AMD EPYC 9115 vs Intel Core Ultra 9 275HX Comparison

FAQ

Q: Which processor wins more head-to-head benchmarks?

A: The Intel Core Ultra 9 275HX wins 12 of the 17 recorded head-to-head tests, while the AMD EPYC 9115 wins 5. The Intel part dominates in most workload categories, but the AMD server chip takes decisive victories in several specific tests.

Q: How do their average benchmark scores compare?

A: The AMD EPYC 9115 has an average benchmark score of 69288 and sits in the 94th percentile of all CPUs. The Intel Core Ultra 9 275HX has an average benchmark score of 67469, also in the 94th percentile. The AMD part’s closest rival is the Intel Core i7-14700K at a delta of -0.1%, while the Intel part’s nearest rival is the Intel Xeon w5-3525 at -0.3%.

Q: What are the core and thread counts for each processor?

A: The AMD EPYC 9115 has 16 cores and 32 threads. The Intel Core Ultra 9 275HX has 24 cores and 24 threads. Despite having more physical cores, the Intel part lacks simultaneous multithreading, so its thread count equals its core count.

Q: Which processor has a higher boost clock?

A: The Intel Core Ultra 9 275HX boosts to 5.40 GHz, which is substantially higher than the AMD EPYC 9115’s 4.10 GHz boost. The base clocks are closer, with the Intel part at 2.70 GHz and the AMD part at 2.60 GHz.

Q: Do both processors support DDR5 memory?

A: Yes, both support DDR5. However, their memory channels differ significantly: the AMD EPYC 9115 uses a twelve-channel memory bus with 576.0 GB/s bandwidth, while the Intel Core Ultra 9 275HX uses a dual-channel bus with 102.4 GB/s bandwidth. The AMD part also supports ECC memory, while the Intel part does not.

Q: Which processor includes integrated graphics?

A: Only the Intel Core Ultra 9 275HX includes integrated graphics, specifically Arc Xe-LPG Graphics 64EU. The AMD EPYC 9115 has no integrated graphics (N/A), which is typical for its server/workstation market segment.

Where Each One Wins

The Intel Core Ultra 9 275HX is the clear winner in multi-threaded throughput across most Cinebench and PassMark workloads. Its 24 physical cores deliver a 32.8% higher Cinebench R15 multicore score (5619.5 vs 4233) and a 12.8% higher Cinebench R20 multicore score (19899 vs 17641). In PassMark, the Intel part wins multithread (55759 vs 48936, a 12.2% margin), floating point math (191186 vs 113853, a 40.4% margin), and prime number finding (448 vs 289, a 35.5% margin). It also leads in data encryption (47112 vs 33489, a 28.9% margin), data compression (608381 vs 600099, a 1.4% margin), extended instructions (47016 vs 45477, a 3.3% margin), and random string sorting (74320 vs 70151, a 5.6% margin).

The AMD EPYC 9115 wins where its SMT capability and architecture favor specific workloads. Its Cinebench R23 multicore score of 42003 beats the Intel part’s 35589 by 18%, and its R23 single-core score of 5929 crushes the Intel part’s 2204 by 169%. The AMD part also wins PassMark integer math (181807 vs 155218, a 17.1% margin) and PassMark physics (4188 vs 3338, a 25.5% margin). Notably, the AMD part wins Cinebench R15 single-core (597 vs 334, a 78.7% margin) despite the Intel part having a higher boost clock.

For use-case guidance: if the workload is heavily multi-threaded rendering, encryption, or floating-point math, the Intel part is the stronger choice. If the workload relies on integer math, physics simulation, or the specific Cinebench R23 test suite, the AMD part wins. The AMD part also wins the newer Cinebench R23 multicore test, suggesting its architecture scales better in that particular benchmark generation.

Architecture Differences

The AMD EPYC 9115 is built on the Zen 5 architecture, codenamed Turin, from the EPYC 9005 series. It uses a 4 nm process node from TSMC, with 16,630 million transistors spread across a dual-die design measuring 2x 70.6 mm². The Intel Core Ultra 9 275HX uses the Arrow Lake architecture from the Core Ultra Series 2, on a 3 nm TSMC process node, with 17,800 million transistors on a single 243 mm² die.

Cache layouts differ substantially. The AMD part has 80 KB of L1 cache per core, 1 MB of L2 per core, and a shared 64 MB L3 cache. The Intel part has 192 KB of L1 per core, 3 MB of L2 per core, and a shared 36 MB L3 cache. The AMD part’s larger L3 pool is a key advantage for server workloads, while the Intel part’s larger per-core L1 and L2 caches benefit its mobile performance profile.

The AMD EPYC 9115 is a server/workstation processor on AMD Socket SP5 with 128 PCIe Gen 5 lanes (CPU only). The Intel Core Ultra 9 275HX is a mobile processor on Intel BGA 2114 with 20 PCIe Gen 5 lanes (CPU only). This explains the massive difference in memory bandwidth: the AMD part’s twelve-channel DDR5 interface delivers 576.0 GB/s, while the Intel part’s dual-channel interface delivers 102.4 GB/s. The AMD part supports ECC memory; the Intel part does not.

The Intel part features integrated Arc Xe-LPG Graphics 64EU, while the AMD part has no integrated graphics. The Intel part is multiplier unlocked; the AMD part is not. The AMD part launched on 2024-10-09 with a launch MSRP of $726, while the Intel part launched later on 2025-01-12.

Specification Differences

| Specification | AMD EPYC 9115 | Intel Core Ultra 9 275HX |

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

| Cores | 16 | 24 |

| Threads | 32 | 24 |

| Base clock | 2.60 GHz | 2.70 GHz |

| Boost clock | 4.10 GHz | 5.40 GHz |

| TDP | 125 W | 55 W |

| Socket | AMD Socket SP5 | Intel BGA 2114 |

| Architecture | Zen 5 | Arrow Lake |

| Process node | 4 nm | 3 nm |

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

| Die size | 2x 70.6 mm² | 243 mm² |

| L1 cache | 80 KB per core | 192 KB per core |

| L2 cache | 1 MB per core | 3 MB per core |

| L3 cache | 64 MB shared | 36 MB shared |

| Memory bus | Twelve-channel | Dual-channel |

| Memory bandwidth | 576.0 GB/s | 102.4 GB/s |

| ECC memory | Yes | No |

| PCIe lanes | 128 Gen 5 | 20 Gen 5 |

| Integrated graphics | N/A | Arc Xe-LPG 64EU |

| Multiplier unlocked | No | Yes |

| Market segment | Server/Workstation | Mobile |

| Part number | 100-000001552 | SRVFK |

The TDP difference is stark: the AMD part consumes 125 W, while the Intel part is rated at 55 W. This means the Intel part delivers its performance at less than half the thermal envelope, a critical factor for mobile use.

Head-to-Head Benchmarks

The largest victory for the AMD EPYC 9115 is in Cinebench R23 single-core, where it scores 5929 against the Intel part’s 2204, a 169% advantage. This is an extraordinary margin, especially given that the Intel part boosts 1.3 GHz higher. The AMD part also wins Cinebench R15 single-core by 78.7% (597 vs 334) and Cinebench R23 multicore by 18% (42003 vs 35589). In PassMark, the AMD part wins integer math by 17.1% (181807 vs 155218) and physics by 25.5% (4188 vs 3338).

The Intel Core Ultra 9 275HX’s biggest win is PassMark floating point math, where it scores 191186 against the AMD part’s 113853, a 40.4% advantage. It also wins PassMark prime number finding by 35.5% (448 vs 289) and PassMark data encryption by 28.9% (47112 vs 33489). In Cinebench R15 multicore, the Intel part leads by 24.7% (5619.5 vs 4233), and in Cinebench R20 multicore by 11.3% (19899 vs 17641). The Intel part also wins PassMark single-thread by 28.7% (4713 vs 3360) and PassMark multithread by 12.2% (55759 vs 48936).

The single-thread results are particularly telling. In Cinebench R15 and R23, the AMD part wins decisively, but in PassMark single-thread, the Intel part wins by 28.7%. This divergence suggests the two architectures handle different instruction mixes differently, with the AMD part excelling in Cinebench’s rendering workload while the Intel part wins the more varied PassMark single-thread suite.

The Cinebench R20 single-core test goes to the Intel part by 11.4% (2809 vs 2490), the only Cinebench single-core test the Intel part wins. This shows that even within the same vendor’s benchmark suite, the relative performance flips between test versions, with the AMD part winning R15 and R23 single-core but losing R20 single-core.

The Verdict

The data supports a clear split based on use case. For server and workstation deployments where memory bandwidth matters, the AMD EPYC 9115 is the choice. Its twelve-channel memory bus with 576.0 GB/s bandwidth, ECC support, and 128 PCIe Gen 5 lanes make it a data-center grade part. Its 64 MB shared L3 cache and 32 threads give it a strong edge in Cinebench R23 multicore (18% ahead) and integer math (17.1% ahead). The launch MSRP of $726 positions it as a specific server entry point, though the database does not track price-performance comparisons.

For mobile workstations and high-performance laptops, the Intel Core Ultra 9 275HX is the stronger pick. It wins 12 of 17 head-to-head tests, including all major PassMark throughput metrics. Its 55 W TDP is less than half the AMD part’s 125 W, making it viable for battery-powered systems. The integrated Arc Xe-LPG Graphics 64EU adds display and acceleration capability that the AMD part lacks entirely. Its 5.40 GHz boost clock and 24 physical cores drive strong performance in floating-point math (40.4% ahead), encryption (28.9% ahead), and multithread (12.2% ahead).

The overall percentiles are identical: both sit at 94% of all CPUs. Their average benchmark scores are close, with the AMD part at 69288 and the Intel part at 67469, a difference of roughly 2.6%. The AMD part’s nearest rivals are all desktop and HX-class processors, while the Intel part’s nearest rivals include workstation Xeons and Threadripper PRO parts, confirming that both processors compete above typical consumer levels.

Choose the AMD EPYC 9115 if the workload demands maximum memory bandwidth, ECC reliability, or the specific integer and physics strengths of Zen 5. Choose the Intel Core Ultra 9 275HX if the workload is mobile-oriented, requires integrated graphics, or benefits from the floating-point and encryption performance of Arrow Lake. The Cinebench R23 multicore result is the outlier where the AMD part wins by 18%, so any workload modeled on that test will favor the server chip.