Intel Arc A380E vs Intel Arc B390 Comparison

Head-to-Head Benchmarks

The recorded data contains only one benchmark score for the Intel Arc B390, a 3DMark Steel Nomad DX12 result of 1482 points. The Intel Arc A380E has no benchmark entries in the database, so there are no direct head-to-head measurements to compare. The Arc B390's score places it at the 9th percentile among all GPUs, which indicates it sits in the lower performance tier of the database. Against its nearest rivals, the Arc B390 leads the NVIDIA GeForce GT 520MX by 1.3%, the GeForce 800M by 1.5%, the GeForce GT 625 OEM by 2.5%, and the GeForce GT 710 by 2.7%. These margins are narrow, showing that the Arc B390 is competitive with those older discrete parts but does not deliver a substantial generational leap in this particular test.

The Arc A380E cannot be assessed through benchmark scores because the database has no recorded measurements for it. Its percentile ranking of 50 places it at the midpoint of all GPUs in the database, but that ranking is not backed by an average benchmark score. The absence of head-to-head results means that any performance comparison between the Arc A380E and the Arc B390 must rely on architectural specifications and compute throughput figures rather than direct test data.

Architecture Differences

The two parts come from entirely different design lineages. The Intel Arc A380E uses the DG2-128 chip built on the Xe-HPG architecture, part of the Alchemist generation for Arc 3 products. It is fabricated on a 6 nm process at TSMC, with 7,200 million transistors on a 157 mm² die, giving a transistor density of 45.9 million per square millimeter. The Arc B390 uses the Panther Lake chip with the Xe3-LPG architecture, part of the Arc Graphics-M generation for Panther Lake. It is built on a 3 nm process at Intel, though its transistor count and die size are listed as unknown. The manufacturing transition from TSMC to Intel and from 6 nm to 3 nm is a major architectural shift, but without die size data the density comparison cannot be quantified.

The compute layout differs significantly. The Arc A380E has 1,024 shading units, 64 texture mapping units, 32 raster output pipelines, and 8 ray tracing cores. The Arc B390 has 1,536 shading units, 48 TMUs, 24 ROPs, and 12 ray tracing cores. The B390 carries 50% more shading units and 50% more ray tracing cores, but the A380E has more texture units and more ROPs. This creates an interesting split: the B390 has more raw shader throughput potential, while the A380E has stronger fixed-function pixel and texture processing resources.

The floating point figures reflect this. The Arc A380E delivers 4.096 TFLOPS FP32 and 8.192 TFLOPS FP16 (2:1 ratio), while the Arc B390 delivers 7.680 TFLOPS FP32 and 15.36 TFLOPS FP16 (2:1 ratio). The B390 offers 87.5% higher FP32 throughput and 87.5% higher FP16 throughput than the A380E. Pixel rate favors the A380E at 64.00 GPixel/s versus 60.00 GPixel/s for the B390, a 6.7% advantage. Texture rate favors the A380E at 128.0 GTexel/s versus 120.0 GTexel/s for the B390, a 6.7% advantage. So the B390 wins in shader compute and ray tracing core count, while the A380E wins in rasterization throughput.

Clock behavior is another major difference. The A380E runs at a fixed 2000 MHz base and boost, with memory at 1937 MHz (15.5 Gbps effective). The B390 has a 300 MHz base clock and a 2500 MHz boost clock, a much wider range that reflects its integrated nature and power management. The B390's boost clock is 25% higher than the A380E's fixed clock.

Memory architecture is fundamentally different. The A380E has 6 GB of dedicated GDDR6 on a 96 bit bus with 186.0 GB/s bandwidth. The B390 uses system shared memory, with shared type, shared bus width, and system dependent bandwidth. This is the most consequential difference for gaming and compute workloads that are memory sensitive. Dedicated VRAM on the A380E provides predictable bandwidth, while the B390's performance depends entirely on the host system's memory configuration.

The A380E is a discrete, single slot card with no power connectors, a 75 W TDP, and a suggested PSU of 250 W. It connects via PCIe 4.0 x8 and has four DisplayPort 2.0 outputs. The B390 is an integrated graphics processor with an 80 W TDP, no slot width, no power connectors, no suggested PSU, an IGP bus interface, and display outputs that are portable device dependent. The A380E is end-of-life, released on March 31, 2024, with a predecessor of Xe Graphics and a successor of Battlemage. The B390 is active, released on January 26, 2026, with no predecessor or successor listed.

Where Each One Wins

The Arc A380E wins in memory capacity and bandwidth. Six gigabytes of dedicated GDDR6 at 186.0 GB/s gives it a fixed, reliable memory subsystem. The B390's system shared memory cannot guarantee bandwidth, and its performance will vary with the host platform. Workloads that require large textures or high memory throughput favor the A380E.

The A380E also wins in rasterization throughput. Its 64.00 GPixel/s pixel rate and 128.0 GTexel/s texture rate both exceed the B390's figures. Older games and workloads that rely on pixel fill and texture filtering rather than shader complexity will see an advantage on the A380E.

The Arc B390 wins in shader compute. Its 7.680 TFLOPS FP32 and 15.36 TFLOPS FP16 are far ahead of the A380E. Applications that scale with raw shader throughput, such as modern DirectX 12 titles with heavy compute shaders, will favor the B390. The B390 also has more ray tracing cores, 12 versus 8, which suggests better ray tracing performance in supported titles.

The B390 wins in process technology and power efficiency per shader. The 3 nm Intel process versus the 6 nm TSMC process is a significant node advantage. The B390 delivers 87.5% more FP32 throughput while drawing only 6.7% more TDP (80 W versus 75 W). The A380E produces 4.096 TFLOPS at 75 W, while the B390 produces 7.680 TFLOPS at 80 W. That is a substantial efficiency advantage for the B390 in raw compute per watt.

The A380E wins on connectivity options for desktop use. Four DisplayPort 2.0 outputs and a PCIe 4.0 x8 interface make it suitable for multi monitor setups and standard desktop motherboards. The B390's portable device dependent display outputs and IGP bus interface tie it to specific mobile or compact platforms.

FAQ

Q: Which GPU has a higher FP32 compute throughput?

A: The Intel Arc B390 delivers 7.680 TFLOPS FP32, which is 87.5% higher than the Intel Arc A380E's 4.096 TFLOPS.

Q: How does the memory configuration differ between the two?

A: The Arc A380E has 6 GB of dedicated GDDR6 on a 96 bit bus with 186.0 GB/s bandwidth. The Arc B390 uses system shared memory with system dependent bandwidth.

Q: Which GPU has more ray tracing cores?

A: The Arc B390 has 12 ray tracing cores, while the Arc A380E has 8 ray tracing cores.

Q: What is the production status of each GPU?

A: The Arc A380E is end-of-life, released on March 31, 2024, with Battlemage as its successor. The Arc B390 is active, released on January 26, 2026, with no successor listed.

Q: Which GPU has a higher boost clock?

A: The Arc B390 has a 2500 MHz boost clock, while the Arc A380E is fixed at 2000 MHz for both base and boost.

Q: How does the Arc B390 compare to its nearest rival in the database?

A: The Arc B390 scores 1482 in 3DMark Steel Nomad DX12, leading the NVIDIA GeForce GT 520MX by 1.3%, the GeForce 800M by 1.5%, the GeForce GT 625 OEM by 2.5%, and the GeForce GT 710 by 2.7%.

Specification Differences

The two GPUs differ in nearly every major specification category. The A380E uses the DG2-128 chip with Xe-HPG architecture from the Alchemist generation, built on TSMC's 6 nm process with 7,200 million transistors and a 157 mm² die. The B390 uses the Panther Lake chip with Xe3-LPG architecture from the Arc Graphics-M generation, built on Intel's 3 nm process with unknown transistor count and die size. The A380E's transistor density is 45.9M per square millimeter; the B390's is not listed.

Clock speeds differ sharply. The A380E has a 2000 MHz base and 2000 MHz boost. The B390 has a 300 MHz base and 2500 MHz boost. Memory clocks are 1937 MHz (15.5 Gbps effective) for the A380E, while the B390 uses system shared memory.

Shading units: 1024 for the A380E, 1536 for the B390. TMUs: 64 versus 48. ROPs: 32 versus 24. Ray tracing cores: 8 versus 12. Pixel rate: 64.00 GPixel/s versus 60.00 GPixel/s. Texture rate: 128.0 GTexel/s versus 120.0 GTexel/s. FP32: 4.096 TFLOPS versus 7.680 TFLOPS. FP16: 8.192 TFLOPS versus 15.36 TFLOPS. TDP: 75 W versus 80 W.

Form factor and interface are entirely different. The A380E is single slot, 254 mm long, 127 mm high, and 20 mm wide, with PCIe 4.0 x8 and four DisplayPort 2.0 outputs. The B390 is an IGP with no dimensions, no bus interface beyond IGP, and portable device dependent display outputs. Both have no power connectors. The A380E has a suggested PSU of 250 W; the B390 has none listed.

Both support DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. Both lack tensor cores. Release dates differ: March 31, 2024 for the A380E, January 26, 2026 for the B390. Production status is end-of-life for the A380E and active for the B390.

The Verdict

The data supports a clear split by use case. The Arc A380E is a discrete desktop GPU with dedicated memory, fixed clocks, and a standard PCIe interface. Its 6 GB GDDR6 at 186.0 GB/s and higher pixel and texture rates make it the better choice for workloads that depend on memory bandwidth and rasterization throughput. Its four DisplayPort 2.0 outputs suit multi monitor desktop configurations. The end-of-life status means it is a mature product with a defined successor in Battlemage.

The Arc B390 is an integrated processor with a 3 nm node, higher shader counts, higher FP32 and FP16 throughput, and more ray tracing cores. Its 7.680 TFLOPS FP32 at 80 W TDP shows strong compute efficiency. The system shared memory and IGP bus interface tie its real world performance to the host platform. The 1482 score in 3DMark Steel Nomad DX12 and the 9th percentile ranking indicate modest absolute performance, narrowly ahead of several older NVIDIA parts. The B390 is the better choice for compact or portable systems where the CPU and GPU are integrated, and where shader compute matters more than dedicated memory capacity. The A380E is the better choice for a discrete desktop card with predictable memory performance and fixed rasterization throughput. Neither part dominates the other across all categories; the selection depends on whether the workload favors dedicated VRAM and pixel processing on the A380E or raw shader compute and ray tracing core count on the B390.