AMD EPYC 9115 vs Intel Core Ultra 7 265KF Comparison

# Intel Core Ultra 7 265KF vs AMD EPYC 9115

The Intel Core Ultra 7 265KF and AMD EPYC 9115 represent two fundamentally different approaches to high-performance computing, and the benchmark data reflects a clear split. The Intel part wins 15 of 17 head-to-head comparisons, dominating in rendering, encryption, and single-threaded workloads, while the AMD EPYC 9115 takes only 2 wins, but those wins are in integer math and physics simulations. The Intel Core Ultra 7 265KF posts an average benchmark score of 71910 against the EPYC 9115's 69288, a difference of roughly 3.8%, yet both sit at the 94th percentile among all CPUs. The real story is not overall performance, but where each processor is optimized to excel.

Where Each One Wins

The Intel Core Ultra 7 265KF is the clear winner for tasks that depend on high clock speeds and strong per-core efficiency. Its boost clock of 5.50 GHz versus the EPYC 9115's 4.10 GHz gives it a massive advantage in single-threaded work, and the data confirms this: it wins the single-thread PassMark test with a score of 4928 against 3360, a 46.7% margin. Every Cinebench iteration, from R15 to R23, goes to Intel by an identical 18.4% in both single-core and multi-core testing. This makes it the obvious choice for desktop workloads like content creation, gaming, and any application that relies on one or a few fast cores.

The AMD EPYC 9115, despite losing most tests, wins in two specific areas that reveal its server-oriented design. It beats the Intel part in PassMark integer math with a score of 181807 versus 143351, a 21.2% advantage. This suggests the EPYC's Zen 5 architecture handles integer-heavy calculations more efficiently, which is common in database operations, financial modeling, and certain scientific workloads. It also wins PassMark physics with 4188 versus 3633, a 13.3% margin, indicating better performance in physics simulations that leverage its 32 threads. The EPYC 9115 is not a general-purpose winner, but for integer-bound or physics-heavy server tasks, it has a genuine edge.

Architecture Differences

The architectural split is stark. The Intel Core Ultra 7 265KF uses Arrow Lake-S, built on a 3 nm process at TSMC, with 20 cores and 20 threads. It has no hyper-threading, meaning each core handles one thread, yet it still outperforms the EPYC in most multi-threaded tests thanks to higher clocks and a more efficient core design. Its cache hierarchy includes 192 KB of L1 per core, 3 MB of L2 per core, and 30 MB of shared L3. Memory support is dual-channel DDR5 with a bandwidth of 102.4 GB/s, and it does not support ECC memory. The socket is Intel Socket 1851, and it uses 20 PCIe Gen 5 lanes.

The AMD EPYC 9115 is built on Zen 5 (Turin), using a 4 nm process at TSMC, with 16 cores and 32 threads. The extra threads come from simultaneous multithreading, which allows each core to process two threads. Its cache is smaller per core: 80 KB of L1 and 1 MB of L2, but the shared L3 is more than double at 64 MB. The EPYC supports twelve-channel DDR5 memory with a bandwidth of 576.0 GB/s, and it includes ECC memory support, which is critical for server reliability. It uses AMD Socket SP5 and provides 128 PCIe Gen 5 lanes. The EPYC's die is composed of two 70.6 mm² chiplets with 16,630 million transistors, while the Intel part uses a single 243 mm² die with 17,800 million transistors. The Intel part has an unlocked multiplier, while the EPYC's is locked.

Head-to-Head Benchmarks

The Cinebench results are uniform and decisive. In Cinebench R15 multi-core, the Intel scores 5013 against 4233, a 18.4% lead. The single-core R15 shows 707 versus 597, also 18.4%. This pattern repeats exactly in R20 (20889 vs 17641 multi-core, 2948 vs 2490 single-core) and R23 (49736 vs 42003 multi-core, 7021 vs 5929 single-core). The consistency of the 18.4% delta across all Cinebench versions indicates a fundamental clock-speed advantage rather than a workload-specific quirk.

PassMark reveals a more nuanced picture. The Intel part wins data compression with 666589 versus 600099, an 11.1% margin, and data encryption with 48198 versus 33489, a 43.9% blowout. Extended instructions go to Intel by 19.9% (54513 vs 45477). The biggest Intel win is in find prime numbers, where it scores 486 versus 289, a 68.2% advantage, and floating point math, where it leads 189431 versus 113853, a 66.4% margin. The multithread test goes to Intel 58518 versus 48936, a 19.6% lead, and random string sorting favors Intel 79735 versus 70151, a 13.7% gap. Single-thread PassMark confirms the pattern: Intel wins 4928 versus 3360, a 46.7% margin.

The EPYC's two wins are notable for their specificity. In integer math, it posts 181807 against Intel's 143351, a 21.2% reversal. In physics, it scores 4188 versus 3633, a 13.3% lead. These are not small margins; they indicate that the EPYC's architecture, with its 32 threads and larger L3 cache, is better suited to certain parallel integer operations and physics calculations. The overall benchmark record shows 15 wins for Intel and 2 for AMD, but the two AMD wins are in categories where server workloads often live.

FAQ

Q: Which processor has higher single-thread performance?

A: The Intel Core Ultra 7 265KF wins every single-thread test. In Cinebench R23 single-core, it scores 7021 against the EPYC's 5929, an 18.4% lead. In PassMark single-thread, the margin expands to 46.7% (4928 vs 3360).

Q: Does the AMD EPYC 9115 win any benchmark?

A: Yes, it wins two of the 17 head-to-head tests. It leads in PassMark integer math with 181807 versus 143351 (21.2% ahead) and in PassMark physics with 4188 versus 3633 (13.3% ahead).

Q: How do the core and thread counts differ?

A: The Intel part has 20 cores and 20 threads, while the AMD EPYC has 16 cores and 32 threads. The EPYC relies on simultaneous multithreading to double its thread count, whereas the Intel part runs one thread per core.

Q: What is the memory bandwidth difference?

A: The AMD EPYC 9115 supports twelve-channel DDR5 memory with a bandwidth of 576.0 GB/s, while the Intel Core Ultra 7 265KF uses dual-channel DDR5 at 102.4 GB/s. The EPYC's bandwidth is 5.6 times higher.

Q: Which processor supports ECC memory?

A: Only the AMD EPYC 9115 supports ECC memory. The Intel Core Ultra 7 265KF does not have ECC support.

Q: What are the process nodes for each?

A: The Intel Core Ultra 7 265KF is built on a 3 nm process at TSMC, while the AMD EPYC 9115 uses a 4 nm process, also at TSMC. The smaller node gives Intel a density advantage.

The Verdict

The data points to a clear conclusion: the Intel Core Ultra 7 265KF is the superior processor for most workloads, but the AMD EPYC 9115 is the right choice for specific server tasks. If the workload involves rendering, encryption, floating-point math, or any single-threaded application, the Intel part wins decisively. Its 18.4% advantage across all Cinebench tests and 46.7% lead in single-thread PassMark make it the obvious pick for desktop and workstation use.

The AMD EPYC 9115 earns its place in environments where integer math and physics simulations dominate. Its 21.2% lead in integer math and 13.3% lead in physics, combined with 576.0 GB/s of memory bandwidth and 128 PCIe Gen 5 lanes, make it a strong candidate for database servers, financial analytics, and scientific computing. The 32 threads and 64 MB of L3 cache support heavily parallel integer workloads, and ECC memory support is essential for reliability in data centers.

The Intel part also wins on average benchmark score (71910 vs 69288), but the EPYC's two wins are not marginal. They are substantial advantages in categories that matter for its target market. The Intel Core Ultra 7 265KF is the general-purpose champion; the AMD EPYC 9115 is the specialized server tool. For a desktop user or a workstation running mixed workloads, the Intel part is the clear recommendation. For a server administrator running integer-heavy or physics-based applications, the EPYC 9115's specific strengths justify its selection.

Specification Differences

| Specification | Intel Core Ultra 7 265KF | AMD EPYC 9115 |

|---|---|---|

| Cores | 20 | 16 |

| Threads | 20 | 32 |

| Base Clock | 3.90 GHz | 2.60 GHz |

| Boost Clock | 5.50 GHz | 4.10 GHz |

| TDP | 125 W | 125 W |

| Socket | Intel Socket 1851 | AMD Socket SP5 |

| Process Node | 3 nm | 4 nm |

| L1 Cache | 192 KB (per core) | 80 KB (per core) |

| L2 Cache | 3 MB (per core) | 1 MB (per core) |

| L3 Cache | 30 MB (shared) | 64 MB (shared) |

| Memory Bus | Dual-channel | Twelve-channel |

| Memory Bandwidth | 102.4 GB/s | 576.0 GB/s |

| ECC Memory | No | Yes |

| PCIe Lanes | 20 (Gen 5) | 128 (Gen 5) |

| Multiplier | Unlocked | Locked |

| Foundry | TSMC | TSMC |

| Transistors | 17,800 million | 16,630 million |

| Die Size | 243 mm² | 2x 70.6 mm² |

| Launch MSRP | $379 | $726 |