AMD Ryzen AI Max PRO 390 socket: FP11
The AMD Ryzen AI Max PRO 390 uses FP11, soldered with 2,077 balls. It is the 12-core PRO version of the Strix Halo APU with a 55 W default TDP, and the soldered chip cannot be replaced or upgraded.
- Socket
- FP11
- Launched
- January 2025
- Cores and threads
- 12 and 24
- Default TDP
- 55 W
- Codename
- Strix Halo
- Segment
- Mobile (soldered)
- Family
- AMD Ryzen AI Max PRO
- Clocks
- 3.2 / 5.0 GHz
AMD Ryzen AI Max PRO 390 specifications
| Cores | 12 (uniform Zen 5 cores) |
|---|---|
| Threads | 24 |
| Core architecture | Zen 5, chiplet CCDs with large I/O die |
| Clocks | 3.2 GHz base, up to 5.0 GHz maximum boost |
| Cache | 64 MB L3 and 12 MB L2 (76 MB total) |
| Default TDP | 55 W |
| Configurable TDP | 45 to 120 W |
| Manufacturing process | TSMC 4 nm |
| Memory support | Soldered LPDDR5X up to 8000 MT/s, 256-bit bus, up to 128 GB unified |
| PCI Express | 16 PCIe Gen4 lanes |
| Integrated graphics | AMD Radeon 8050S Graphics, 32 compute units, up to 2800 MHz |
| NPU | XDNA 2, up to 50 TOPS |
| ECC memory | Supported with LPDDR5X |
| Package | FP11 (BGA2077, soldered) |
| Cooler in the box | None (soldered, system thermal solution required) |
Motherboard compatibility
AMD Ryzen AI Max PRO 390 needs a motherboard with a FP11 socket. Chipset by chipset data for this codename is not recorded yet. See the FP11 chipset list and the board maker's CPU support list.
FP11 at a glance
| Type | BGA, 2,077 contacts |
|---|---|
| Package size | 37.5 × 45 mm |
| Introduced | January 2025 |
| Memory | Soldered LPDDR5X only, 256-bit bus (8 channels), up to LPDDR5X-8533 and up to 128 GB of unified memory shared by CPU, GPU and NPU, with Link-ECC on embedded boards |
| Cooler mounting holes | Not applicable (soldered) |
What the Ryzen AI Max PRO 390 is
The Ryzen AI Max PRO 390 is a 12-core, 24-thread mobile workstation APU that AMD launched in January 2025 at the top of the PRO branch of the Strix Halo silicon. Its twelve uniform Zen 5 cores run at 3.2 GHz and boost to 5.0 GHz, backed by 12 MB of L2 and a large 64 MB of L3 cache. Graphics come from AMD Radeon 8050S Graphics with 32 compute units at up to 2800 MHz, and an XDNA 2 NPU delivers up to 50 TOPS for local AI work.
The PRO suffix marks the business stack. The chip carries AMD PRO Technologies for fleet deployment and maintenance, while its clocks, core count and graphics match the consumer Ryzen AI Max 390. Error correcting memory is supported on LPDDR5X, which suits workstations that hold large data sets in memory through long runs. That combination of 12 cores, ECC and a 32 compute unit GPU is unusual in a soldered laptop processor.
Strix Halo is a chiplet design. Zen 5 core chiplets sit beside a large I/O die that holds the graphics engine, the NPU and the memory controller. AMD builds the compute silicon on the TSMC 4 nm process. The laptop maker sets the cooling and power policy inside a 45 to 120 W configurable window, with 55 W as the default TDP for this SKU.
Against its PRO siblings, the 390 doubles the L3 cache of the 8-core PRO 385 to 64 MB and adds four CPU cores at the same 5.0 GHz boost. Multithreaded compiles, simulations and data preparation scale with those cores, while the unchanged 32 compute unit GPU keeps graphics throughput level across the two SKUs. Buyers choose between them on thread count and cache rather than graphics.
Socket and motherboard compatibility
The Ryzen AI Max PRO 390 is soldered to its board in the FP11 ball grid array, so there is no socket to plug into and the chip cannot be replaced or upgraded later. The 2,077-ball joint is made once at the factory on a 37.5 × 45 mm package, and the laptop design is qualified around this exact SKU for its whole service life.
There is no board or chipset choice in the retail sense. The vendor selects the soldered memory capacity, tunes the firmware and validates the thermal design for the 45 to 120 W window. Moving to the 8-core AMD Ryzen AI Max PRO 385 or the 6-core AMD Ryzen AI Max PRO 380 means a different board build, not a part swap, even though all three share the FP11 package.
The FP11 footprint is shared with the embedded branch of the same silicon, including the industrial AMD Ryzen AI Embedded X188i, which matches this SKU at 12 cores and 5.0 GHz boost for equipment with long lifecycles. The smaller FP8 package covers lower power laptops with smaller integrated graphics. Halo boards stay distinct from those mainstream designs by memory width and power class.
Memory, PCIe and cooling
Memory is soldered LPDDR5X up to 8000 MT/s on a 256-bit bus, with up to 128 GB in one unified pool shared by the CPU cores, the 32 compute unit graphics engine and the NPU. Error correcting memory is supported, so long running technical workloads get protection against single bit errors in the large shared pool. Capacity and speed are fixed when the board is built, so they rank with the SKU as permanent choices.
Expansion comes from 16 PCIe Gen4 lanes from the SoC, which typically serve NVMe storage and high speed peripherals. Display and parallel compute run on the integrated Radeon graphics, so the lanes stay free for storage and I/O duties. The exact lane split and port list are set per laptop design, so the vendor specification governs.
Cooling is a system responsibility across the 45 to 120 W configurable range. Twelve Zen 5 cores plus 32 graphics compute units make this the hardest PRO SKU to cool, so chassis choice decides how long the chip holds peak clocks under combined load. There is no boxed cooler and no retail heatsink list for a soldered mobile APU.
Upgrade paths
There is no CPU upgrade path on a soldered package. When twelve cores stop being enough, the only step inside mobile Halo silicon is a different class of machine, since this SKU already tops the PRO branch. When the workload shrinks, the 8-core AMD Ryzen AI Max PRO 385 is the step down with the same 32 compute unit GPU.
Moving off FP11 means a new platform. The FP8 family serves lower power business laptops with smaller integrated graphics, and desktop builders meet related Zen 5 silicon on socketed boards. Because memory is soldered on FP11, even a memory size change means a new board, so capacity planning belongs at the start of the project.
Frequently asked questions
What socket does the Ryzen AI Max PRO 390 use?
It uses FP11, a 2,077-ball grid array. The chip is soldered and cannot be moved to another board.
Does the PRO 390 support ECC memory?
Yes, error correcting memory is supported on LPDDR5X. That suits workstations that keep large data sets resident through long runs.
What memory does the PRO 390 use?
Soldered LPDDR5X up to 8000 MT/s on a 256-bit bus, up to 128 GB shared by the CPU, GPU and NPU. The size is fixed at board manufacture.
Does the PRO 390 include an NPU?
Yes, an XDNA 2 NPU with up to 50 TOPS, alongside 32 RDNA 3.5 graphics compute units and the twelve Zen 5 CPU cores.
Sources and verification
- AMD Ryzen AI Max PRO 390 (AMD)
- AMD announces expanded consumer and commercial AI PC portfolio at CES 2025 (AMD newsroom)
- AMD Ryzen AI Max PRO 390 specs (TechPowerUp CPU database)
Facts last checked against these sources on . Page last edited on .