AMD EPYC 9554P vs Intel Core Ultra 5 135U Comparison

The Intel Core Ultra 5 135U and the AMD EPYC 9554P occupy opposite ends of the CPU landscape, and the recorded data reflects that divide in every measurement. The Core Ultra 5 135U is a Meteor Lake mobile processor with a 15 W TDP, 12 cores and 14 threads, while the EPYC 9554P is a Genoa server part with 64 cores, 128 threads and a 360 W TDP. In the six head-to-head benchmarks in the database, the EPYC wins all six, often by margins exceeding 80 percent. Yet the two chips land in nearly the same percentile against the full CPU population, at 76 for the Core Ultra 5 135U and 75 for the EPYC 9554P, and their average benchmark scores sit close together at 22618 and 21899 respectively. That apparent parity is an artifact of different test suites: the mobile chip carries a broad set of Passmark results, the server chip a pair of Geekbench entries, and the average score blends them. The direct comparisons tell the real story.

Where Each One Wins

Strictly on the recorded head-to-head results, the EPYC 9554P wins everywhere it was measured. It takes Cinebench R15 multi-core 9020 to 1634, Cinebench R20 multi-core 37585 to 5934, and Cinebench R23 multi-core 89490 to 9315, an 89.6 percent gap that is the largest in the dataset. The single-core results follow the same direction: R15 single-core goes to the EPYC 1273 to 228, R20 single-core 5305 to 837, and R23 single-core 12633 to 1563.

The Core Ultra 5 135U records wins only in the tests where it stands alone, the Passmark suite, where the database holds no EPYC 9554P entries. There it posts a single-thread score of 3343, an integer math result of 56864, a floating point math result of 41232, a data compression score of 196712, and a physics score of 1122. These numbers place it in the 76th percentile of all CPUs in the database and put it in the company of rivals like the Intel Core i5-12500H, which matches its average score within 0 percent delta, the Intel Core i7-1270P at a 0.5 percent delta, the Intel Core i5-13500H at 0.7 percent, and the AMD Ryzen 5 7535U at -0.8 percent. In other words, the Ultra 5 135U performs like a mainstream mobile chip, which is exactly what it is.

FAQ

Q: Does the Core Ultra 5 135U win any head-to-head benchmark against the EPYC 9554P?

A: No. The database records six direct comparisons, all Cinebench tests, and the EPYC 9554P wins all six. Its smallest margin is 81.9 percent in Cinebench R15 multi-core and its largest is 89.6 percent in Cinebench R23 multi-core.

Q: Why do the two chips have similar percentile rankings if one dominates the head-to-head results?

A: The percentile fields compare each CPU against the entire database population, not against each other. The Core Ultra 5 135U sits in the 76th percentile and the EPYC 9554P in the 75th, nearly identical standings, but each ranking is built from a different mix of tests: Passmark entries for the mobile chip and Cinebench plus Geekbench entries for the server chip.

Q: How do their average benchmark scores compare?

A: The Core Ultra 5 135U averages 22618 and the EPYC 9554P averages 21899. The mobile chip's average is actually the higher of the two, again because the averages are computed over different test suites rather than shared tests.

Q: Which chip has the stronger single-core performance in the shared tests?

A: The EPYC 9554P, by a wide margin. In Cinebench R23 single-core it scores 12633 against 1563, an 87.6 percent advantage, and in Cinebench R20 single-core it leads 5305 to 837.

Q: Do both processors support ECC memory?

A: No. The EPYC 9554P supports ECC memory, consistent with its server market segment. The Core Ultra 5 135U does not.

Q: Which processor is more power-efficient on paper?

A: The recorded TDP figures are 15 W for the Core Ultra 5 135U and 360 W for the EPYC 9554P. The mobile part is designed around a dramatically lower power envelope, which is the core reason the performance gap in the shared benchmarks is so large.

Head-to-Head Benchmarks

Walking through the six shared tests in order, the pattern is uniform but worth examining test by test.

Cinebench R15 multi-core: the EPYC 9554P scores 9020 against 1634 for the Core Ultra 5 135U, an 81.9 percent gap. This is the closest result of the series, which still means the server chip delivers more than five times the throughput. With 64 cores and 128 threads versus 12 cores and 14 threads, the raw thread count alone explains most of the distance.

Cinebench R15 single-core: 1273 to 228 in favor of the EPYC, an 82.1 percent gap. This is the more notable result, because single-core tests neutralize the core-count advantage. The EPYC's higher base and boost clocks in the recorded data, 3.10 GHz base and 3.75 GHz boost against 1600 MHz base and 4.40 GHz boost for the mobile part, do not by themselves explain an 82 percent lead, indicating the Zen 4 server cores simply sustain far more work per thread under load than the low-power Meteor Lake cores in this configuration.

Cinebench R20 widens the gap on both fronts. Multi-core goes to the EPYC 37585 to 5934, an 84.2 percent margin, and single-core 5305 to 837, also 84.2 percent. The identical deltas across the two R20 tests suggest the gap is structural rather than workload-specific.

Cinebench R23 produces the largest discrepancies in the dataset. Multi-core lands at 89490 for the EPYC versus 9315 for the Core Ultra 5 135U, an 89.6 percent gap, meaning the server chip delivers nearly ten times the multi-threaded render throughput. Single-core reaches 12633 versus 1563, an 87.6 percent gap.

For context on the EPYC side, its nearest rivals in the database include the AMD EPYC 9534 at a 0 percent delta, the Intel Core i7-11700 also at 0 percent, the AMD Ryzen 5 PRO 6650U at 0.1 percent, and the Intel Core Ultra 5 236V at -0.2 percent. Its Geekbench results of 17978 multi-core and 1911 single-core round out the server chip's recorded profile.

Specification Differences

The two processors differ on nearly every specification field in the database.

• Cores and threads: 12 cores and 14 threads for the Core Ultra 5 135U versus 64 cores and 128 threads for the EPYC 9554P.
• Clocks: the mobile part records a 1600.00 base clock and a 4.40 boost clock, while the EPYC records 3.10 base and 3.75 boost. The Ultra has the higher boost figure; the EPYC has the higher base.
• TDP: 15 W versus 360 W.
• Socket: Intel BGA 2049 versus AMD Socket SP5, a soldered mobile interface against a server socket.
• Cache: the Ultra carries 112 KB of L1 per core, 2 MB of L2 per core and 12 MB of shared L3. The EPYC carries 64 KB of L1 per core, 1 MB of L2 per core and 256 MB of shared L3. Per-core cache favors the mobile chip; total cache massively favors the server part.
• Memory: the Ultra supports DDR5 over a dual-channel bus with 89.6 GB/s of bandwidth and no ECC. The EPYC supports DDR5 over a twelve-channel bus with 460.8 GB/s of bandwidth and ECC support.
• PCIe: Gen 4 with 12 CPU lanes on the Ultra versus Gen 5 with 128 CPU lanes on the EPYC.
• Graphics: the Ultra integrates Arc Xe-LPG 64EU graphics; the EPYC has none recorded.
• Market segment and status: mobile versus server/workstation, both active.
• Release dates: the EPYC entered the database with a 2022-11-09 release date, the Ultra with 2023-12-13.
• Launch MSRP: $332 for the Core Ultra 5 135U and $7104 for the EPYC 9554P.
• Multiplier: neither part is unlocked.
• Transistors and die size: recorded for the EPYC as 52,560 million transistors on an 8x 72 mm² die configuration; not recorded for the Ultra.

Architecture Differences

The Core Ultra 5 135U is built on Intel's Meteor Lake architecture on a 7 nm process at Intel's own foundry. Its 12 cores and 14 threads reflect a hybrid mobile design, with a thread count only slightly above the core count, and its Arc Xe-LPG 64EU integrated graphics are part of the same package. The part number is SRN6C.

The EPYC 9554P uses AMD's Zen 4 architecture under the Genoa codename, manufactured on a 5 nm TSMC process. It packs 52,560 million transistors across a chiplet layout recorded as 8x 72 mm², with 256 MB of shared L3 cache spread across that structure. Its 128 threads across 64 cores, twelve-channel DDR5 with ECC, and 128 lanes of Gen 5 PCIe are all hallmarks of a platform built for server workloads rather than client mobility. The part number is 100-100000804.

The core-count asymmetry is the defining architectural difference. The Meteor Lake chip allocates silicon to efficiency, integrated graphics and a 15 W envelope. The Genoa part allocates silicon almost entirely to parallel compute and memory bandwidth, with no integrated graphics at all.

The Verdict

The data frames this as a category mismatch rather than a contest. The EPYC 9554P wins every shared benchmark by margins between 81.9 and 89.6 percent, including single-core tests where its core-count advantage is irrelevant, so any workload resembling the recorded Cinebench suite belongs on the server chip. It is also the only one of the two with ECC memory, twelve-channel memory bandwidth of 460.8 GB/s, and 128 Gen 5 PCIe lanes.

The Core Ultra 5 135U is the pick where its recorded strengths align with the need: a 15 W TDP, integrated Arc Xe-LPG 64EU graphics, dual-channel DDR5, and Passmark results placing it alongside mainstream mobile rivals like the Core i5-12500H and Ryzen 5 7535U within a fraction of a percent. Its 76th percentile ranking and 22618 average score describe a capable thin-and-light processor, not a competitor to a 64-core Genoa part. Choose the EPYC for throughput and platform bandwidth; choose the Ultra for a power-constrained mobile system where integrated graphics and efficiency dominate the requirements.