AMD Ryzen AI Embedded P164 vs Intel Core Ultra 5 245 Comparison

FAQ

Q: Which processor has the higher average benchmark score?

A: The AMD Ryzen AI Embedded P164 records an average benchmark score of 52901, while the Intel Core Ultra 5 245 records 48995. Despite this, the Intel part wins all 11 recorded head-to-head comparisons.

Q: What does the multithread score difference indicate?

A: The Intel Core Ultra 5 245 scores 38706 in PassMark multithread, which is 33.1% ahead of the AMD Ryzen AI Embedded P164's 25889. The Intel processor also holds a 90th percentile rank versus the AMD's 91st percentile, meaning both sit near the top of the database.

Q: How do the two compare in single-thread performance?

A: The Intel Core Ultra 5 245 leads with a single-thread score of 4394, which is 8.3% higher than the AMD Ryzen AI Embedded P164's 4029. This is the narrowest margin among all head-to-head tests.

Q: Which processor uses the smaller manufacturing process?

A: The Intel Core Ultra 5 245 uses a 3 nm process, while the AMD Ryzen AI Embedded P164 uses a 4 nm process. Both are fabricated by TSMC.

Q: What is the memory bandwidth difference?

A: The Intel Core Ultra 5 245 provides 102.4 GB/s of memory bandwidth, while the AMD Ryzen AI Embedded P164 provides 89.6 GB/s. Both support dual-channel memory configurations.

Q: Are both processors unlocked for overclocking?

A: No. Both the AMD Ryzen AI Embedded P164 and the Intel Core Ultra 5 245 have locked multipliers, indicating neither is multiplier unlocked.

Architecture Differences

The AMD Ryzen AI Embedded P164 uses the Gorgon Point codename with a generation labeled "Ryzen AI Embedded (Zen 5 / Zen 5c)". It is built on a 4 nm process at TSMC with a die size of 233 mm². The Intel Core Ultra 5 245 belongs to the Core Ultra Series 2, uses the Arrow Lake-S codename with the Arrow Lake architecture, and is fabricated on a 3 nm process at TSMC with a die size of 243 mm². The Intel die is slightly larger despite the smaller process, and it carries 17,800 million transistors, a figure not recorded for the AMD part.

The core configurations diverge sharply. The AMD processor has 8 cores and 16 threads, indicating simultaneous multithreading. The Intel processor has 14 cores and 14 threads, meaning it relies entirely on physical cores without hyperthreading. The AMD part's hybrid Zen 5 / Zen 5c layout suggests a mix of performance and efficiency cores, while the Intel part's Arrow Lake architecture employs its own core type separation.

Cache hierarchies also differ. The AMD Ryzen AI Embedded P164 provides 80 KB of L1 per core, 1 MB of L2 per core, and 8 MB of L3. The Intel Core Ultra 5 245 provides 192 KB of L1 per core, 3 MB of L2 per core, and 24 MB of shared L3. The Intel part has substantially more L3 cache, which likely contributes to its performance in data-heavy workloads.

Memory support shows another split. The AMD part supports both DDR5 and LPDDR5X, while the Intel part supports only DDR5. Both have dual-channel memory buses and ECC memory support. The Intel part offers PCIe Gen 5 with 20 lanes, whereas the AMD part offers PCIe Gen 4 with 16 lanes. The integrated graphics differ: the AMD uses Radeon 880M, and the Intel uses Arc Xe-LPG Graphics 64EU.

The Verdict

The benchmark data favors the Intel Core Ultra 5 245 in every recorded test. The Intel part wins all 11 head-to-head comparisons, with the largest margin in find prime numbers (80.5% ahead) and the smallest in integer math (3.6% ahead). The AMD Ryzen AI Embedded P164 holds a higher average benchmark score of 52901 versus 48995, but that average includes a wider set of workloads beyond the head-to-head list. The AMD part also holds a slightly higher percentile rank at 91 versus 90.

For users prioritizing raw throughput in the tested PassMark workloads, the Intel Core Ultra 5 245 is the clear choice. The data shows it leads in floating-point math by 53.7%, in physics by 52.9%, and in data encryption by 46.9%. For users who need a mobile processor with ECC memory support and a lower 28 W TDP, the AMD Ryzen AI Embedded P164 fits that profile, but the recorded performance metrics consistently favor Intel.

The AMD part's higher average score suggests it may excel in workloads not covered by the head-to-head tests, but within the measured PassMark suite, the Intel processor dominates. The Intel part also has a launch MSRP of $270, which is noted once here for reference. The AMD part has no recorded launch MSRP.

Specification Differences

The two processors differ in nearly every core specification. The AMD Ryzen AI Embedded P164 has 8 cores and 16 threads, while the Intel Core Ultra 5 245 has 14 cores and 14 threads. Base clocks are 2.00 GHz for AMD and 3.50 GHz for Intel. Boost clocks are 5.00 GHz for AMD and 5.10 GHz for Intel. The TDP is 28 W for AMD and 65 W for Intel.

The process nodes differ: 4 nm for AMD versus 3 nm for Intel, both at TSMC. The die sizes are 233 mm² for AMD and 243 mm² for Intel. The Intel part records 17,800 million transistors, while the AMD part does not list a transistor count. Cache sizes differ across all levels: L1 is 80 KB per core for AMD versus 192 KB per core for Intel; L2 is 1 MB per core for AMD versus 3 MB per core for Intel; L3 is 8 MB for AMD versus 24 MB shared for Intel.

Memory support differs: AMD supports DDR5 and LPDDR5X, while Intel supports only DDR5. Memory bandwidth is 89.6 GB/s for AMD and 102.4 GB/s for Intel. PCIe support differs: AMD uses Gen 4 with 16 lanes, while Intel uses Gen 5 with 20 lanes. The integrated graphics are Radeon 880M for AMD and Arc Xe-LPG Graphics 64EU for Intel. The AMD part uses AMD Socket FP8; the Intel part uses Intel Socket 1851. The AMD part has a release date of 2026-03-08, while the Intel part has a release date of 2025-01-06. The Intel part has part number SRVFE; the AMD part lists part number as unknown.

Head-to-Head Benchmarks

The Intel Core Ultra 5 245 wins all 11 recorded head-to-head benchmarks, so the analysis focuses on the magnitude of each win. The largest margin is in passmark_find_prime_numbers, where Intel scores 365 versus AMD's 71, a delta of -80.5% from the AMD perspective. This indicates the Intel processor is over five times faster in this specific primality test, which is highly sensitive to integer arithmetic efficiency and core count.

The second-largest margin appears in passmark_floating_point_math, where Intel scores 120548 versus AMD's 55799, a delta of -53.7%. This more than doubles the AMD score, suggesting the Intel part's floating-point execution units and higher clock speeds provide a decisive advantage. The physics test shows a similar pattern: Intel scores 2569 versus AMD's 1210, a delta of -52.9%.

Data encryption shows Intel at 30236 versus AMD's 16055, a delta of -46.9%. This workload benefits from AES-NI and other cryptographic instructions, where the Intel architecture appears stronger. The multithread test shows Intel at 38706 versus AMD's 25889, a delta of -33.1%, reflecting the Intel part's higher physical core count. Random string sorting shows Intel at 49140 versus AMD's 34801, a delta of -29.2%. Extended instructions show Intel at 33304 versus AMD's 24193, a delta of -27.4%.

Data compression shows Intel at 400942 versus AMD's 327891, a delta of -18.2%. This is a substantial absolute difference, with Intel exceeding 400 thousand points. Single-thread performance shows Intel at 4394 versus AMD's 4029, a delta of -8.3%, the smallest margin. Integer math shows Intel at 91187 versus AMD's 87940, a delta of -3.6%, the narrowest gap of all. The two single-thread entries (passmark_single_thread and passmark_singlethread) are duplicates with identical scores.

Where Each One Wins

The Intel Core Ultra 5 245 wins in every measured workload, so the differentiation comes from the size of each win and the workload characteristics. The Intel part shows its largest advantages in prime number finding (80.5% ahead), floating-point math (53.7% ahead), and physics (52.9% ahead). These are compute-intensive, multi-core workloads that respond to the Intel part's 14 physical cores and higher 5.10 GHz boost clock. For users running scientific simulations, physics calculations, or heavy floating-point code, the Intel processor delivers more than double the throughput in some cases.

The Intel part also dominates in data encryption (46.9% ahead) and extended instructions (27.4% ahead), indicating strong cryptographic and SIMD instruction processing. The multithread win of 33.1% confirms the Intel part's advantage in highly parallel workloads. The data compression win of 18.2% shows better handling of memory-bandwidth-heavy tasks, consistent with its 102.4 GB/s memory bandwidth versus the AMD's 89.6 GB/s.

The AMD Ryzen AI Embedded P164 has no benchmark wins in the head-to-head set, but it does hold a higher average benchmark score of 52901, meaning the database records it performing better across its full benchmark suite. The AMD part also holds a 91st percentile rank versus Intel's 90th, indicating it sits slightly higher among all CPUs. The AMD part's 28 W TDP makes it suited for low-power mobile applications, while its support for LPDDR5X memory and the Radeon 880M integrated graphics targets embedded and mobile designs. The AMD part's smaller 8-core, 16-thread configuration with a 5.00 GHz boost clock still delivers competitive single-thread performance, trailing Intel by only 8.3% in that test.

For workloads outside the PassMark head-to-head suite, the AMD part's higher average score suggests it may hold advantages in certain embedded or AI-specific tasks, particularly given its "Ryzen AI Embedded" designation. However, the recorded data does not include such tests. Within the measured benchmarks, the Intel Core Ultra 5 245 is the stronger processor across the board.