AMD EPYC 7F72 vs Intel Core Ultra 9 290K Plus Comparison

The AMD EPYC 7F72 and Intel Core Ultra 9 290K Plus present a stark contrast in design philosophy, with the former being a 24-core/48-thread server processor built on the Zen 2 architecture for Socket SP3, and the latter a 24-core/24-thread desktop chip on the Arrow Lake Refresh architecture for Socket 1851. While both processors achieve the 96th percentile in the benchmark database, their average scores are remarkably close, with the EPYC 7F72 averaging 85,072 points against the Intel's 84,003, a margin of just 1.3%. Yet beneath these similar averages lies a deeply split performance profile, with the Intel part winning 12 of the 17 head-to-head benchmarks while the AMD part takes 5, often by significant margins.

Head-to-Head Benchmarks

The most decisive victory for the Intel Core Ultra 9 290K Plus comes in single-threaded workloads. In PassMark single-thread testing, the Intel part scores 4,823 versus the EPYC's 2,384, a 50.6% advantage. This pattern repeats in Cinebench single-core tests, where the Intel chip leads by 13.4-13.5% across R15, R20, and R23, with scores of 736, 3,067, and 7,303 respectively, compared to the EPYC's 637, 2,657, and 6,328.

Multi-threaded Cinebench results also favor Intel, though by a smaller margin. The Core Ultra 9 290K Plus posts 5,214 in Cinebench R15 multi-core, 21,727 in R20, and 51,731 in R23, each representing a 13.3% lead over the EPYC 7F72's 4,518, 18,828, and 44,829. This consistency across three Cinebench versions suggests a fundamental throughput advantage in this specific rendering workload.

PassMark floating-point math is where Intel's dominance becomes extreme, with the Core Ultra 9 290K Plus scoring 214,760 against the EPYC's 108,437, a 49.5% delta. The Intel chip also shows a 10.3% lead in extended instructions (52,338 vs 46,936) and a narrow 1% advantage in prime number finding (503 vs 498). PassMark multithread follows the Cinebench pattern, with Intel ahead at 60,860 versus 52,740, a 13.3% gap.

However, the AMD EPYC 7F72 claims several important wins, particularly in memory-related and integer workloads. Data compression is the largest AMD victory, with the EPYC scoring 808,795 against Intel's 698,346, a 15.8% advantage. Random string sorting goes to AMD by 28.5% (102,436 vs 79,744), and integer math favors the EPYC by 9% (181,103 vs 166,194). Data encryption is also an AMD win, albeit narrower at 7% (56,261 vs 52,563).

The most striking single result is PassMark physics, where the EPYC 7F72 scores 6,459 against Intel's 3,315, a 94.8% advantage—nearly double the performance. This particular workload appears to respond strongly to the EPYC's thread count and memory architecture, as the data shows no other benchmark with such a lopsided result.

Architecture Differences

The two processors diverge fundamentally in their construction. The AMD EPYC 7F72 uses a 7 nm process node from TSMC, packing 3,800 million transistors onto a 74 mm² die. In contrast, the Intel Core Ultra 9 290K Plus uses a 3 nm TSMC process, with 17,800 million transistors on a 243 mm² die. This represents a significantly denser and larger chip for Intel.

Cache hierarchies are dramatically different. The EPYC 7F72 features 192 KB of L1 per core, 512 KB of L2 per core, and a massive 192 MB of shared L3 cache. The Intel part offers 192 KB of L1 per core, 3 MB of L2 per core, but only 36 MB of shared L3. The EPYC's six-fold L3 advantage is likely a major factor in its data compression and string sorting wins, where large working sets can reside in cache.

Memory architecture also separates them. The EPYC 7F72 supports DDR4 over an eight-channel memory bus, delivering 204.8 GB/s of bandwidth. The Intel Core Ultra 9 290K Plus uses DDR5 over a dual-channel bus, providing 115.2 GB/s. Despite the newer DDR5 standard, the EPYC's eight-channel configuration offers 78% more raw bandwidth. Both support ECC memory, but the EPYC targets server workloads while Intel's part is a desktop chip with integrated Arc Xe-LPG Graphics 64EU; the EPYC has no integrated graphics.

The core counts match at 24, but threading differs: the EPYC 7F72 supports 48 threads through simultaneous multithreading, while the Intel chip runs 24 threads with no hyperthreading. Clock speeds favor Intel, with a 3.70 GHz base and 5.80 GHz boost versus the EPYC's 3.20 GHz base and 3.70 GHz boost. Power envelopes also differ, with the EPYC rated at 240W TDP and Intel at 125W. The EPYC uses PCIe Gen 4, while Intel provides PCIe Gen 5 with 20 CPU lanes. The Intel part also has an unlocked multiplier, unlike the locked EPYC.

Where Each One Wins

The AMD EPYC 7F72's strengths lie in workloads that leverage its massive L3 cache and high memory bandwidth. Data compression and random string sorting are classic memory-bound tasks, and the EPYC's 15.8% and 28.5% wins respectively reflect this. Integer math and encryption also favor AMD by 9% and 7%, suggesting the EPYC's architecture is better suited to these specific ALU and cryptographic operations. The physics result, at 94.8% ahead, indicates a scenario where the combination of 48 threads and high bandwidth produces outsized returns.

The Intel Core Ultra 9 290K Plus dominates single-threaded performance, as expected from its 5.80 GHz boost clock. The 50.6% lead in PassMark single-thread and 13.4% lead in Cinebench single-core make it the clear choice for latency-sensitive applications that depend on one or few cores. Floating-point math shows a 49.5% advantage, suggesting Intel's FPU design is significantly more capable. Cinebench multi-core wins by 13.3% indicate that even in multi-threaded rendering, Intel's higher clock speeds on its 24 threads outweigh AMD's 48 threads in this specific workload.

PassMark multithread follows the same 13.3% pattern, reinforcing that in general multi-threaded throughput, Intel's higher frequency per core compensates for fewer threads. Extended instructions (10.3% lead) and prime number finding (1% lead) round out Intel's edge in compute-heavy integer and SIMD workloads.

FAQ

Q: Which processor has better single-threaded performance?

A: The Intel Core Ultra 9 290K Plus is decisively ahead, winning PassMark single-thread by 50.6% (4,823 vs 2,384) and all Cinebench single-core tests by 13.4-13.5%.

Q: How do they compare in multi-threaded Cinebench performance?

A: Intel wins consistently across all three versions, with a 13.3% lead in R15 (5,214 vs 4,518), R20 (21,727 vs 18,828), and R23 (51,731 vs 44,829).

Q: Where does the AMD EPYC 7F72 show its biggest advantage?

A: The EPYC's largest win is in PassMark physics, where it scores 6,459 versus Intel's 3,315, a 94.8% advantage. It also leads in random string sorting by 28.5% and data compression by 15.8%.

Q: Which processor has more L3 cache?

A: The AMD EPYC 7F72 has 192 MB of shared L3 cache, while the Intel Core Ultra 9 290K Plus has only 36 MB, a six-fold difference.

Q: Do both processors support ECC memory?

A: Yes, both the AMD EPYC 7F72 and Intel Core Ultra 9 290K Plus support ECC memory, though the EPYC uses DDR4 over eight channels while Intel uses DDR5 over two channels.

Q: What is the overall benchmark win count?

A: The Intel Core Ultra 9 290K Plus wins 12 of the 17 head-to-head benchmarks, while the AMD EPYC 7F72 wins 5.

The Verdict

The data presents a clear split based on workload type. For users prioritizing single-threaded performance, floating-point math, or general Cinebench rendering, the Intel Core Ultra 9 290K Plus is the superior choice, with leads ranging from 13.3% to 50.6% across these categories. Its higher boost clock of 5.80 GHz and 3 nm process node provide a substantial performance edge in frequency-sensitive tasks.

However, for workloads that stress memory bandwidth and cache capacity, the AMD EPYC 7F72 is the better option. Its 192 MB L3 cache and 204.8 GB/s eight-channel bandwidth deliver wins of 15.8% in data compression, 28.5% in string sorting, and a remarkable 94.8% in physics. These are not marginal advantages; they represent category-level dominance.

The average benchmark scores show the EPYC 7F72 at 85,072 versus Intel's 84,003, a 1.3% overall edge for AMD. Yet this narrow average masks the divergent strengths. The Intel part is the desktop performance leader for most users, especially given its integrated graphics and unlocked multiplier. The AMD part is a server/workstation chip that excels in specific data-intensive operations. Given the EPYC's 240W TDP against Intel's 125W, the Intel chip also offers better efficiency for its performance profile. The choice ultimately depends on whether the workload leans toward Intel's frequency advantage or AMD's cache and bandwidth superiority.

Specification Differences

| Specification | AMD EPYC 7F72 | Intel Core Ultra 9 290K Plus |

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

| Threads | 48 | 24 |

| Base Clock | 3.20 GHz | 3.70 GHz |

| Boost Clock | 3.70 GHz | 5.80 GHz |

| TDP | 240W | 125W |

| Socket | AMD Socket SP3 | Intel Socket 1851 |

| Process Node | 7 nm | 3 nm |

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

| Die Size | 74 mm² | 243 mm² |

| 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 | 115.2 GB/s |

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

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

| Multiplier Unlocked | No | Yes |

| Market Segment | Server/Workstation | Desktop |

| Release Date | 2020-04-13 | Not specified |