AMD Radeon AI PRO R9700S vs Intel Arc B390 Comparison

The Verdict

The data shows two graphics products built for entirely different purposes. The AMD Radeon AI PRO R9700S is a high-performance discrete workstation GPU with 4096 shading units, 32 GB of GDDR6 memory, and a 300 W TDP. The Intel Arc B390 is an integrated graphics processor (IGP) with 1536 shading units, shared system memory, and an 80 W TDP. The R9700S is designed for demanding compute and professional workloads, while the B390 targets portable devices requiring basic graphics capability. Benchmark results indicate the B390 sits in the 9th percentile of all GPUs, with its nearest rivals including the NVIDIA GeForce GT 710 and GT 625 OEM, scoring only 1.3% to 2.7% higher. The R9700S, by contrast, reports a 50th percentile position with no recorded benchmark scores or nearest rivals in the database, meaning direct numerical comparison is not possible from recorded measurements. The verdict is straightforward: the R9700S is for professionals needing high-throughput rendering and AI workloads, while the B390 serves as an integrated solution for lightweight graphics in portable systems.

Architecture Differences

The R9700S uses AMD's Navi 48 chip built on the RDNA 4.0 architecture, fabricated on a 4 nm process at TSMC. It packs 53,900 million transistors into a 357 mm² die, yielding a transistor density of 151.0 million per mm². The B390 uses Intel's Panther Lake chip with the Xe3-LPG architecture on a 3 nm process at Intel's own foundry, with transistor count and die size listed as unknown. The process node difference (4 nm versus 3 nm) indicates Intel has a smaller manufacturing geometry, but the R9700S carries a massive transistor budget by comparison.

The R9700S features 4096 shading units, 256 texture mapping units, and 128 raster operation pipelines. Its 64 ray tracing cores support hardware-accelerated ray tracing. The B390 has 1536 shading units, 48 TMUs, and 24 ROPs, with 12 ray tracing cores. In raw compute throughput, the R9700S delivers 47.84 TFLOPS FP32 and 47.84 TFLOPS FP16 at a 1:1 ratio. The B390 provides 7.680 TFLOPS FP32 and 15.36 TFLOPS FP16 at a 2:1 ratio. The R9700S has over six times the FP32 throughput and over three times the FP16 throughput.

Memory architecture differs fundamentally. The R9700S has 32 GB of dedicated GDDR6 memory on a 256-bit bus with 644.6 GB/s bandwidth. The B390 uses system shared memory, meaning its bus width and bandwidth are system dependent and not fixed values. The R9700S also has a dedicated memory clock of 2518 MHz (20.1 Gbps effective), while the B390's memory clock is listed as "System Shared."

Both support DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The R9700S uses a PCIe 5.0 x16 interface, while the B390 uses an IGP bus interface. The R9700S is dual-slot with a 1x 16-pin power connector and requires a 700 W suggested PSU; the B390 has no power connectors and no suggested PSU because it is integrated.

Head-to-Head Benchmarks

The database records no head-to-head benchmark results between the R9700S and the B390. The R9700S has zero recorded benchmark scores and zero wins, while the B390 has a single recorded benchmark: 3DMark Steel Nomad DX12 with a score of 1482. This puts the B390 in the 9th percentile of all GPUs. Its nearest rivals are the NVIDIA GeForce GT 520MX scoring 1463 (1.3% behind), GeForce 800M scoring 1460 (1.5% behind), GeForce GT 625 OEM scoring 1446 (2.5% behind), and GeForce GT 710 scoring 1443 (2.7% behind). The B390 leads its nearest recorded rivals, but those rivals are themselves low-end GPUs from years past. The R9700S, with no benchmark entries, cannot be positioned against the B390 or any other GPU through recorded scores. The closest available comparison comes from architecture specifications: the R9700S has 2.67 times the shading units, 5.33 times the TMUs, 5.33 times the ROPs, and 5.33 times the ray tracing cores of the B390. Its pixel rate of 373.8 GPixel/s versus 60.00 GPixel/s indicates roughly 6.2 times the fill rate. Its texture rate of 747.5 GTexel/s versus 120.0 GTexel/s indicates roughly 6.2 times the texturing throughput.

FAQ

Q: Which GPU has more shading units?

A: The AMD Radeon AI PRO R9700S has 4096 shading units, while the Intel Arc B390 has 1536 shading units.

Q: What is the memory configuration of each product?

A: The R9700S has 32 GB of GDDR6 memory on a 256-bit bus with 644.6 GB/s bandwidth. The B390 uses system shared memory, with size, type, bus width, and bandwidth all listed as system dependent.

Q: How do their FP32 compute performances compare?

A: The R9700S delivers 47.84 TFLOPS FP32, while the B390 delivers 7.680 TFLOPS FP32. The R9700S is approximately 6.2 times higher.

Q: What are their power requirements?

A: The R9700S has a 300 W TDP, uses a 1x 16-pin power connector, and requires a 700 W suggested PSU. The B390 has an 80 W TDP, no power connectors, and no suggested PSU.

Q: What is the recorded benchmark score for the B390?

A: The B390 scores 1482 in 3DMark Steel Nomad DX12, placing it in the 9th percentile of all GPUs. Its nearest rival, the NVIDIA GeForce GT 520MX, scores 1463, which is 1.3% lower.

Q: Which product uses a smaller manufacturing process?

A: The Intel Arc B390 uses a 3 nm process at Intel, while the AMD Radeon AI PRO R9700S uses a 4 nm process at TSMC.

Where Each One Wins

The R9700S wins decisively in compute performance categories. Its 47.84 TFLOPS FP32 throughput dwarfs the B390's 7.680 TFLOPS. Its FP16 performance of 47.84 TFLOPS also exceeds the B390's 15.36 TFLOPS. The R9700S has higher pixel rate (373.8 GPixel/s versus 60.00 GPixel/s) and higher texture rate (747.5 GTexel/s versus 120.0 GTexel/s). It has 64 ray tracing cores versus 12, and 4096 shading units versus 1536. The R9700S also offers dedicated memory with 644.6 GB/s bandwidth, whereas the B390 depends on system memory performance. The R9700S supports PCIe 5.0 x16, a full-bandwidth interface for maximum data transfer. It provides four DisplayPort 2.1a outputs, while the B390's display outputs are portable device dependent.

The B390 wins in power efficiency and physical integration. Its 80 W TDP is far lower than the R9700S's 300 W TDP. The B390 has no power connectors and requires no suggested PSU, making it suitable for compact portable devices. It uses an IGP bus interface, meaning no separate card slot is needed. Its 3 nm process node is smaller than the R9700S's 4 nm node, which contributes to lower power draw. The B390 also has a lower base clock of 300 MHz versus 1660 MHz on the R9700S, but its boost clock reaches 2500 MHz versus 2920 MHz on the R9700S. The B390's release date of January 26, 2026, comes after the R9700S's December 10, 2025 release. The B390 supports FP16 at a 2:1 ratio, meaning it can process half-precision data at twice the FP32 rate, a feature the R9700S does not offer since its FP16 runs at 1:1.

Specification Differences

The two products differ across nearly every recorded specification.

• Chip and architecture: R9700S uses Navi 48 with RDNA 4.0. B390 uses Panther Lake with Xe3-LPG.
• Process node: R9700S is 4 nm at TSMC. B390 is 3 nm at Intel.
• Transistors: R9700S has 53,900 million. B390 is unknown.
• Die size: R9700S is 357 mm². B390 is unknown.
• Transistor density: R9700S is 151.0M per mm². B390 has no recorded value.
• Base clock: R9700S is 1660 MHz. B390 is 300 MHz.
• Boost clock: R9700S is 2920 MHz. B390 is 2500 MHz.
• Game clock: R9700S is 2350 MHz. B390 has no game clock.
• Memory clock: R9700S is 2518 MHz (20.1 Gbps effective). B390 is system shared.
• Memory size: R9700S is 32 GB GDDR6. B390 is system shared.
• Memory bus width: R9700S is 256 bit. B390 is system shared.
• Memory bandwidth: R9700S is 644.6 GB/s. B390 is system dependent.
• Shading units: R9700S has 4096. B390 has 1536.
• TMUs: R9700S has 256. B390 has 48.
• ROPs: R9700S has 128. B390 has 24.
• Ray tracing cores: R9700S has 64. B390 has 12.
• Pixel rate: R9700S is 373.8 GPixel/s. B390 is 60.00 GPixel/s.
• Texture rate: R9700S is 747.5 GTexel/s. B390 is 120.0 GTexel/s.
• FP32 performance: R9700S is 47.84 TFLOPS. B390 is 7.680 TFLOPS.
• FP16 performance: R9700S is 47.84 TFLOPS (1:1). B390 is 15.36 TFLOPS (2:1).
• TDP: R9700S is 300 W. B390 is 80 W.
• Slot width: R9700S is dual-slot. B390 is IGP.
• Power connectors: R9700S uses 1x 16-pin. B390 has none.
• Suggested PSU: R9700S is 700 W. B390 has none.
• Bus interface: R9700S is PCIe 5.0 x16. B390 is IGP.
• Display outputs: R9700S has 4x DisplayPort 2.1a. B390 is portable device dependent.
• Dimensions: R9700S is 267 mm long, 109 mm high, 39 mm wide. B390 has no recorded dimensions.
• Release date: R9700S is December 10, 2025. B390 is January 26, 2026.
• Benchmark score: R9700S has none recorded. B390 has 1482 in 3DMark Steel Nomad DX12.
• Percentile: R9700S is 50th percentile. B390 is 9th percentile.
• Average benchmark score: R9700S is 0. B390 is 1482.
• Nearest rivals: R9700S has none recorded. B390 lists NVIDIA GeForce GT 520MX, GeForce 800M, GeForce GT 625 OEM, and GeForce GT 710.