AMD EPYC 9334 vs Intel Core Ultra 5 115U Comparison

At first glance this matchup looks like a mismatch, and the recorded data confirms it: the AMD EPYC 9334, a 32-core server processor, against the Intel Core Ultra 5 115U, a 15-watt mobile chip. Yet both parts sit at the 70th percentile versus all CPUs in the database, a reminder that percentile placement is computed within each part's competitive context, not across segments. The EPYC 9334 sweeps all six shared Cinebench tests with margins that hover around 408 percent, and the interesting question is not who wins but what the pattern of those numbers reveals about two fundamentally different design philosophies.

Where Each One Wins

The EPYC 9334 wins everywhere the two chips overlap. The shared benchmark set consists of six Cinebench tests spanning R15, R20 and R23, in both single-core and multi-core variants, and the server processor takes all six. There is no benchmark in the recorded data where the Core Ultra 5 115U comes out ahead.

The scale of the sweep is remarkably uniform. Multi-core margins run from 407.7 percent in R20 and R23 to 407.8 percent in R15, while single-core margins sit between 407.8 and 409.1 percent. That uniformity is itself informative: the EPYC is not merely exploiting its core count in threaded tests, it is roughly five times faster even when only one thread does the work. Whatever the workload, the database shows no scenario in the shared test set where the mobile chip closes the gap.

Where the Core Ultra 5 115U wins is in territory the EPYC simply cannot enter. It has integrated graphics, the Arc Xe-LPG 48EU, while the EPYC 9334 lists none. Its thermal envelope is 15 watts against 210 watts, its socket is a soldered Intel BGA 2049 rather than the AMD Socket SP5, and it carries Passmark coverage the EPYC lacks entirely, including single-thread scores of 3057 and a multithread result of 13024. Its victories are categorical, not numerical: portability, efficiency by design, and graphical output.

Architecture Differences

These two parts come from opposite ends of the industry. The EPYC 9334 is a Zen 4 design, codenamed Genoa, built on TSMC's 5 nm process. It packs 52,560 million transistors across a chiplet layout described in the database as 4x 72 mm². The Core Ultra 5 115U is Intel's Meteor Lake architecture on a 7 nm Intel process, with transistor count and die size unlisted in the database.

The core topology gap is enormous. The EPYC pairs 32 cores with 64 threads, a two-to-one ratio enabled by simultaneous multithreading. The Core Ultra 5 115U offers 8 cores and 10 threads, an asymmetric ratio that hints at a hybrid arrangement, though the database does not break the cores down further. Clock behavior inverts the resource picture: the EPYC runs a 2.70 GHz base and 3.90 GHz boost, while the Intel chip boosts higher at 4.20 GHz, trading frequency for its far smaller core count.

Cache architecture also diverges sharply. The EPYC carries 64 KB of L1 and 1 MB of L2 per core, plus a 128 MB shared L3. Per-core figures, the Intel chip actually lists more L1 at 112 KB and more L2 at 2 MB, but its shared L3 is only 10 MB. The server part's cache pool is more than twelve times larger in aggregate L3 terms, a scale difference that matches its throughput ambitions.

Memory and platform connectivity separate the two even further. The EPYC supports twelve-channel DDR5 with 460.8 GB/s of bandwidth and ECC memory, plus PCIe Gen 5 with 128 CPU-only lanes. The Core Ultra 5 115U is dual-channel DDR5 with 89.6 GB/s of bandwidth, no ECC, and PCIe Gen 4 with 12 CPU-only lanes. For memory-bound or I/O-heavy server workloads, that platform gap is as decisive as the core count.

Head-to-Head Benchmarks

The head-to-head record is a clean 6-0 sweep for the EPYC 9334. In Cinebench R15 multi-core the EPYC scores 5555 against 1094, a 407.8 percent margin. R15 single-core reads 784 versus 154, a 409.1 percent gap, the largest in the dataset.

Cinebench R20 follows the same script: 23146 against 4559 in multi-core (407.7 percent) and 3267 against 643 in single-core (408.1 percent). R23 pushes absolute numbers higher but keeps ratios identical, with the EPYC posting 55110 versus 10855 in multi-core (407.7 percent) and 7780 versus 1532 in single-core (407.8 percent).

What should catch an analyst's eye is how little the margin moves between single-core and multi-core tests. If the EPYC's advantage came purely from its 64 threads, the multi-core gap would dwarf the single-core gap. Instead both sit near 408 percent, which suggests the per-thread performance of the Zen 4 server cores, running at up to 3.90 GHz with generous cache behind them, already exceeds what the Meteor Lake cores deliver in these rendering tests. The core count merely sustains the ratio across scaled workloads.

The Core Ultra 5 115U's own Passmark data adds context the head-to-head lacks. Its single-thread score of 3057 and multithread score of 13024 place it in a cluster of thin-and-light rivals: the Intel Core i5-1235U sits 0.1 percent above it on average score, the AMD Ryzen 3 7335U is 0.4 percent ahead, the AMD Ryzen 3 8440U trails it by 0.5 percent, and the Intel Core i3-12100E leads by 0.6 percent. That is a tightly packed mobile field, and the 115U holds its own within it, with an average benchmark score of 16753 that actually edges the EPYC's 15940 average, a quirk of averaging across different test mixes rather than evidence of real-world parity.

The EPYC's own rival set is equally instructive. Its average score of 15940 lands within 0.2 percent of the Intel Core Ultra 5 134U, 0.4 percent of the AMD Ryzen 5 40, 0.5 percent of the AMD EPYC 75F3, and 0.7 percent of the AMD EPYC 9354P. Within its database neighborhood the 9334 is competitive but not dominant, which frames the head-to-head result correctly: the sweep over the 115U reflects segment differences, not a claim that the EPYC outclasses every peer.

FAQ

Q: Does the Intel Core Ultra 5 115U win any benchmark against the EPYC 9334?

A: No. The EPYC 9334 wins all six shared Cinebench tests, with margins between 407.7 and 409.1 percent in every one.

Q: Is the EPYC's advantage only in multi-core workloads?

A: No. Single-core margins (407.8 to 409.1 percent) are as large as multi-core margins (407.7 to 407.8 percent), showing the server chip's per-thread lead matches its threaded lead in these tests.

Q: How do the two compare on power draw?

A: The EPYC 9334 has a 210 watt TDP; the Core Ultra 5 115U has a 15 watt TDP. The mobile chip is designed for a fundamentally lower power envelope.

Q: Which chip has integrated graphics?

A: The Core Ultra 5 115U, with Arc Xe-LPG 48EU graphics. The EPYC 9334 lists no integrated graphics.

Q: Which processors are the closest rivals to each chip?

A: The EPYC 9334's nearest rivals are the Intel Core Ultra 5 134U (0.2 percent), AMD Ryzen 5 40 (0.4 percent), AMD EPYC 75F3 (0.5 percent) and AMD EPYC 9354P (0.7 percent). The Core Ultra 5 115U's nearest rivals are the Intel Core i5-1235U (-0.1 percent), AMD Ryzen 3 7335U (-0.4 percent), AMD Ryzen 3 8440U (0.5 percent) and Intel Core i3-12100E (-0.6 percent).

Q: Do both chips support ECC memory?

A: The EPYC 9334 does. The Core Ultra 5 115U does not, consistent with its mobile segment positioning.

Specification Differences

• Cores/Threads: 32 cores, 64 threads (EPYC) versus 8 cores, 10 threads (Core Ultra 5 115U).
• Clocks: 2.70 GHz base and 3.90 GHz boost (EPYC) versus a 4.20 GHz boost on the Intel chip.
• TDP: 210 watts versus 15 watts.
• Socket: AMD Socket SP5 versus Intel BGA 2049.
• Architecture/Node: Zen 4 (Genoa) on TSMC 5 nm versus Meteor Lake on Intel 7 nm.
• Cache: 64 KB L1 and 1 MB L2 per core with 128 MB shared L3 (EPYC) versus 112 KB L1 and 2 MB L2 per core with 10 MB shared L3.
• Memory: twelve-channel DDR5 at 460.8 GB/s with ECC versus dual-channel DDR5 at 89.6 GB/s without ECC.
• PCIe: Gen 5 with 128 CPU-only lanes versus Gen 4 with 12 CPU-only lanes.
• Graphics: none versus Arc Xe-LPG 48EU.
• Segment: Server/Workstation versus Mobile.
• Release dates: November 2022 for the EPYC versus December 2023 for the Core Ultra.
• Launch MSRP: the EPYC 9334 lists a launch MSRP of $2990; no launch MSRP is recorded for the Core Ultra 5 115U.

The Verdict

The data describes two processors that should never be cross-shopped. For sustained multi-threaded compute, memory-bandwidth-hungry workloads, ECC-dependent reliability requirements, or anything needing vast PCIe connectivity, the EPYC 9334 is the only candidate here: 32 cores, 460.8 GB/s of bandwidth, 128 Gen 5 lanes, and Cinebench scores roughly five times higher across the board. Its launch MSRP of $2990 reflects that server positioning.

The Core Ultra 5 115U is for an entirely different buyer. It fits a 15 watt envelope, ships with integrated Arc Xe-LPG graphics, and slots into a BGA mobile platform. Its benchmark profile shows parity with its actual mobile rivals, all within 0.6 percent on average score, which is the comparison that matters for a thin-and-light laptop.

One caveat deserves emphasis: the EPYC's higher average benchmark score of 15940 versus 16753 actually favors the Intel chip, but that figure aggregates different test suites and should not be read as head-to-head evidence. The direct measurements are unambiguous. Choosing between these two is really a choice between a server platform and a mobile platform, and once that segment decision is made, the benchmark data leaves little ambiguity about what each chip delivers.