AMD EPYC 9115 vs Intel Xeon Platinum 8270 Comparison

Head-to-Head Benchmarks

The benchmark data delivers a decisive verdict: the AMD EPYC 9115 wins 14 of the 17 recorded head-to-head comparisons, while the Intel Xeon Platinum 8270 takes only 3. The margin is not narrow. In Cinebench testing, the EPYC 9115 leads by a consistent 32% across all six tests, including both single-core and multi-core workloads. The R23 multi-core score of 42,003 versus the Xeon's 28,561 illustrates a substantial performance gap, and the single-core R23 result of 5,929 versus 4,032 confirms the advantage is not limited to heavily threaded tasks.

PassMark results show a similar pattern. The multithread test gives the EPYC 9115 a 38.7% lead (48,936 versus 29,986). Physics performance is even more lopsided: the EPYC scores 4,188 against the Xeon's 903, a 78.4% deficit for Intel. Prime number finding favors the AMD part by 70.9% (289 versus 84). Data encryption shows the EPYC at 33,489 versus 8,798, a 73.7% margin. Floating point math favors AMD by 12.7% (113,853 versus 99,432), and integer math by 11.8% (181,807 versus 160,322). Even single-thread performance, an area where older architectures often remain competitive, shows the EPYC 9115 ahead by 30.8% (3,360 versus 2,325).

The Intel Xeon Platinum 8270 does retain three wins, and they are worth noting. Data compression favors Intel by 21.3% with a score of 728,144 versus 600,099. Extended instructions go to Intel by 12.6% (51,191 versus 45,477). Random string sorting also goes Intel's way by 14.3% (80,208 versus 70,151). These results indicate that specific workload types, particularly those that stress memory access patterns or instruction-level parallelism in certain algorithms, can still favor the older Intel design. However, these three wins are narrow in scope compared to the breadth of AMD's victories.

The average benchmark score tells a similar story, though with less drama. The Xeon Platinum 8270 averages 71,370, while the EPYC 9115 averages 69,288. This puts the Intel part 0.4% ahead of the Xeon 6511P and 0.7% ahead of the Core Ultra 7 265K, while trailing the Ryzen 7 8840HX by 0.6%. The EPYC 9115 sits nearly level with the Core i7-14700K, which is only 0.1% ahead, and trails the Ryzen 9 7950X by just 0.3%. Both processors sit at the 94th percentile of all CPUs in the database, meaning their overall standings are comparable, but the distribution of wins is not.

Architecture Differences

The architectural gap between these two processors explains the benchmark results. The Intel Xeon Platinum 8270 uses the Cascade Lake architecture on a 14 nm process from Intel's own foundry. It packs 26 cores and 52 threads. The AMD EPYC 9115 uses the Zen 5 architecture, codenamed Turin, built on a 4 nm process at TSMC. It has fewer cores, 16, and fewer threads, 32, yet it outperforms the Intel part in most tests. This is a clear indication that per-core efficiency and IPC improvements in Zen 5 outweigh the raw core count advantage of the Xeon.

Cache configurations differ sharply. The Xeon Platinum 8270 provides 64 KB of L1 per core, 1 MB of L2 per core, and 35.75 MB of shared L3. The EPYC 9115 has 80 KB of L1 per core, 1 MB of L2 per core, and a much larger 64 MB of shared L3. The larger L3 cache on the AMD part likely contributes to its strength in floating point and integer math, where data reuse is common. The Xeon's smaller L3, despite more cores, cannot compensate for the architectural efficiency difference.

Memory support also diverges. The Xeon uses DDR4, while the EPYC 9115 uses DDR5 with a twelve-channel memory bus. The EPYC's memory bandwidth is listed at 576.0 GB/s, a figure that dwarfs what the DDR4-based Xeon can offer. The EPYC also brings PCIe Gen 5 with 128 lanes (CPU only), while the Xeon's PCIe capability is not recorded in the database. The EPYC 9115 has an integrated graphics field marked as "N/A," while the Xeon's is absent from the records entirely. Both support ECC memory.

The transistor counts differ by a factor of roughly two: the Xeon uses 8,000 million transistors, while the EPYC 9115 uses 16,630 million. The EPYC's die size is listed as 2x 70.6 mm², while the Xeon's die size is not recorded. The EPYC 9115 belongs to the EPYC 9005 series, while the Xeon's series field is null. The EPYC is marked as an active production part, while the Xeon's production status is not recorded. The release dates are far apart: the Xeon launched in December 2018, the EPYC in October 2024.

The Verdict

The data points to the AMD EPYC 9115 as the stronger processor for nearly all recorded workloads. It wins every Cinebench iteration by 32%, wins the PassMark multithread test by 38.7%, and dominates in physics, encryption, and prime number finding by margins of 70% or more. The Xeon Platinum 8270 cannot match the EPYC's efficiency per core. Even though the Xeon has 10 more cores and 20 more threads, the EPYC's 16 Zen 5 cores deliver higher scores in multi-threaded tests. The only areas where the Xeon wins are data compression, extended instructions, and random string sorting, and those wins are smaller in magnitude than AMD's lead in most other tests.

For workloads that emphasize compression, sorting, or specific extended instruction patterns, the Xeon Platinum 8270 still holds a niche advantage, with leads of 21.3%, 14.3%, and 12.6% respectively. But for general computing, rendering, physics simulation, encryption, and single-threaded responsiveness, the EPYC 9115 is the clear choice based on recorded data. The EPYC also operates with a lower TDP: 125 watts versus the Xeon's 205 watts. The EPYC 9115 has a launch MSRP of $726, which can be stated once as the recorded launch price.

Specification Differences

| Specification | Intel Xeon Platinum 8270 | AMD EPYC 9115 |

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

| Cores | 26 | 16 |

| Threads | 52 | 32 |

| Base Clock | 2.70 GHz | 2.60 GHz |

| Boost Clock | 4.00 GHz | 4.10 GHz |

| TDP | 205 W | 125 W |

| Socket | Intel Socket 3647 | AMD Socket SP5 |

| Architecture | Cascade Lake | Zen 5 |

| Codename | Cascade Lake-SP | Turin |

| Process Node | 14 nm | 4 nm |

| Foundry | Intel | TSMC |

| Transistors | 8,000 million | 16,630 million |

| Die Size | Not recorded | 2x 70.6 mm² |

| L1 Cache | 64 KB (per core) | 80 KB (per core) |

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

| L3 Cache | 35.75 MB (shared) | 64 MB (shared) |

| Memory Support | DDR4 | DDR5 |

| Memory Bus | Not recorded | Twelve-channel |

| Memory Bandwidth | Not recorded | 576.0 GB/s |

| PCIe | Not recorded | Gen 5, 128 Lanes (CPU only) |

| Integrated Graphics | Not recorded | N/A |

| Release Date | 2018-12-10 | 2024-10-09 |

| Part Number | SRF96CD8069504195201 | 100-000001552 |

| Multiplier Unlocked | No | No |

FAQ

Q: Which processor has more cores?

A: The Intel Xeon Platinum 8270 has 26 cores and 52 threads, while the AMD EPYC 9115 has 16 cores and 32 threads.

Q: How do they compare in Cinebench R23 multi-core performance?

A: The AMD EPYC 9115 scores 42,003, which is 32% higher than the Intel Xeon Platinum 8270's 28,561.

Q: In which tests does the Intel Xeon Platinum 8270 win?

A: The Xeon wins in three tests: data compression with a 21.3% lead, extended instructions with a 12.6% lead, and random string sorting with a 14.3% lead.

Q: What is the difference in memory bandwidth?

A: The AMD EPYC 9115 has a twelve-channel memory bus with 576.0 GB/s bandwidth and supports DDR5, while the Intel Xeon Platinum 8270 supports DDR4 and has no memory bandwidth recorded in the database.

Q: Which processor has a smaller process node?

A: The AMD EPYC 9115 is built on a 4 nm process at TSMC, while the Intel Xeon Platinum 8270 uses a 14 nm process from Intel.

Q: What is the TDP difference?

A: The Intel Xeon Platinum 8270 has a TDP of 205 watts, while the AMD EPYC 9115 has a TDP of 125 watts.

Where Each One Wins

The AMD EPYC 9115 is the winner in rendering, physics simulation, encryption, prime number computation, floating point math, integer math, and all recorded single-threaded and multi-threaded Cinebench tests. Its 32% lead across every Cinebench test, combined with a 38.7% lead in PassMark multithread, makes it the default choice for general compute-heavy server or workstation tasks. The 78.4% physics lead and 73.7% encryption lead are particularly strong, suggesting the Zen 5 architecture excels in these specific workloads.

The Intel Xeon Platinum 8270 wins only in data compression, extended instructions, and random string sorting. These are narrower victories, with margins of 21.3%, 12.6%, and 14.3% respectively. For workloads that process large amounts of data requiring compression or sorting, the older Intel part still demonstrates a measurable advantage. However, these three wins do not offset the EPYC's dominance across the remaining 14 benchmark categories. The EPYC also achieves its performance with a lower TDP of 125 watts versus 205 watts, and it uses a modern DDR5 memory interface with 576.0 GB/s bandwidth. The choice is clear for most users: the EPYC 9115 is the higher-performing processor in the database's recorded tests, with the Xeon retaining only a narrow set of specialized wins.