Intel Arc B570 vs Intel Arc Pro B390 Comparison

Head-to-Head Benchmarks

The Intel Arc B570 and Intel Arc Pro B390 occupy very different positions in the recorded database. The B570 has a full suite of benchmark results, while the B390 has no recorded scores in any test. This makes direct numeric comparison impossible for most workloads. The B570’s average benchmark score across all tests is 20556, placing it at the 65th percentile of all GPUs. The B390, with no benchmark entries, holds a 50th percentile rating based on its architectural classification rather than measured performance.

Looking at the B570’s individual results provides context for what a dedicated discrete card delivers. In 3DMark Steel Nomad DX12, the B570 scores 2649. Geekbench OpenCL returns 83514, while Geekbench Vulkan comes in at 96844. Passmark scores show 65 in DirectX 10, 118 in DirectX 11, 72 in DirectX 12, and 164 in DirectX 9. The G2D score is 661, G3D is 14195, and GPU compute is 7281.

The nearest rivals for the B570 show how tightly grouped its performance is. The NVIDIA GeForce RTX 3070 Mobile averages 20534, which is 0.1 percent below the B570. The Intel Arc A750 averages 20582, putting it 0.1 percent ahead. The NVIDIA Quadro M4000M averages 20480, 0.4 percent behind. The AMD Radeon R9 M390X averages 20662, 0.5 percent ahead. These deltas are small, indicating the B570 sits in a competitive band where no single result separates it clearly from its peers.

Since the B390 has no recorded benchmarks, the head-to-head section relies on the B570’s measured data and the architectural expectations for the B390. The B570’s FP32 throughput is 11.52 TFLOPS, while the B390 is rated at 7.680 TFLOPS. That is a substantial gap on paper. Texture rate for the B570 is 360.0 GTexel/s versus 120.0 GTexel/s for the B390. Pixel rate shows 200.0 GPixel/s against 60.00 GPixel/s. These figures suggest the B570 delivers significantly higher raw throughput in every rasterization category.

The B390’s boost clock matches the B570 at 2500 MHz, but its base clock is only 300 MHz compared to 2500 MHz for the B570. Memory bandwidth cannot be compared because the B390 uses system shared memory with bandwidth listed as system dependent, while the B570 has dedicated 380.0 GB/s GDDR6. The B570 carries 10 GB of GDDR6 on a 160 bit bus, while the B390 has no dedicated memory of its own.

FAQ

Q: How does the Intel Arc B570 compare to its closest rival, the Intel Arc A750?

A: The B570 averages 20556 across all benchmarks. The Arc A750 averages 20582, which is 0.1 percent higher. The difference is statistically minor, placing the two cards effectively at parity in aggregate performance.

Q: What is the performance percentile of each GPU in the database?

A: The Intel Arc B570 sits at the 65th percentile of all GPUs. The Intel Arc Pro B390 is rated at the 50th percentile.

Q: Does the Intel Arc Pro B390 have any recorded benchmark scores?

A: No. The database contains no benchmark entries for the B390. Its average benchmark score is listed as 0, and it has no nearest rivals recorded.

Q: What are the FP32 compute ratings for both GPUs?

A: The Intel Arc B570 delivers 11.52 TFLOPS of FP32 performance. The Intel Arc Pro B390 delivers 7.680 TFLOPS. The B570 offers roughly one and a half times the raw FP32 throughput of the B390 based on these figures.

Q: How do the memory configurations differ?

A: The B570 has 10 GB of GDDR6 memory on a 160 bit bus with 380.0 GB/s of bandwidth. The B390 uses system shared memory with no dedicated VRAM, and its bandwidth is listed as system dependent.

Q: What is the TDP difference between the two?

A: The B570 has a TDP of 150 W and requires a 1x 8-pin power connector with a suggested 450 W PSU. The B390 has a TDP of 80 W, uses no power connectors, and has no suggested PSU listed because it is an integrated graphics processor.

Architecture Differences

The two GPUs come from different architectural generations and manufacturing processes. The Intel Arc B570 uses the BMG-G21 chip built on Xe2-HPG architecture, part of the Battlemage (Arc 5) generation. It is fabricated on a 5 nm process at TSMC with 19,600 million transistors on a 272 mm² die. The transistor density works out to 72.1M per mm².

The Intel Arc Pro B390 uses the Panther Lake chip built on Xe3-LPG architecture, part of the Arc Graphics-WM (Panther Lake) generation. It is fabricated on a 3 nm process at Intel. Transistor count and die size are listed as unknown in the database. No transistor density is recorded.

The B570 has 2304 shading units, 144 texture mapping units, and 80 ROPs. It includes 18 ray tracing cores. The B390 has 1536 shading units, 48 TMUs, and 24 ROPs. It includes 12 ray tracing cores. The B570 has more than double the TMUs and more than triple the ROPs of the B390, which explains the large gap in texture and pixel rates.

Both GPUs support DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. Neither lists tensor core counts in the database. The B570’s FP16 rating is 23.04 TFLOPS (2:1), while the B390’s FP16 rating is 15.36 TFLOPS (2:1).

The B570 is a discrete dual-slot card measuring 272 mm in length and 115 mm in height. It uses PCIe 4.0 x8 as its bus interface and provides 1x HDMI 2.1a and 3x DisplayPort 2.1 outputs. The B390 is an integrated graphics processor with no slot width, no dimensions, no power connectors, and a bus interface listed as IGP. Its display outputs are portable device dependent.

The B570’s predecessor is Alchemist. The B390’s predecessor is HD Graphics-WM. The B570 was released on January 15, 2025. The B390 was released on January 26, 2026. Both are listed as Active in production status.

The architecture shift from Xe2-HPG to Xe3-LPG appears to prioritize integration and efficiency over raw throughput. The B390’s 3 nm process at Intel versus the B570’s 5 nm process at TSMC indicates a newer fabrication node, but the B390’s lower shading unit count and reduced ROP count show that it targets a different performance envelope.

Specification Differences

The clock behavior differs significantly. The B570 runs at a constant 2500 MHz base and 2500 MHz boost. The B390 has a 300 MHz base clock and a 2500 MHz boost clock. The wide gap between base and boost on the B390 suggests aggressive power management, while the B570 maintains a steady clock.

Memory is the most fundamental split. The B570 uses 10 GB GDDR6 at 2375 MHz with 19 Gbps effective speed, a 160 bit bus, and 380.0 GB/s bandwidth. The B390 uses system shared memory with no dedicated size, type, bus width, or bandwidth figures. Its memory clock is listed as system shared as well.

Compute resources differ across the board. The B570 has 2304 shading units, 144 TMUs, 80 ROPs, and 18 RT cores. The B390 has 1536 shading units, 48 TMUs, 24 ROPs, and 12 RT cores. The B570 leads in every category.

Rasterization rates follow the hardware counts. The B570 produces 200.0 GPixel/s and 360.0 GTexel/s. The B390 produces 60.00 GPixel/s and 120.0 GTexel/s. FP32 performance is 11.52 TFLOPS for the B570 and 7.680 TFLOPS for the B390. FP16 performance is 23.04 TFLOPS for the B570 and 15.36 TFLOPS for the B390.

Power requirements are starkly different. The B570 draws 150 W, uses a dual-slot cooler, requires a 1x 8-pin connector, and suggests a 450 W PSU. The B390 draws 80 W, is an IGP, uses no connectors, and has no suggested PSU. The B390’s launch MSRP is not recorded in the database, while the B570 launched with a 219 USD MSRP.

The bus interface separates the two as well. The B570 uses PCIe 4.0 x8. The B390 uses IGP, meaning it connects directly to the processor rather than through a PCIe slot. Display outputs for the B570 are fixed at 1x HDMI 2.1a and 3x DisplayPort 2.1. The B390’s outputs depend on the portable device it is integrated into.

Where Each One Wins

The B570 wins in every measurable performance category. Its FP32 throughput of 11.52 TFLOPS is significantly higher than the B390’s 7.680 TFLOPS. Its texture rate of 360.0 GTexel/s is triple the B390’s 120.0 GTexel/s. Its pixel rate of 200.0 GPixel/s is more than triple the B390’s 60.00 GPixel/s. In rasterization-heavy workloads, the B570 has a clear advantage.

The B570 also wins on memory. Dedicated 10 GB GDDR6 with 380.0 GB/s bandwidth provides consistent performance regardless of system memory pressure. The B390 relies on system shared memory, which introduces variable performance depending on the host device’s memory configuration. For workloads that demand predictable bandwidth, the B570 is the stronger choice.

The B390 wins on power efficiency and integration. Its 80 W TDP is substantially lower than the B570’s 150 W. It requires no power connectors and no suggested PSU, making it suitable for compact portable systems where discrete cards cannot fit. The B390’s 3 nm Intel process node is newer than the B570’s 5 nm TSMC node, which may contribute to its lower power draw.

The B570 wins on connectivity for desktop use. It offers fixed HDMI 2.1a and DisplayPort 2.1 outputs, enabling multi-display setups with known specifications. The B390’s outputs are portable device dependent, meaning users cannot rely on specific ports.

In terms of API support, neither card has an edge. Both support DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. Software compatibility should be equivalent for applications that target these APIs.

The B570 also wins on release timing for desktop users. It launched in January 2025 as part of the Battlemage generation. The B390 launched in January 2026 as part of the Panther Lake generation, but it is an integrated solution, not a standalone upgrade path for existing desktops.

For compute tasks, the B570’s higher shading unit count, TMU count, and ROP count give it more parallel execution resources. The B390’s lower counts make it less suited for GPU compute, despite its newer architecture.

The Verdict

The recorded data points to a straightforward conclusion. The Intel Arc B570 is a discrete desktop GPU with measured benchmark results, a 65th percentile ranking, and nearest rival deltas within 0.5 percent of its average score. The Intel Arc Pro B390 is an integrated GPU with no benchmark entries, a 50th percentile rating, and no recorded rival comparisons.

Users who need a dedicated graphics card for gaming, rendering, or compute workloads should choose the B570. Its 11.52 TFLOPS FP32 performance, 380.0 GB/s memory bandwidth, and 150 W power envelope make it a capable discrete solution. Its 65th percentile ranking places it above the median GPU population, and its average score of 20556 puts it in direct competition with the RTX 3070 Mobile and Arc A750.

Users who require an integrated graphics solution for a portable device should consider the B390. Its 80 W TDP, lack of power connectors, and system shared memory make it suitable for low-power environments. Its 3 nm process node and Xe3-LPG architecture represent a newer design, but the database shows no measured performance to validate its capabilities.

The B570 has a launch MSRP of 219 USD. No launch MSRP is recorded for the B390.

The B570’s 272 mm length and dual-slot design require chassis space and a 450 W PSU. The B390 has no dimensions recorded and no PSU requirement, reflecting its integrated nature.

Performance parity with rivals defines the B570’s position. A 0.1 percent lead over the RTX 3070 Mobile, a 0.1 percent deficit to the Arc A750, a 0.4 percent lead over the Quadro M4000M, and a 0.5 percent deficit to the Radeon R9 M390X all indicate a tightly contested segment. The B390 has no such comparisons, leaving its performance unquantified.

The choice depends on the target system. Desktop builders with room for a discrete card get the full suite of B570 features: dedicated memory, fixed display outputs, PCIe 4.0 x8, and measured performance. Portable device designers get the B390’s integration benefits but accept unverified performance and shared memory. The database supports the B570 for measured performance and the B390 for low-power integrated deployments.