AMD EPYC 7F72 vs Intel Core Ultra 9 285K Comparison

The AMD EPYC 7F72 and Intel Core Ultra 9 285K are both 24-core processors sitting at the 96th percentile of all CPUs, yet they achieve that status through radically different designs. The benchmark data reveals a clear split: the EPYC 7F72 dominates in single-core Cinebench tests and several specialized server workloads, while the Core Ultra 9 285K wins the majority of head-to-head matchups, particularly in floating-point math and newer multi-threaded benchmarks. Their average benchmark scores are separated by just 1.5%, with the EPYC 7F72 posting an 85072 average against the Intel chip's 83807, making this a contest decided by workload rather than overall capability.

Head-to-Head Benchmarks

The most striking result in the entire comparison is the Cinebench R23 single-core test, where the EPYC 7F72 scores 6328 against the Core Ultra 9's 2377, a staggering 166.2% advantage. This pattern repeats in Cinebench R15 single-core, with the EPYC winning 637 to 359, a 77.4% margin. The EPYC also wins Cinebench R23 multi-core, scoring 44829 versus 42522, a 5.4% edge that shows its 48 threads overcoming the Intel chip's higher clock speeds.

The Core Ultra 9 285K fights back decisively in other Cinebench versions. It wins Cinebench R15 multi-core with 6494 against 4518, a 30.4% margin, and takes Cinebench R20 multi-core 24003 to 18828, a 21.6% lead. The Intel chip also wins Cinebench R20 single-core 3388 to 2657, another 21.6% margin, showing that the EPYC's single-core advantage is not universal across Cinebench versions.

PassMark results paint a more complex picture. The Intel chip dominates floating-point math, scoring 224324 against the EPYC's 108437, a 51.7% blowout. It also wins extended instructions by 24.6% (62277 to 46936), single-thread performance by 53.1% (5087 to 2384), and multi-thread by 21.6% (67260 to 52740). The EPYC counters with wins in integer math (181103 to 172379, a 5.1% edge), data compression (808795 to 790052, a 2.4% margin), random string sorting (102436 to 94927, a 7.9% lead), and physics (6459 to 3938, a 64% advantage). The Intel chip narrowly takes data encryption 57745 to 56261, a 2.6% margin, and find prime numbers 541 to 498, a 7.9% edge. Overall, the Intel chip wins 10 of 17 head-to-head benchmarks, but the EPYC's wins include some of the largest margins in either direction.

Architecture Differences

The two processors come from opposite ends of the computing spectrum. The AMD EPYC 7F72 is a server/workstation part built on Zen 2 architecture with the Rome codename, manufactured on TSMC's 7 nm process. It uses the AMD Socket SP3 platform and supports DDR4 memory across an eight-channel bus, delivering 204.8 GB/s of memory bandwidth. The Intel Core Ultra 9 285K is a desktop processor from the Core Ultra Series 2, built on Arrow Lake architecture with the Arrow Lake-S codename, manufactured on TSMC's 3 nm process. It uses Intel Socket 1851, supports DDR5 memory on a dual-channel bus, and provides 102.4 GB/s of bandwidth.

Cache configurations differ dramatically. The EPYC 7F72 has 96 KB of L1 per core, 512 KB of L2 per core, and a massive 192 MB of shared L3 cache. The Core Ultra 9 285K has 192 KB of L1 per core, 3 MB of L2 per core, and only 36 MB of shared L3 cache. The transistor counts reflect their different scales: the EPYC uses 3,800 million transistors on a 74 mm² die, while the Intel chip uses 17,800 million transistors on a 243 mm² die.

Clock speeds favor Intel significantly. The Core Ultra 9 285K has a base clock of 3.70 GHz and a boost clock of 5.70 GHz, while the EPYC 7F72 runs at 3.20 GHz base and 3.70 GHz boost. The Intel chip has a 125 W TDP versus the EPYC's 240 W, and it includes integrated Arc Xe-LPG Graphics with 64 execution units, whereas the EPYC has no integrated graphics. Both support ECC memory, but the EPYC uses PCIe Gen 4 while the Intel chip uses PCIe Gen 5 with 20 lanes from the CPU. The Intel chip is multiplier unlocked, while the EPYC is not. The EPYC launched in April 2020, while the Core Ultra 9 arrived in October 2024.

FAQ

Q: Which processor has more threads?

A: The AMD EPYC 7F72 has 48 threads from its 24 cores, while the Intel Core Ultra 9 285K has 24 threads from its 24 cores, meaning the EPYC offers simultaneous multithreading that the Intel chip lacks.

Q: What is the largest single benchmark margin between the two?

A: The EPYC 7F72's 166.2% advantage in Cinebench R23 single-core is the largest margin, followed by its 77.4% lead in Cinebench R15 single-core and the Intel chip's 51.7% win in PassMark floating-point math.

Q: How do their average benchmark scores compare?

A: The EPYC 7F72 has an average benchmark score of 85072, while the Core Ultra 9 285K averages 83807, giving the EPYC a 1.5% advantage in overall average performance.

Q: Which processor has a higher boost clock?

A: The Intel Core Ultra 9 285K boosts to 5.70 GHz, while the AMD EPYC 7F72 boosts to only 3.70 GHz, a 2.0 GHz difference in favor of Intel.

Q: Do both processors support ECC memory?

A: Yes, both the AMD EPYC 7F72 and the Intel Core Ultra 9 285K support ECC memory, though the EPYC uses DDR4 across eight channels while the Intel chip uses DDR5 across two channels.

Q: How many benchmark wins does each processor have?

A: The Intel Core Ultra 9 285K wins 10 of the 17 head-to-head benchmarks, while the AMD EPYC 7F72 wins 7.

Specification Differences

| Specification | AMD EPYC 7F72 | Intel Core Ultra 9 285K |

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

| Threads | 48 | 24 |

| Base Clock | 3.20 GHz | 3.70 GHz |

| Boost Clock | 3.70 GHz | 5.70 GHz |

| TDP | 240 W | 125 W |

| Socket | AMD Socket SP3 | Intel Socket 1851 |

| Architecture | Zen 2 | Arrow Lake |

| Codename | Rome | Arrow Lake-S |

| Process Node | 7 nm | 3 nm |

| Transistors | 3,800 million | 17,800 million |

| Die Size | 74 mm² | 243 mm² |

| L1 Cache | 96 KB (per core) | 192 KB (per core) |

| L2 Cache | 512 KB (per core) | 3 MB (per core) |

| L3 Cache | 192 MB (shared) | 36 MB (shared) |

| Memory Support | DDR4 | DDR5 |

| Memory Bus | Eight-channel | Dual-channel |

| Memory Bandwidth | 204.8 GB/s | 102.4 GB/s |

| PCIe | Gen 4 | Gen 5, 20 Lanes (CPU only) |

| Integrated Graphics | None | Arc Xe-LPG Graphics 64EU |

| Market Segment | Server/Workstation | Desktop |

| Release Date | 2020-04-13 | 2024-10-23 |

| Multiplier Unlocked | No | Yes |

The Verdict

The benchmark data shows two different philosophies. The AMD EPYC 7F72 is built for throughput on server workloads, with 48 threads, 192 MB of L3 cache, and eight-channel DDR4 memory delivering 204.8 GB/s of bandwidth. Its wins in integer math, data compression, and physics reflect a design optimized for database and enterprise tasks. The Intel Core Ultra 9 285K is a desktop flagship with a 5.70 GHz boost clock, 3 nm process, and integrated graphics, winning floating-point math by 51.7% and single-thread PassMark by 53.1%.

The EPYC 7F72's Cinebench R23 single-core score of 6328 versus the Intel chip's 2377 is an anomaly that flips the expected desktop-versus-server narrative, but the Intel chip wins the newer Cinebench R20 tests by 21.6% in both single and multi-core. For users prioritizing raw multi-threaded performance in Cinebench R15 and R20, the Core Ultra 9 285K is the clear choice, winning by 30.4% and 21.6% respectively. For integer-heavy server work and massive cache workloads, the EPYC 7F72 holds advantages of 5.1% in integer math and 2.4% in data compression.

The Intel chip's 125 W TDP versus the EPYC's 240 W makes it the more power-efficient option, and its unlocked multiplier allows overclocking that the EPYC cannot offer. The EPYC's eight-channel memory subsystem provides double the bandwidth of the Intel chip's dual-channel setup, which matters for memory-bound server applications. The 1.5% average score difference favors the EPYC, but the Intel chip wins more individual benchmarks. The verdict depends entirely on whether the workload resembles a server environment or a desktop environment.

Where Each One Wins

The AMD EPYC 7F72 wins in Cinebench R15 single-core by 77.4%, Cinebench R23 single-core by 166.2%, Cinebench R23 multi-core by 5.4%, PassMark data compression by 2.4%, PassMark integer math by 5.1%, PassMark physics by 64%, and PassMark random string sorting by 7.9%. These results point to workloads involving integer calculations, compression algorithms, physics simulations, and string processing, where the EPYC's 48 threads and 192 MB L3 cache provide a decisive advantage.

The Intel Core Ultra 9 285K wins in Cinebench R15 multi-core by 30.4%, Cinebench R20 multi-core by 21.6%, Cinebench R20 single-core by 21.6%, PassMark data encryption by 2.6%, PassMark extended instructions by 24.6%, PassMark find prime numbers by 7.9%, PassMark floating-point math by 51.7%, PassMark multi-thread by 21.6%, and PassMark single-thread by 53.1%. The floating-point and extended instruction results indicate superiority in scientific computing, encryption workloads, and single-threaded desktop applications, while the multi-thread PassMark win shows the Intel chip's 24 threads outperform the EPYC's 48 threads in that specific test.

For server and workstation deployments requiring maximum cache, memory bandwidth, and integer throughput, the EPYC 7F72 is the data-backed choice. For desktop users, overclockers, and workloads dominated by floating-point math, single-thread performance, or encryption, the Core Ultra 9 285K delivers superior results. The two processors share the same core count and percentile ranking, but the data shows they are optimized for different worlds.