AMD Ryzen AI Max 390 vs Intel Core Ultra 5 338H Comparison

The AMD Ryzen AI Max 390 and Intel Core Ultra 5 338H are both 12-core mobile processors, but the benchmark data positions them in different performance classes. The Ryzen AI Max 390 wins 15 of the 17 recorded head-to-head tests, while the Core Ultra 5 338H takes only 2. The average benchmark score for the AMD part is 56,273, placing it at the 91st percentile of all CPUs, whereas the Intel chip averages 33,989 and sits at the 84th percentile. This gap is substantial, but the specific tests each processor wins reveal distinct strengths that go beyond a simple overall performance ranking.

Where Each One Wins

The Ryzen AI Max 390 dominates in nearly every multi-threaded and compute-heavy workload. In Cinebench R23 multi-core, it scores 36,064 against 16,331 for the Intel part, a 120.8% advantage. Cinebench R20 multi-core shows a 48.3% lead (15,146 vs 10,213), and Cinebench R15 multi-core shows a 45.2% lead (3,635 vs 2,504). The PassMark integer math test shows a 125.6% difference (146,519 vs 64,934), and data compression is 76.2% higher (487,145 vs 276,539). These results indicate the AMD processor is built for sustained parallel workloads, such as video rendering, code compilation, and scientific simulations.

The Intel Core Ultra 5 338H wins only in the PassMark single-thread test and its duplicate, the PassMark singlethread test. Its score of 4,180 beats the AMD's 4,028 by 3.6%. This is a narrow win, and the Cinebench single-core tests tell a different story. In Cinebench R23 single-core, the AMD scores 5,091 versus 2,044 for Intel, a 149.1% lead. The R20 single-core test shows a 48.4% lead (2,138 vs 1,441), and R15 single-core shows a 68.2% lead (513 vs 305). The single PassMark win suggests the Intel part has a specific advantage in that particular workload, but the Cinebench results indicate the AMD architecture is faster in most single-threaded scenarios as well.

Architecture Differences

The two processors use fundamentally different designs. The AMD Ryzen AI Max 390 uses the Zen 5 architecture on a 4 nm TSMC process, with the codename Strix Halo. It has 12 cores and 24 threads, meaning it supports simultaneous multithreading. The Intel Core Ultra 5 338H uses the Panther Lake architecture on a 3 nm Intel process, with 12 cores and 12 threads, which means no multithreading per core. This thread count difference is a primary reason for the large multi-core benchmark gap.

Cache configurations also differ sharply. The AMD part has 80 KB of L1 per core, 1 MB of L2 per core, and 64 MB of shared L3 cache. The Intel part has 192 KB of L1 per core, 2.5 MB of L2 per core, but only 18 MB of shared L3. The AMD's 64 MB L3 cache is more than three times larger, which helps with data-heavy workloads and explains its strong showing in the PassMark data compression test.

Memory architecture adds another layer of separation. Both support LPDDR5X, but the AMD runs a quad-channel memory bus with 256.0 GB/s bandwidth, while the Intel runs a dual-channel bus with 136.5 GB/s. The AMD also supports ECC memory, while the Intel does not. PCIe connectivity differs as well: the AMD offers Gen 4 with 16 lanes, while the Intel offers Gen 5 with only 4 lanes. The integrated graphics are the Radeon 8050S on the AMD side and the Arc B370 on the Intel side, though the database does not provide graphics benchmarks for comparison.

FAQ

Q: Why does the AMD Ryzen AI Max 390 have such a large lead in multi-core benchmarks?

A: The AMD processor has 24 threads versus 12 threads on the Intel part, and it also has a 64 MB shared L3 cache compared to 18 MB on the Intel. The combination of simultaneous multithreading and a larger cache allows it to handle more parallel work and keep more data on-chip.

Q: The Intel Core Ultra 5 338H wins the PassMark single-thread test. Does that mean it is faster in single-core tasks?

A: The PassMark single-thread score of 4,180 is 3.6% higher than the AMD's 4,028. However, the Cinebench R23 single-core test shows the AMD at 5,091 versus 2,044 for Intel, a 149.1% difference. The PassMark result is an isolated win, while the Cinebench tests indicate the AMD has the higher single-core capability in those workloads.

Q: What is the difference in memory bandwidth between the two processors?

A: The AMD Ryzen AI Max 390 uses a quad-channel memory bus with 256.0 GB/s bandwidth. The Intel Core Ultra 5 338H uses a dual-channel bus with 136.5 GB/s. The AMD has nearly double the memory bandwidth, which benefits tasks that move large data sets.

Q: Do both processors support ECC memory?

A: No. The AMD Ryzen AI Max 390 supports ECC memory, while the Intel Core Ultra 5 338H does not. This makes the AMD part more suitable for error-sensitive computing environments.

Q: How do the cache sizes compare?

A: The AMD has 80 KB of L1 per core, 1 MB of L2 per core, and 64 MB of shared L3. The Intel has 192 KB of L1 per core, 2.5 MB of L2 per core, and 18 MB of shared L3. The AMD has a much larger L3 cache, while the Intel has larger L1 and L2 caches per core.

Q: What is the difference in process node and foundry?

A: The AMD uses a 4 nm process from TSMC. The Intel uses a 3 nm process from Intel. The Intel process is smaller, but the benchmark results do not show a clear performance advantage from this difference.

Specification Differences

The two processors differ on several key specifications. The AMD Ryzen AI Max 390 has a base clock of 3.20 GHz and a boost clock of 5.00 GHz, while the Intel Core Ultra 5 338H has a base clock of 1.90 GHz and a boost clock of 4.70 GHz. The AMD has a 55 W TDP, while the Intel has a 25 W TDP, indicating the AMD is designed for higher power consumption and likely requires more robust cooling.

The AMD uses AMD Socket FP11, while the Intel uses Intel BGA 2540. The AMD has a die size of 2x 70.6 mm², while the Intel die size is not recorded. The AMD supports PCIe Gen 4 with 16 lanes, while the Intel supports PCIe Gen 5 with 4 lanes. The AMD has a quad-channel memory bus, the Intel a dual-channel bus. The AMD supports ECC, the Intel does not. The AMD's part number is 100-000001423, and the Intel's is SA4REQ9EW. The AMD release date is 2025-01-05, while the Intel release date is 2026-01-04. Both are active production parts, both are mobile segments, and neither has an unlocked multiplier.

Head-to-Head Benchmarks

The largest win for the AMD Ryzen AI Max 390 comes in Cinebench R23 single-core, where it scores 5,091 against 2,044 for the Intel, a 149.1% delta. This is a striking result for a single-threaded test, as it suggests the Zen 5 core design provides a major instruction-per-clock advantage in this workload. The Cinebench R23 multi-core test shows a 120.8% delta (36,064 vs 16,331), which is the second-largest gap. PassMark integer math shows a 125.6% delta (146,519 vs 64,934), nearly matching the R23 multi-core gap. These three tests alone establish the AMD as the clear leader in both single-threaded and multi-threaded Cinebench workloads, as well as in general integer arithmetic.

Other notable AMD wins include PassMark data compression at 76.2% (487,145 vs 276,539), PassMark extended instructions at 62% (38,716 vs 23,906), and PassMark random string sorting at 55.8% (53,113 vs 34,082). Cinebench R20 multi-core shows a 48.3% delta (15,146 vs 10,213), and R15 multi-core shows a 45.2% delta (3,635 vs 2,504). PassMark multithread shows a 45.3% delta (41,737 vs 28,717). The AMD also wins PassMark data encryption by 17.5% (25,097 vs 21,367), floating point math by 7.8% (90,594 vs 84,067), find prime numbers by 3.9% (316 vs 304), and physics by 2.4% (2,761 vs 2,697).

The Intel Core Ultra 5 338H wins only the PassMark single-thread test, scoring 4,180 versus 4,028 for the AMD, a 3.6% delta. The duplicate PassMark singlethread test shows the same result. These are the only two wins for the Intel part, and they are narrow. The AMD's lead in the Cinebench single-core tests, which show a 149.1% delta, far outweighs the Intel's PassMark single-thread win. The data clearly indicates that the AMD processor is faster in almost every measured workload, with the Intel's only edge being a modest lead in one specific synthetic single-thread test.

The Verdict

The benchmark data points to the AMD Ryzen AI Max 390 as the superior processor for compute-intensive tasks. Its wins in Cinebench R23 multi-core (120.8% ahead), PassMark integer math (125.6% ahead), and data compression (76.2% ahead) show a consistent pattern of dominance in parallel and data-processing workloads. The 24 threads, 64 MB L3 cache, and 256.0 GB/s memory bandwidth provide the resources needed for these tasks. The Intel Core Ultra 5 338H, with 12 threads, 18 MB L3, and 136.5 GB/s bandwidth, is not competitive in these areas.

For users focused on single-threaded performance, the Cinebench results also favor the AMD, with a 149.1% lead in R23 single-core. The Intel's PassMark single-thread win (3.6%) is an outlier that does not appear in the Cinebench tests. The Intel part does have a lower TDP of 25 W versus 55 W, which may make it appealing for power-constrained designs, but the database does not include battery life or thermal measurements to verify a real-world advantage. The Intel part also uses a 3 nm process from Intel, while the AMD uses a 4 nm process from TSMC, yet the smaller process node does not translate into better benchmark scores. The AMD Ryzen AI Max 390 is the appropriate choice for workloads that demand high multi-core throughput, large cache capacity, and high memory bandwidth. The Intel Core Ultra 5 338H offers a single PassMark single-thread win and a lower TDP, but the recorded data shows it trailing in 15 of 17 tests.