AMD Ryzen AI Embedded P164 vs Intel Core Ultra 5 235 Comparison

Head-to-Head Benchmarks

The recorded data shows a decisive sweep for the Intel Core Ultra 5 235 across all eleven shared benchmark tests. The Intel part wins every head-to-head comparison, with the AMD Ryzen AI Embedded P164 failing to claim a single victory. The most significant gap appears in passmark_find_prime_numbers, where Intel scores 371 against AMD's 71, a delta of -80.9% for the AMD processor. This is not a close contest by any measure; the Intel chip delivers more than five times the throughput in that specific workload.

The second-largest margin is in passmark_physics, where Intel posts 2570 versus AMD's 1210, a -52.9% delta. Floating point math shows a similar story: Intel reaches 117951 while AMD manages 55799, a -52.7% difference. Data encryption also favors Intel heavily, with 29293 against 16055, a -45.2% gap. Extended instructions follow at -26.1%, with Intel scoring 32752 and AMD scoring 24193.

Random string sorting shows Intel ahead by -28.9%, recording 48980 versus 34801. Multithread performance favors Intel by -31.5%, with scores of 37816 and 25889 respectively. Data compression shows Intel at 390711 against AMD's 327891, a -16.1% delta. Even single-thread performance, often a strength of smaller-core designs, goes to Intel: 4516 versus 4029, a -10.8% advantage.

The only test where the two processors effectively tie is passmark_integer_math. Intel scores 87948, AMD scores 87940, and the delta is exactly 0%. That near-identical result suggests the per-core integer execution capability of the two architectures is comparable, despite the overall performance gap elsewhere.

Looking at the broader database context, the AMD Ryzen AI Embedded P164 sits at the 91st percentile of all CPUs, while the Intel Core Ultra 5 235 sits at the 89th percentile. The AMD part's average benchmark score is 52901, which places it among rivals like the AMD Ryzen 5 9500F (52873, delta 0.1%), the Intel Xeon 634 (52974, delta -0.1%), the AMD EPYC 7313P (53206, delta -0.6%), and the AMD Ryzen 9 7900X (53288, delta -0.7%). The Intel part's average score is 46062, with rivals including the AMD Ryzen AI 9 HX 375 (46030, delta 0.1%), the Intel Core i9-13900HX (46098, delta -0.1%), the AMD EPYC 4364P (45970, delta 0.2%), and the AMD EPYC 7303 (45960, delta 0.2%).

The percentile discrepancy is interesting. The AMD chip ranks higher overall, yet it loses every head-to-head test to the Intel chip. This occurs because the AMD part's average score is boosted by its own benchmark suite, which differs from the shared head-to-head set. The Intel part's average score is dragged down by its inclusion of Cinebench results, which are not present for the AMD part. The data must be read carefully: the head-to-head table is the direct comparison, and it favors Intel unambiguously.

The Verdict

From the recorded data, the Intel Core Ultra 5 235 is the stronger processor for raw compute throughput. It wins all eleven shared benchmarks, often by very large margins. The AMD Ryzen AI Embedded P164 cannot match it in any measured head-to-head workload, despite holding a higher percentile rank in the full database. The Intel part's multithread score of 37816 versus 25889, its physics score of 2570 versus 1210, and its floating point score of 117951 versus 55799 all indicate a substantially higher performance ceiling.

The AMD part does have its own profile. It is an embedded mobile processor with a 28 W TDP, while the Intel part is a desktop processor with a 65 W TDP. The database shows the AMD chip at the 91st percentile against all CPUs, which reflects well on its efficiency class. The Intel chip, at the 89th percentile, still ranks highly but draws more power to achieve its results.

For workloads that appear in the head-to-head table, the Intel Core Ultra 5 235 is the clear choice. The data shows no scenario where the AMD part wins. For users prioritizing the lower power envelope of an embedded mobile design, the AMD part remains relevant, but not because it outperforms Intel in any shared test. The verdict from the measurements is straightforward: Intel dominates on performance, AMD competes on platform characteristics.

Architecture Differences

The two processors come from different architectural lineages. The AMD Ryzen AI Embedded P164 uses the Gorgon Point codename and belongs to the Ryzen AI Embedded generation built on Zen 5 and Zen 5c cores. It is fabricated on a 4 nm process at TSMC, with a die size of 233 mm². The Intel Core Ultra 5 235 uses Arrow Lake-S architecture, belongs to the Core Ultra Series 2 generation, and is built on a 3 nm process, also at TSMC. Its die size is 243 mm², and it contains 17,800 million transistors, a figure not listed for the AMD part.

Core counts differ significantly. The AMD chip has 8 cores and 16 threads, while the Intel chip has 14 cores and 14 threads. The Intel part therefore has more physical cores but no hyperthreading, while the AMD part uses simultaneous multithreading to reach 16 threads from 8 cores. Base clocks differ as well: AMD runs at 2.00 GHz, Intel at 3.40 GHz. Both boost to 5.00 GHz.

Cache hierarchies are distinct. The AMD part has 80 KB of L1 per core, 1 MB of L2 per core, and 8 MB of L3. The Intel part has 192 KB of L1 per core, 3 MB of L2 per core, and 24 MB of shared L3. The Intel chip's larger L3 cache aligns with its higher performance in cache-sensitive workloads like data compression and random string sorting.

Memory support also differs. The AMD processor supports DDR5 and LPDDR5X, while the Intel processor supports DDR5 only. Both use dual-channel memory buses. The Intel chip has a higher memory bandwidth at 102.4 GB/s versus 89.6 GB/s. ECC memory is supported on the AMD part but not on the Intel part. PCIe connectivity favors Intel: Gen 5 with 20 lanes versus Gen 4 with 16 lanes on the AMD side.

Integrated graphics differ. The AMD part uses a Radeon 880M, while the Intel part uses Arc Xe-LPG Graphics with 24 execution units. The AMD part's socket is AMD Socket FP8, while the Intel part uses Intel Socket 1851. The AMD part is classified as a mobile processor, while the Intel part is a desktop processor. Neither has an unlocked multiplier.

The release dates differ. The Intel part launched on January 6, 2025, while the AMD part launched on March 8, 2026. The AMD part's launch MSRP is not listed in the database; the Intel part carries a launch MSRP of $257.

FAQ

Q: Which processor has more cores?

A: The Intel Core Ultra 5 235 has 14 cores and 14 threads. The AMD Ryzen AI Embedded P164 has 8 cores and 16 threads.

Q: Does the AMD chip support ECC memory?

A: Yes, the AMD Ryzen AI Embedded P164 supports ECC memory. The Intel Core Ultra 5 235 does not.

Q: What is the single-thread performance difference?

A: The Intel Core Ultra 5 235 scores 4516 in passmark_single_thread, while the AMD Ryzen AI Embedded P164 scores 4029. The Intel part leads by 10.8%.

Q: Which processor has the higher memory bandwidth?

A: The Intel Core Ultra 5 235 has a memory bandwidth of 102.4 GB/s. The AMD Ryzen AI Embedded P164 has 89.6 GB/s.

Q: What process nodes are used?

A: The AMD Ryzen AI Embedded P164 is built on a 4 nm process. The Intel Core Ultra 5 235 is built on a 3 nm process. Both are fabricated by TSMC.

Q: Are both processors unlocked?

A: No. Neither the AMD Ryzen AI Embedded P164 nor the Intel Core Ultra 5 235 has an unlocked multiplier.

Where Each One Wins

The Intel Core Ultra 5 235 wins every measured head-to-head benchmark. Its largest advantages come in prime number finding, physics simulation, floating point math, and data encryption. These results indicate strong performance in computationally intensive, parallel-friendly workloads. The multithread score of 37816 and the physics score of 2570 both exceed the AMD part by more than 30%, and the floating point score nearly doubles it. For users running heavy number-crunching tasks, the Intel part is the clear winner per the recorded data.

The AMD Ryzen AI Embedded P164 does not win any shared benchmark, but it holds the higher percentile rank in the full database at 91 versus 89. Its average benchmark score of 52901 is also higher than the Intel part's 46062, driven by differences in the benchmark sets each part carries. The AMD part's advantages lie outside the head-to-head table: it supports ECC memory, uses a lower 28 W TDP, supports LPDDR5X memory, and is built for the mobile embedded segment. These are platform-level attributes, not performance wins.

For integer math, the two parts are effectively tied at 87940 versus 87948, a 0% delta. That single result shows parity in one narrow domain. Every other shared test favors Intel, often by double-digit margins.

Specification Differences

The two processors differ across nearly every major specification field. The AMD Ryzen AI Embedded P164 has 8 cores and 16 threads; the Intel Core Ultra 5 235 has 14 cores and 14 threads. Base clocks are 2.00 GHz for AMD and 3.40 GHz for Intel. Both boost to 5.00 GHz. TDP ratings are 28 W for AMD and 65 W for Intel.

Sockets differ: AMD uses Socket FP8, Intel uses Socket 1851. The AMD part uses the Gorgon Point codename with Zen 5 and Zen 5c cores, while the Intel part uses Arrow Lake-S. Process nodes are 4 nm for AMD and 3 nm for Intel, both at TSMC. The Intel part lists 17,800 million transistors and a 243 mm² die size; the AMD part lists a 233 mm² die size with no transistor count.

Cache configurations diverge: AMD has 80 KB L1 per core, 1 MB L2 per core, and 8 MB L3. Intel has 192 KB L1 per core, 3 MB L2 per core, and 24 MB shared L3. Memory support shows AMD with DDR5 and LPDDR5X, Intel with DDR5 only. Memory bandwidth is 89.6 GB/s for AMD and 102.4 GB/s for Intel. ECC memory is supported on AMD only.

PCIe connectivity favors Intel with Gen 5 and 20 lanes versus AMD's Gen 4 and 16 lanes. Integrated graphics are Radeon 880M on AMD and Arc Xe-LPG Graphics with 24 execution units on Intel. The AMD part is mobile, the Intel part is desktop. The AMD part launched on March 8, 2026; the Intel part launched on January 6, 2025. The Intel part lists a launch MSRP of $257; the AMD part lists none. Neither part has an unlocked multiplier.