AMD Ryzen AI Embedded P164 vs Intel Core Ultra 5 225H Comparison

The AMD Ryzen AI Embedded P164 and Intel Core Ultra 5 225H are both 28-watt mobile processors, but their benchmark results show distinctly different strengths. Across the 11 head-to-head PassMark tests, the Intel part wins seven, while the AMD part wins four. However, the margins tell a more nuanced story than the raw win count.

Head-to-Head Benchmarks

The AMD Ryzen AI Embedded P164 delivers its largest victory in integer math, scoring 87940 against the Intel Core Ultra 5 225H’s 68520, a 28.3% advantage. This is the single biggest delta in either direction. Data compression also favors AMD decisively: 327891 versus 283451, a 15.7% lead. Extended instructions show a 10.4% edge for AMD (24193 vs 21915), and random string sorting is close but still an AMD win at 34801 versus 33654, a 3.4% margin.

The Intel Core Ultra 5 225H counters with two notable wins. In floating-point math, it scores 86540 versus 55799, a 35.5% advantage. The same 35.5% delta appears in the physics test, where Intel scores 1877 and AMD scores 1210. Intel also dominates prime number finding with 220 versus 71, a massive 67.7% lead. Data encryption goes to Intel at 21428 versus 16055, a 25.1% margin. The multithread score favors Intel by 7.9% (28119 vs 25889), and single-thread performance gives Intel a smaller 5.4% edge (4258 vs 4029).

The overall average benchmark scores diverge sharply. The AMD part sits at 52901, placing it in the 91st percentile of all CPUs. The Intel part averages 31508, placing it in the 82nd percentile. This difference stems from the benchmark sets: AMD’s average includes only PassMark tests, while Intel’s average includes Cinebench and Geekbench results, which pull its average down relative to its PassMark performance.

Architecture Differences

The two processors use different silicon approaches. AMD’s Gorgon Point generation combines Zen 5 and Zen 5c cores on a 4 nm TSMC process with a 233 mm² die. The Intel Core Ultra 5 225H uses Arrow Lake-H architecture on a 3 nm TSMC process, part of the Core Ultra Series 2 family.

Core configurations differ significantly. The AMD chip has 8 cores and 16 threads, indicating simultaneous multithreading. Intel’s part has 14 cores but only 14 threads, meaning no hyperthreading. This explains why AMD’s multithread score stays competitive despite having fewer physical cores.

Cache hierarchies are built differently. AMD allocates 80 KB of L1 per core and 1 MB of L2 per core, with 8 MB of shared L3. Intel allocates 192 KB of L1 per core and 3 MB of L2 per core, with 18 MB of shared L3. The larger L3 cache on Intel likely contributes to its lead in memory-sensitive workloads.

Memory bandwidth also differs. Both support DDR5 and LPDDR5X with dual-channel buses, but Intel reaches 102.4 GB/s while AMD reaches 89.6 GB/s. AMD supports ECC memory; Intel does not. PCIe connectivity favors AMD with Gen 4 and 16 lanes, while Intel offers Gen 5 with 8 lanes.

Integrated graphics differ as well. AMD uses the Radeon 880M, while Intel uses Arc Graphics 130T. The socket and release dates diverge: AMD uses Socket FP8 and launched in March 2026, while Intel uses BGA 2049 and launched in January 2025.

The Verdict

The data indicates two different usage profiles. The AMD Ryzen AI Embedded P164 excels at workloads dominated by integer operations, data compression, and extended instruction sets. Its 91st percentile ranking and higher average benchmark score suggest it handles general-purpose compute tasks efficiently. The Intel Core Ultra 5 225H, despite its lower percentile ranking, delivers superior floating-point performance, physics simulation, and prime number calculations.

For single-threaded responsiveness, the Intel part holds a 5.4% edge. For heavily multithreaded tasks, Intel’s 14 physical cores give it a 7.9% advantage over AMD’s 8 cores with SMT. The Intel chip also leads in encryption throughput by 25.1%, which matters for security-sensitive applications.

The AMD chip is the better choice for integer-heavy workloads, compression pipelines, and scenarios where the 91st percentile overall standing reflects broad competence. The Intel chip is preferable for floating-point math, physics simulations, and tasks that scale across many physical cores.

Specification Differences

• Cores: AMD has 8; Intel has 14.
• Threads: AMD has 16; Intel has 14.
• Base Clock: AMD runs at 2.00 GHz; Intel at 1.70 GHz.
• Boost Clock: AMD reaches 5.00 GHz; Intel reaches 4.90 GHz.
• Process Node: AMD uses 4 nm; Intel uses 3 nm.
• Die Size: AMD measures 233 mm²; Intel has no recorded die size.
• L1 Cache: AMD provides 80 KB per core; Intel provides 192 KB per core.
• L2 Cache: AMD provides 1 MB per core; Intel provides 3 MB per core.
• L3 Cache: AMD has 8 MB; Intel has 18 MB shared.
• ECC Memory: AMD supports it; Intel does not.
• PCIe: AMD uses Gen 4 with 16 lanes; Intel uses Gen 5 with 8 lanes.
• Integrated Graphics: AMD uses Radeon 880M; Intel uses Arc Graphics 130T.
• Socket: AMD uses Socket FP8; Intel uses BGA 2049.
• Release Date: AMD launched in March 2026; Intel in January 2025.
• Part Number: AMD lists unknown; Intel lists SRQAM.

FAQ

Q: Which processor has better single-thread performance?

A: The Intel Core Ultra 5 225H scores 4258 in PassMark single-thread tests, 5.4% ahead of the AMD Ryzen AI Embedded P164’s 4029.

Q: How do the multithread scores compare?

A: Intel leads with 28119 versus AMD’s 25889, a 7.9% advantage. Intel’s 14 physical cores without SMT outperform AMD’s 8 cores with 16 threads in this metric.

Q: Which processor handles data compression better?

A: AMD wins decisively with 327891 versus Intel’s 283451, a 15.7% margin in the PassMark data compression test.

Q: What is the biggest performance gap between the two?

A: The largest delta is in prime number finding, where Intel scores 220 versus AMD’s 71, a 67.7% difference in Intel’s favor.

Q: Do both processors support ECC memory?

A: No. The AMD Ryzen AI Embedded P164 supports ECC memory, while the Intel Core Ultra 5 225H does not.

Q: Which processor has a higher overall benchmark percentile?

A: AMD ranks in the 91st percentile of all CPUs, while Intel ranks in the 82nd percentile. AMD’s average benchmark score is 52901 versus Intel’s 31508.

Where Each One Wins

The AMD Ryzen AI Embedded P164 wins in four head-to-head PassMark tests: data compression, extended instructions, integer math, and random string sorting. These results point to workloads involving data processing, integer arithmetic, and instruction-heavy algorithms. The 28.3% integer math lead and 15.7% compression advantage make it the stronger choice for database operations, file archiving, and general application logic.

The Intel Core Ultra 5 225H wins seven tests: data encryption, prime number finding, floating-point math, multithread, physics, and both single-thread variants. The 35.5% floating-point lead and 35.5% physics advantage indicate superiority in scientific computing, 3D rendering, and simulation tasks. The 25.1% encryption edge supports secure communication workloads, and the 7.9% multithread win favors parallel processing with many physical cores.

The decision comes down to workload type. AMD’s architecture prioritizes integer throughput and compression efficiency. Intel’s architecture prioritizes floating-point throughput and core count. The recorded data shows no overall winner, only different optimizations for different tasks.