AMD Ryzen AI Embedded P164 vs Intel Core Ultra X7 368H Comparison

The Verdict

The recorded benchmark data presents a clear split: the Intel Core Ultra X7 368H is the dominant performer for heavily parallel workloads, taking 8 of 11 head-to-head tests. The AMD Ryzen AI Embedded P164 counters with a slender single-thread win and a decisive advantage in data compression. For users prioritizing multi-core throughput, floating-point math, or encryption, the Intel part is the obvious choice from the data. For single-thread responsiveness or compression-heavy tasks, the AMD part holds the edge.

Looking at the aggregate scores, the AMD part posts an average benchmark score of 52,901, placing it at the 91st percentile of all CPUs in the database. Its nearest rival, the AMD Ryzen 5 9500F, scores 52,873, a delta of 0.1%. The Intel Core Ultra X7 368H averages 40,518, ranking at the 87th percentile. Its nearest rival, the Intel Core 7 253PE, scores 40,557, a delta of -0.1%. The AMD part is therefore the higher-ranked chip overall, although the Intel part holds a commanding lead in many individual tests.

The Intel part wins the PassMark multi-thread test with a score of 32,956 versus 25,889 for AMD, a 21.4% gap. The Intel part also wins physics simulation by 57.6%, floating-point math by 47.2%, and encryption by 36.4%. The AMD part wins single-thread by a hair, 4,029 versus 4,005, and data compression by 4.8%.

Where Each One Wins

The Intel Core Ultra X7 368H is the clear leader in computational throughput. In the PassMark multi-thread benchmark, it scores 32,956, which is 21.4% ahead of the AMD part's 25,889. The physics test shows an even larger gap: Intel scores 2,857 versus 1,210, a 57.6% advantage. Floating-point math also heavily favors Intel, with 105,681 versus 55,799, a 47.2% lead. Prime number finding is the most lopsided test: Intel scores 321 versus 71, a 77.9% margin. Data encryption favors Intel at 25,228 versus 16,055, a 36.4% lead. Extended instruction performance favors Intel by 5.8%, integer math is effectively tied at 0.2% in Intel's favor, and random string sorting favors Intel by 8.6%.

The AMD Ryzen AI Embedded P164 wins in two specific areas. Single-thread performance is 4,029 versus 4,005, a 0.6% margin. Data compression is 327,891 versus 312,927, a 4.8% margin. These are narrow wins, but they are consistent across the two recorded single-thread entries. The AMD part also carries a higher overall percentile rank, 91 versus 87, and a higher average benchmark score, 52,901 versus 40,518.

Architecture Differences

The AMD Ryzen AI Embedded P164 is built on a 4 nm TSMC process and uses the Gorgon Point codename. It belongs to the Ryzen AI Embedded generation based on Zen 5 and Zen 5c cores. It has 8 cores and 16 threads, with a base clock of 2.00 GHz and a boost clock of 5.00 GHz. The thermal design power is 28 W. The cache layout includes 80 KB of L1 per core, 1 MB of L2 per core, and 8 MB of L3 cache. The die size is 233 mm². Memory support includes DDR5 and LPDDR5X over a dual-channel bus, with 89.6 GB/s of bandwidth and ECC support. The integrated graphics are Radeon 880M. The socket is AMD Socket FP8, and the CPU provides Gen 4 PCIe with 16 lanes.

The Intel Core Ultra X7 368H uses a 3 nm Intel process and the Panther Lake codename, specifically the Panther Lake-H variant in the Core Ultra Series 3. It has 16 cores and 16 threads, meaning it does not use simultaneous multithreading. Base clock is 2.00 GHz and boost clock is 5.00 GHz, identical to the AMD part. The thermal design power is 25 W, slightly lower than AMD's 28 W. The cache is significantly larger: 192 KB of L1 per core, 2.5 MB of L2 per core, and 18 MB of shared L3. Memory support is limited to LPDDR5X over a dual-channel bus, with 153.6 GB/s of bandwidth, notably higher than the AMD part, but without ECC. The integrated graphics are Arc B390. The socket is Intel BGA 2540, and the CPU provides Gen 5 PCIe with 4 lanes.

The core count difference is the most obvious architectural factor. The Intel part has twice the cores, 16 versus 8, which aligns with its strong multi-thread performance. The AMD part uses simultaneous multithreading to reach 16 threads from 8 cores, while the Intel part reaches 16 threads from 16 physical cores. The Intel part also has more than double the L3 cache, 18 MB versus 8 MB, and higher memory bandwidth, 153.6 GB/s versus 89.6 GB/s. The AMD part counters with ECC memory support and a lower power draw per core, though the total TDP is only 3 W higher.

FAQ

Q: Which processor has better single-thread performance?

A: The AMD Ryzen AI Embedded P164 scores 4,029 in the PassMark single-thread test, while the Intel Core Ultra X7 368H scores 4,005. The AMD part wins by 0.6%.

Q: How large is the multi-thread performance gap?

A: The Intel Core Ultra X7 368H scores 32,956 in the PassMark multi-thread test, which is 21.4% higher than the AMD part's 25,889.

Q: Which processor supports ECC memory?

A: The AMD Ryzen AI Embedded P164 supports ECC memory. The Intel Core Ultra X7 368H does not.

Q: What are the core and thread counts for each processor?

A: The AMD part has 8 cores and 16 threads. The Intel part has 16 cores and 16 threads.

Q: How do the processors compare in memory bandwidth?

A: The Intel Core Ultra X7 368H provides 153.6 GB/s of memory bandwidth, while the AMD Ryzen AI Embedded P164 provides 89.6 GB/s. Intel leads by a substantial margin.

Q: Which processor has the higher overall benchmark ranking?

A: The AMD Ryzen AI Embedded P164 ranks at the 91st percentile of all CPUs, with an average benchmark score of 52,901. The Intel part ranks at the 87th percentile, with an average score of 40,518.

Head-to-Head Benchmarks

The PassMark data compression test is the AMD part's strongest win. It scores 327,891 against Intel's 312,927, a 4.8% margin. This is the only multi-threaded test where AMD takes the lead, aside from the narrow single-thread result.

Single-thread performance is nearly identical. The AMD part scores 4,029 in both recorded single-thread entries, while the Intel part scores 4,005. The 0.6% margin is small but consistent across both entries. This suggests the AMD part holds a slight edge in lightly threaded workloads.

The Intel part dominates the remaining tests. In floating-point math, Intel scores 105,681 versus 55,799, a 47.2% lead. In physics, Intel scores 2,857 versus 1,210, a 57.6% lead. In prime number finding, Intel scores 321 versus 71, a 77.9% lead. In encryption, Intel scores 25,228 versus 16,055, a 36.4% lead. In extended instructions, Intel scores 25,669 versus 24,193, a 5.8% lead. In random string sorting, Intel scores 38,093 versus 34,801, an 8.6% lead. Integer math is the closest contest: Intel scores 88,083 versus 87,940, a 0.2% margin.

The multi-thread test confirms the overall trend. Intel's 32,956 score is 21.4% higher than AMD's 25,889. The Intel part's 16 physical cores and larger L3 cache appear to support this result, although the recorded data does not explicitly link the two. The AMD part's 8 cores with simultaneous multithreading cannot close the gap in most parallel workloads.

The aggregate benchmark scores place the AMD part ahead despite losing most head-to-head tests. The AMD part averages 52,901 across all benchmarks, while the Intel part averages 40,518. This discrepancy suggests the AMD part performs relatively better in benchmarks outside the PassMark head-to-head set, although the database does not specify which tests those are. The percentile ranks reflect this: AMD at 91, Intel at 87.

The Intel part's nearest rival, the Intel Core 7 253PE, averages 40,557, a 0.1% difference. The AMD part's nearest rival, the AMD Ryzen 5 9500F, averages 52,873, a 0.1% difference. Both parts sit at the top of their respective clusters, with no rival within a full percentage point. The AMD part also sits near the AMD EPYC 7313P and Ryzen 9 7900X, which average 53,206 and 53,288 respectively, deltas of -0.6% and -0.7%. The Intel part sits near the Intel Core 5 223PE, AMD Ryzen 9 7940H, and Intel Xeon 6507P, with deltas of -0.2%, 0.2%, and 0.2% respectively.

The recorded data shows a processor pairing where architectural choices strongly predict performance outcomes. The Intel part's extra cores, larger cache, and higher memory bandwidth translate directly into multi-thread wins. The AMD part's higher average score and percentile rank show that single-thread and compression performance, plus other benchmarks outside the head-to-head set, can offset those losses in the aggregate.