AMD EPYC 9634 vs Intel Xeon 6980P Comparison

The Intel Xeon 6980P and AMD EPYC 9634 are both 99th-percentile server processors, yet their benchmark profiles are sharply divergent. The data shows the AMD EPYC 9634 wins 14 of 17 head-to-head tests, securing a 31.2% lead in Cinebench R23 multi-core with a score of 91752 against the Intel’s 63175. However, the Intel Xeon 6980P counters with decisive victories in specialized workloads, including a 56.2% margin in extended instructions and a 41.8% edge in floating-point math, making the choice between them a matter of workload-specific priorities rather than overall performance.

Head-to-Head Benchmarks

The most striking pattern in the head-to-head results is the AMD EPYC 9634’s dominance across Cinebench and PassMark’s general multithreaded tests. In Cinebench R15 multi-core, the AMD scores 9248 versus Intel’s 6367, a 31.2% deficit for the Xeon 6980P. This gap is nearly identical across all Cinebench versions: R20 shows 38535 against 26533 (-31.1%), and R23 shows 91752 against 63175 (-31.1%). The same delta appears in PassMark’s multithread test, where the EPYC 9634 scores 107944 against 74324 (-31.1%). These results indicate that for heavily threaded, render-style workloads, the AMD part is consistently about a third faster than the Intel flagship.

Single-threaded performance follows the same direction with an even larger margin. In Cinebench R23 single-core, the EPYC 9634 scores 12953 versus 8918, a 31.2% advantage. PassMark’s single-thread test shows a more pronounced gap: 2924 versus 1681, a 42.5% lead for AMD. The physics test is the largest single delta in the entire comparison, with the EPYC 9634 scoring 12291 against 3350, a 72.7% advantage for AMD. This suggests that the Zen 4 architecture’s per-core efficiency translates directly into superior latency-sensitive and physics-based calculations.

The Intel Xeon 6980P does not go unanswered. Its most significant victory is in PassMark’s extended instructions test, where it scores 214794 against 137543, a 56.2% advantage. Floating-point math also favors Intel heavily, with 501720 versus 353784, a 41.8% edge. In data compression, Intel wins 2364519 against 2236412, a smaller but still meaningful 5.7% margin. These wins are not anomalies — they point to a processor with a specialized instruction set and FPU design that excels in computational throughput tasks.

The remaining tests split between the two. AMD wins integer math (725356 versus 637476, a 12.1% lead), random string sorting (261134 versus 240792, a 7.8% lead), and data encryption (151943 versus 125246, a 17.6% lead). Intel’s only other win beyond its three main victories is in data compression, as noted. The overall win count is lopsided at 14-3 in favor of AMD, but the magnitude of Intel’s wins in floating-point and extended instructions is large enough to matter in scientific computing and cryptography-adjacent workloads.

Architecture Differences

The architectural divide between these two processors is fundamental. The Intel Xeon 6980P is built on a 5 nm process at Intel’s foundry, using the Granite Rapids architecture with a die size of 3x 598 mm². The AMD EPYC 9634 also uses a 5 nm node, but fabricated by TSMC, with a chiplet design of 12x 72 mm² and a transistor count of 78,840 million. The Intel part is monolithic in comparison, though both are active production parts.

Core counts diverge sharply: the Xeon 6980P has 128 cores and 256 threads, while the EPYC 9634 has 84 cores and 168 threads. Despite having 44 fewer cores, the AMD part wins most multithreaded benchmarks, which indicates a substantial per-core performance advantage. The base clocks tell part of the story — Intel runs at 2.00 GHz base and 3.90 GHz boost, while AMD runs at 2.25 GHz base and 3.70 GHz boost. The AMD part’s higher base clock and lower boost clock suggest a more efficient frequency curve.

Cache hierarchies also differ significantly. The Intel Xeon 6980P has 112 KB of L1 per core, 2 MB of L2 per core, and a massive 504 MB of shared L3 cache. The AMD EPYC 9634 has 64 KB of L1 per core, 1 MB of L2 per core, and 384 MB of shared L3 cache. Intel’s L3 is 120 MB larger, yet it does not translate into a win in most memory-intensive benchmarks, except for extended instructions and compression. The per-core L2 advantage for Intel (2 MB versus 1 MB) is offset by AMD’s superior core efficiency.

Memory support is similar in channel count — both are twelve-channel DDR5 — but bandwidth differs. Intel lists 614.4 GB/s, while AMD lists 460.8 GB/s. This 33% bandwidth advantage for Intel likely contributes to its wins in floating-point and compression, which are often memory-bound. PCIe connectivity also differs: Intel provides Gen 5 with 96 lanes (CPU only), while AMD provides Gen 5 with 128 lanes (CPU only), giving the EPYC 9634 a 32-lane advantage for I/O-heavy configurations.

FAQ

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

A: The AMD EPYC 9634 wins 14 of 17 tests, while the Intel Xeon 6980P wins 3. The AMD part’s wins include all Cinebench versions, PassMark multithread, single-thread, physics, integer math, encryption, and random string sorting.

Q: What is the largest benchmark gap between the two?

A: The largest delta is in PassMark’s physics test, where the AMD EPYC 9634 scores 12291 against the Intel Xeon 6980P’s 3350, a 72.7% advantage. The second-largest is in extended instructions, where Intel wins by 56.2% with 214794 versus 137543.

Q: How do the core counts compare, and does it matter?

A: The Intel Xeon 6980P has 128 cores and 256 threads, while the AMD EPYC 9634 has 84 cores and 168 threads. Despite fewer cores, the AMD part wins most multithreaded benchmarks, including Cinebench R23 multi-core by 31.1%, indicating higher per-core throughput.

Q: What are the cache size differences?

A: Intel provides 112 KB of L1 per core, 2 MB of L2 per core, and 504 MB of shared L3. AMD provides 64 KB of L1 per core, 1 MB of L2 per core, and 384 MB of shared L3. Intel’s L3 cache is 120 MB larger.

Q: Which processor has higher memory bandwidth?

A: The Intel Xeon 6980P lists 614.4 GB/s of memory bandwidth, while the AMD EPYC 9634 lists 460.8 GB/s. Both support twelve-channel DDR5 memory.

Q: What is the average benchmark score for each?

A: The Intel Xeon 6980P has an average benchmark score of 251516, while the AMD EPYC 9634 has 244274. The Intel part is 3% higher in the average score metric, despite losing most head-to-head tests.

The Verdict

The data presents a clear trade-off. For general server workloads — rendering, physics, multi-threaded applications, and single-thread latency — the AMD EPYC 9634 is the better choice. Its 31.2% lead in Cinebench R23 multi-core and 42.5% lead in PassMark single-thread are decisive, and it achieves this with fewer cores. The 72.7% advantage in physics makes it particularly suited for simulation and real-time analytics.

The Intel Xeon 6980P is the right pick for workloads that leverage its specialized strengths. Its 56.2% lead in extended instructions and 41.8% lead in floating-point math, combined with a 5.7% edge in data compression, make it a superior option for scientific computing, numerical analysis, and compression-heavy pipelines. The 33% memory bandwidth advantage (614.4 GB/s versus 460.8 GB/s) supports these high-throughput tasks.

Neither processor is a universal winner. The AMD EPYC 9634’s 14-3 head-to-head record is impressive, but the Intel part’s wins are in categories where AMD cannot compete. The average benchmark score slightly favors Intel (251516 versus 244274, a 3% delta), but this aggregate metric masks the lopsided test-level outcomes. Buyers should match the processor to the workload: AMD for broad general-purpose performance, Intel for specialized computational density.

Specification Differences

| Specification | Intel Xeon 6980P | AMD EPYC 9634 |

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

| Cores | 128 | 84 |

| Threads | 256 | 168 |

| Base Clock | 2.00 GHz | 2.25 GHz |

| Boost Clock | 3.90 GHz | 3.70 GHz |

| TDP | 500 W | 290 W |

| Socket | Intel Socket 7529 | AMD Socket SP5 |

| Architecture | Granite Rapids | Zen 4 |

| Codename | Granite Rapids | Genoa |

| Process Node | 5 nm (Intel) | 5 nm (TSMC) |

| Die Size | 3x 598 mm² | 12x 72 mm² |

| Transistors | Not listed | 78,840 million |

| L1 Cache | 112 KB (per core) | 64 KB (per core) |

| L2 Cache | 2 MB (per core) | 1 MB (per core) |

| L3 Cache | 504 MB (shared) | 384 MB (shared) |

| Memory Bandwidth | 614.4 GB/s | 460.8 GB/s |

| PCIe | Gen 5, 96 Lanes | Gen 5, 128 Lanes |

| Release Date | 2024-09-23 | 2022-11-09 |

| Launch MSRP | $12460 | $10304 |

| Part Number | SRPL2Q5SM | 100-100000797 |