AMD Ryzen AI Embedded P185 vs AMD Ryzen AI Max 390 Comparison

The AMD Ryzen AI Embedded P185 and AMD Ryzen AI Max 390 are both 12-core, 24-thread AMD mobile processors, yet the database records two opposite kinds of leadership. In the direct head-to-head PassMark suite, the Max 390 wins every entry, while the P185 holds the higher aggregate benchmark average and the higher percentile rank. That split makes this comparison less about a single fastest part and more about which metric matters for a given workload.

Head-to-Head Benchmarks

The Max 390's clean sweep is the defining result. The database lists 11 head-to-head PassMark comparisons, and the Max 390 wins all 11. The largest margin is in passmark_find_prime_numbers: the Max 390 scores 316 against the P185's 129, a 59.2% gap. Physics follows at 2761 versus 1772, a 35.8% deficit for the P185. Extended instructions shows 38716 versus 26544, a 31.4% gap. Multithread scores 41737 versus 31817, a 23.8% gap. Random string sorting lands at 53113 versus 40557, a 23.6% gap. Data compression records 487145 versus 374429, a 23.1% gap. Floating point math is 90594 versus 70587, a 22.1% gap. Data encryption is 25097 versus 19612, a 21.9% gap. Integer math is 146519 versus 117832, a 19.6% gap. The smallest margin appears in the single-thread entries: 4028 versus 3977, only 1.3%.

The P185 has zero direct wins. Its best direct showing is the single-thread result, where it trails by just 1.3%. That narrow margin suggests the P185's Zen 5 / Zen 5c core arrangement can stay close on lightly threaded work, even though the Max 390 still leads.

The aggregate average complicates that sweep. The P185 records an average benchmark score of 62839, above the Max 390's 56273, and a percentile of 93 versus 91. In other words, the part that loses every head-to-head PassMark test sits higher in the database-wide average. The database does not specify which tests feed the aggregate average; the measured values simply show the divergence.

The nearest-rival sets reinforce the split. The P185 sits 0.2% above the Intel Core Ultra 7 255HX and 0.5% above the AMD Ryzen AI 9 PRO 465, while trailing the Intel Core i7-13790F by 0.4% and the Intel Core Ultra 7 265HX by 0.5%. The Max 390 sits 0.5% above the Intel Core i9-14900HX and 0.6% above the Intel Core 7 253PQE, while trailing the AMD Ryzen AI 9 HX PRO 470 by 0.1% and the AMD Ryzen Threadripper PRO 3955WX by 0.5%.

FAQ

Q: Which processor wins the direct PassMark comparisons?

A: The AMD Ryzen AI Max 390 wins all 11 head-to-head entries. The largest margin is in passmark_find_prime_numbers, where 316 beats 129 by 59.2%. The smallest margin is in single-thread tests, where 4028 beats 3977 by 1.3%.

Q: Does the P185 beat the Max 390 anywhere?

A: Not in the direct head-to-head list. The P185 has zero wins there. However, the P185 posts a higher aggregate average benchmark score, 62839 versus 56273, and a higher CPU percentile, 93 versus 91.

Q: What are the clock and power differences?

A: The P185 has a 2.00 GHz base clock and a 5.10 GHz boost clock. The Max 390 has a 3.20 GHz base clock and a 5.00 GHz boost clock. The P185 carries a 28 W TDP, while the Max 390 carries a 55 W TDP.

Q: How do the memory subsystems differ?

A: The P185 uses a dual-channel memory bus with 89.6 GB/s of bandwidth and supports DDR5 and LPDDR5X. The Max 390 uses a quad-channel LPDDR5X bus with 256.0 GB/s of bandwidth. Both support ECC memory.

Q: Are the core and thread counts identical?

A: Yes, both processors have 12 cores and 24 threads. They also share 80 KB of L1 cache per core and 1 MB of L2 cache per core. The L3 cache differs: 16 MB on the P185 versus 64 MB shared on the Max 390.

Q: What integrated graphics does each processor list?

A: The P185 lists a Radeon 890M. The Max 390 lists a Radeon 8050S.

The Verdict

The direct benchmark data is unambiguous about compute performance. The Max 390 wins every PassMark sub-test in the head-to-head list, and it does so by wide margins in memory-sensitive and parallel workloads. The P185's only direct bright spot is single-thread performance, where the 1.3% gap is small enough to place the two parts in the same general range.

However, the aggregate metrics flip the ranking. The P185's average score of 62839 and 93rd percentile sit above the Max 390's 56273 and 91st percentile. The nearest-rival table for each part also reflects different neighborhoods: the P185 competes against Intel Core Ultra 7-class parts and the AMD Ryzen AI 9 PRO 465, while the Max 390 competes against the AMD Ryzen AI 9 HX PRO 470, Intel Core i9-14900HX, AMD Ryzen Threadripper PRO 3955WX, and Intel Core 7 253PQE.

The data supports two separate conclusions. For the recorded PassMark workload set, the Max 390 is the stronger processor. For the database's overall average and percentile, the P185 is the higher-ranked part. Neither conclusion cancels the other; they reflect different measurement layers.

Specification Differences

| Field | AMD Ryzen AI Embedded P185 | AMD Ryzen AI Max 390 |

|---|---|---|

| Base clock | 2.00 GHz | 3.20 GHz |

| Boost clock | 5.10 GHz | 5.00 GHz |

| TDP | 28 W | 55 W |

| Socket | AMD Socket FP8 | AMD Socket FP11 |

| Codename | Gorgon Point | Strix Halo |

| Generation | Ryzen AI Embedded (Zen 5 / Zen 5c) | Ryzen AI Max (Zen 5 (Strix Halo)) |

| Architecture field | Not specified | Zen 5 |

| Die size | 233 mm² | 2x 70.6 mm² |

| L3 cache | 16 MB | 64 MB (shared) |

| Memory support | DDR5, LPDDR5X | LPDDR5X |

| Memory bus | Dual-channel | Quad-channel |

| Memory bandwidth | 89.6 GB/s | 256.0 GB/s |

| Integrated graphics | Radeon 890M | Radeon 8050S |

| Release date | 2026-02-28 | 2025-01-05 |

| Part number | unknown | 100-000001423 |

Architecture Differences

Both processors are built on TSMC's 4 nm process and share the same 12-core, 24-thread layout, with 80 KB of L1 cache per core and 1 MB of L2 cache per core. The architectural split begins with the core design. The P185's generation line identifies Zen 5 / Zen 5c cores, so it mixes full Zen 5 cores with denser Zen 5c cores. The Max 390 lists a straight Zen 5 architecture under the Strix Halo codename. Gorgon Point and Strix Halo are therefore different designs despite the shared process.

Cache and memory amplify the difference. The P185 has 16 MB of L3; the Max 390 has 64 MB of shared L3. The Max 390 also uses a quad-channel LPDDR5X memory bus with 256.0 GB/s, while the P185 uses a dual-channel bus supporting DDR5 and LPDDR5X at 89.6 GB/s. Both support ECC memory and use PCIe Gen 4 with 16 CPU-only lanes. The integrated graphics differ as well: Radeon 890M on the P185, Radeon 8050S on the Max 390. The die layout is also distinct: a 233 mm² die for the P185 versus two 70.6 mm² dies for the Max 390.

Where Each One Wins

Based solely on the recorded data, the Max 390 wins every direct workload comparison. Its largest wins are in prime-number finding at 59.2%, physics at 35.8%, extended instructions at 31.4%, multithread at 23.8%, random string sorting at 23.6%, data compression at 23.1%, floating-point math at 22.1%, data encryption at 21.9%, and integer math at 19.6%. Its smallest win is single-thread at 1.3%. This makes the Max 390 the clear winner for the specific PassMark tasks in the head-to-head list.

The P185 does not win any direct head-to-head PassMark test, but it leads in the aggregate metrics. Its average benchmark score is 62839 versus 56273, and its percentile is 93 versus 91. It also has a lower TDP, 28 W versus 55 W, and a higher boost clock, 5.10 GHz versus 5.00 GHz. In terms of use-case split, the Max 390 is the part for the measured compute suite, while the P185 is the part that ranks higher in the database-wide average and carries a lower power envelope.