AMD Ryzen AI Embedded P164 vs Intel Core Ultra 7 255H Comparison

The AMD Ryzen AI Embedded P164 and Intel Core Ultra 7 255H are both active mobile processors with a 28W TDP, dual-channel DDR5/LPDDR5X memory support, and ECC memory support. The AMD part uses 8 cores and 16 threads on TSMC 4 nm, while the Intel part uses 16 cores and 16 threads on TSMC 3 nm. In the recorded Passmark comparisons, Intel takes 8 wins and AMD takes 3, but the database's overall percentile ranks AMD at 91 versus Intel at 83. The two parts split cleanly by workload type.

Head-to-Head Benchmarks

Intel's biggest wins are decisive. In Passmark find prime numbers, Intel scores 303 against AMD's 71, a 76.6% advantage. In physics, Intel leads 2254 to 1210, a 46.3% margin. Floating point math goes Intel's way by 43.5%, 98796 to 55799. Data encryption favors Intel by 31.4%, 23395 to 16055. Multithread throughput is also Intel's, 30703 to 25889, a 15.7% margin. Random string sorting and single-thread tests are narrower: Intel leads 36058 to 34801, a 3.5% margin, and 4317 to 4029, a 6.7% margin.

AMD's wins are concentrated but real. Integer math goes to AMD by 12.8%, 87940 to 77975. Data compression goes to AMD by 9.7%, 327891 to 298850. Extended instructions go to AMD by 1.8%, 24193 to 23755. The recorded win count is 3 for AMD and 8 for Intel.

FAQ

Q: Which processor has more cores and threads?

A: The Intel Core Ultra 7 255H has 16 cores and 16 threads. The AMD Ryzen AI Embedded P164 has 8 cores and 16 threads. Both expose 16 threads.

Q: Which processor has the higher boost clock?

A: Intel's boost clock is 5.10 GHz; AMD's is 5.00 GHz. Both have a 2.00 GHz base clock and a 28W TDP.

Q: Which processor has more cache?

A: Intel has 192 KB L1 and 3 MB L2 per core, plus 24 MB shared L3. AMD has 80 KB L1 and 1 MB L2 per core, plus 8 MB L3.

Q: Do both processors support ECC memory?

A: Yes. Both support ECC memory, and both support DDR5 and LPDDR5X on a dual-channel bus. Intel's recorded memory bandwidth is 102.4 GB/s; AMD's is 89.6 GB/s.

Q: Which processor wins the Passmark single-thread test?

A: Intel wins with 4317 versus AMD's 4029, a 6.7% advantage.

Q: What additional benchmark data is recorded for Intel?

A: Intel has Cinebench R15 multicore 1515 and single-core 251, R20 multicore 6381 and single-core 900, R23 multicore 9240 and single-core 1843, and Geekbench multicore 14024 and single-core 2335. No AMD equivalents are recorded in the database.

Architecture Differences

AMD builds the P164 around Gorgon Point with Zen 5 / Zen 5c cores on TSMC 4 nm, and the package has a recorded 233 mm² die size. Intel builds the 255H on Arrow Lake-H with TSMC 3 nm, and no die size is recorded. The core counts differ: AMD uses 8 cores with 16 threads, Intel uses 16 cores with 16 threads. The cache hierarchy is also different. AMD provides 80 KB L1 and 1 MB L2 per core, with 8 MB L3. Intel provides 192 KB L1 and 3 MB L2 per core, with 24 MB shared L3.

Memory and I/O features differ. Intel's memory bandwidth is 102.4 GB/s versus AMD's 89.6 GB/s. AMD uses PCIe Gen 4 with 16 CPU lanes; Intel uses PCIe Gen 5 with 20 CPU lanes. Both support DDR5 and LPDDR5X, dual-channel memory, and ECC. Integrated graphics differ: AMD uses Radeon 880M, Intel uses Arc Graphics 140T. The sockets are AMD Socket FP8 for the AMD part and Intel BGA 2049 for the Intel part. Intel's release date is 2025-01-12; AMD's is 2026-03-08. The AMD part's part number is unknown, while Intel's is SRQAN.

The Verdict

The database's overall percentile ranks AMD at 91 and Intel at 83, and AMD's average benchmark score is 52901 versus Intel's 33537. Yet the common Passmark comparisons assign Intel 8 wins and AMD 3. The data indicates a workload-dependent verdict rather than a single overall winner. Intel is stronger in prime-number finding, physics, floating-point math, encryption, multithread throughput, random string sorting, and single-thread execution. AMD is stronger in integer math, data compression, and extended instructions.

Nearest-rival data reinforces this. AMD's average score sits 0.1% above the AMD Ryzen 5 9500F, which has an average score of 52873, 0.1% below the Intel Xeon 634 at 52974, 0.6% below the AMD EPYC 7313P at 53206, and 0.7% below the AMD Ryzen 9 7900X at 53288. Intel's average score sits level with the AMD Ryzen 5 240 at 33542, 0.4% below the AMD Ryzen 7 8840HS at 33667 and the AMD Ryzen 5 7645HX at 33668, and 0.5% above the Intel Core i5-12600HX at 33375.

Specification Differences

The recorded specification differences are:

| Attribute | AMD Ryzen AI Embedded P164 | Intel Core Ultra 7 255H |

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

| Series | Not recorded | Core Ultra Series 2 |

| Cores | 8 | 16 |

| Boost clock | 5.00 GHz | 5.10 GHz |

| Socket | AMD Socket FP8 | Intel BGA 2049 |

| Architecture | Not recorded | Arrow Lake |

| Codename | Gorgon Point | Arrow Lake-H |

| Generation | Ryzen AI Embedded (Zen 5 / Zen 5c) | Ultra 7 (Arrow Lake-H) |

| Process node | 4 nm | 3 nm |

| Die size | 233 mm² | Not recorded |

| L1 cache | 80 KB (per core) | 192 KB (per core) |

| L2 cache | 1 MB (per core) | 3 MB (per core) |

| L3 cache | 8 MB | 24 MB (shared) |

| Memory bandwidth | 89.6 GB/s | 102.4 GB/s |

| PCIe | Gen 4, 16 Lanes (CPU only) | Gen 5, 20 Lanes (CPU only) |

| Integrated graphics | Radeon 880M | Arc Graphics 140T |

| Release date | 2026-03-08 | 2025-01-12 |

| Part number | unknown | SRQAN |

Where Each One Wins

The Intel Core Ultra 7 255H wins in workloads that stress prime-number finding, physics, floating-point math, encryption, multithread throughput, random string sorting, and single-thread responsiveness. Its Passmark scores in those tests are 303, 2254, 98796, 23395, 30703, 36058, and 4317, respectively.

The AMD Ryzen AI Embedded P164 wins in integer math, data compression, and extended instruction streams. Its Passmark scores are 87940, 327891, and 24193, respectively.

These are the only head-to-head win categories in the recorded data. The Intel part also has recorded Cinebench and Geekbench scores, but no AMD counterparts are present for direct comparison.