: What is the memory bandwidth difference?

A: The B65 has a bandwidth of 608.0 GB/s, while the B60 has a bandwidth of 456.0 GB/s. The difference is 152.0 GB/s in favor of the B65.

Q: Are there any benchmark scores for the B65?

A: No, the database contains no benchmark entries for the Intel Arc Pro B65. Its average benchmark score is recorded as 0, and its nearest rivals list is empty. The B60 has benchmark scores for 3dmark_3dmark_steel_nomad_dx12 (2646), passmark_directx_10 (61), passmark_directx_11 (122), passmark_directx_12 (76), passmark_directx_9 (179), passmark_g2d (763), passmark_g3d (14580), and passmark_gpu_compute (7029).

Q: What is the PCIe interface difference?

A: The B60 uses PCIe 5.0 x8, while the B65 uses PCIe 5.0 x16. The B65 has double the PCIe lane width.

Q: Do both cards use the same power configuration?

A: Yes, both cards have a TDP of 200 W, use a dual-slot cooling design, require a single 8-pin power connector, and have a suggested PSU of 550 W.

Head-to-Head Benchmarks

There are no recorded head-to-head benchmark results between the Intel Arc Pro B60 and the Intel Arc Pro B65 in the database. The winsA and winsB fields are both 0, and the headToHeadBenchmarks array is empty. This means the two cards have never been tested against each other in a direct comparison. Any performance difference must be inferred from the individual specifications and the B60's standalone benchmark scores.

The B60's average benchmark score is 3182, with a percentile rank of 20 against all GPUs. Its nearest rivals provide context for this score. The NVIDIA Quadro P1000 has an average score of 3163, which is 0.6% lower than the B60. The NVIDIA GeForce GT 640 has an average score of 3210, which is 0.9% higher than the B60. The NVIDIA GeForce 920M has an average score of 3287, which is 3.2% higher. The NVIDIA GeForce RTX 5080 SUPER has an average score of 3075, which is 3.5% lower than the B60. This places the B60 in a narrow performance band around the 3182 mark, within roughly 3.5% of its immediate rivals.

The B60's individual benchmark scores show a specific pattern. In 3dmark_3dmark_steel_nomad_dx12, it scores 2646. In PassMark tests, the highest score is passmark_g3d at 14580, followed by passmark_gpu_compute at 7029. The legacy DirectX tests are lower: passmark_directx_9 scores 179, passmark_directx_11 scores 122, passmark_directx_12 scores 76, and passmark_directx_10 scores 61. The passmark_g2d score is 763. These scores indicate that the B60 performs substantially better in modern 3D workloads and compute tasks than in legacy DirectX 10 or 12 tests.

For the B65, no such analysis is possible from the database because no benchmark data exists. The B65's specifications suggest that it should outperform the B60 in memory-bound scenarios, given its 33% larger capacity and 33% higher bandwidth. However, the database does not confirm this with any measured score. The B65's identical compute core (2560 shading units, 12.29 TFLOPS FP32) means that in pure shader throughput, the two cards should behave identically. The differentiator is entirely in the memory subsystem and the PCIe interface.

The B60's boost clock of 2400 MHz matches the B65's constant clock of 2400 MHz. The B60's base clock is lower at 2000 MHz, but since the boost clock is the same, peak clock behavior is equal. Both cards have the same memory clock of 2375 MHz (19 Gbps effective). The B65's wider 256 bit bus is the sole reason for its higher bandwidth; the memory chips themselves run at the same speed.

The B60's measured performance relative to the NVIDIA Quadro P1000 (0.6% higher average score) and the NVIDIA GeForce RTX 5080 SUPER (3.5% higher average score) gives a sense of where it sits in the broader GPU landscape. The B65, with no recorded scores, cannot be placed in that landscape. Buyers considering the B65 must rely on its memory and interface advantages over the B60, but the database offers no measured evidence of its actual performance level. The B60 is the only one of the two with verified benchmark results, making it the only card with a data-backed performance profile.