AMD EPYC 9655P vs Intel Xeon 696X Comparison

Head-to-Head Benchmarks

The benchmark data is unambiguous: the AMD EPYC 9655P wins all 14 recorded comparisons against the Intel Xeon 696X. The margin varies dramatically by workload, from a narrow single-thread edge to a massive lead in integer math and physics.

Starting with the multi-core rendering tests, the AMD part is consistently about 53% ahead. In Cinebench R15 multicore, the EPYC scores 13,744 versus 8,994, a delta of 52.8%. The same percentage gap repeats in Cinebench R20 (57,268 vs 37,475) and Cinebench R23 (136,354 vs 89,227). This consistency suggests a structural advantage rather than a workload-specific quirk.

The single-thread results tell a different story. In PassMark single-thread, the EPYC 9655P scores 3,849 versus 3,742 for the Xeon, a modest 2.9% advantage. That is a narrow margin, but it still goes to AMD. The Cinebench R15 single-core test was not recorded for the Intel part, so the comparison relies on the PassMark figures.

The most lopsided result is PassMark physics, where the EPYC 9655P scores 25,847 versus 3,382, a staggering 664.3% difference. That is not a typo in the data; it reflects a fundamental capability gap in that specific test. Similarly, integer math shows a 114.2% lead (1,225,251 vs 572,072), and random string sorting shows a 150.5% lead (451,824 vs 180,392).

Data encryption is another standout: the EPYC scores 220,074 versus 112,529, a 95.6% advantage. Prime number finding shows a 97.7% lead (1,686 vs 853). Floating point math is 59% ahead (715,866 vs 450,164). Data compression is 53.9% ahead (3,486,158 vs 2,264,907). Extended instructions are 33.3% ahead (230,609 vs 172,975). Multithread overall is 52.9% ahead (160,490 vs 104,974).

The average benchmark score tells the broader story: the EPYC 9655P averages 397,773 across all recorded tests, while the Xeon 696X averages 286,102. The AMD chip sits at the 100th percentile of all CPUs in the database, while the Intel part sits at the 99th. Both are elite parts, but the gap between them is wide.

Where Each One Wins

The data shows no test where the Intel Xeon 696X comes out ahead. That does not mean it is useless; it means its strengths lie in areas not captured by these benchmarks or in system-level characteristics.

For the AMD EPYC 9655P, the wins are everywhere in compute-heavy workloads. The 52.8% lead across all three Cinebench versions indicates strong multi-threaded rendering performance. The 114.2% lead in integer math and 150.5% lead in string sorting suggest heavy database, compression, and general server workloads will favor it significantly. The 95.6% encryption lead matters for security-focused deployments. The 59% floating-point lead helps scientific computing and financial modeling.

The Intel Xeon 696X has one notable advantage in the raw numbers: its boost clock is higher at 4.80 GHz versus 4.50 GHz. Yet the recorded single-thread benchmark still goes to AMD by 2.9%. That suggests the AMD architecture extracts more work per clock, or the boost behavior differs in practice. The Intel part also has a lower TDP of 350 watts versus 400 watts, which could matter in power-constrained racks, though the database does not include wattage measurements for these tests.

For workloads that are heavily memory-bandwidth dependent, the EPYC 9655P offers 576.0 GB/s over twelve channels, versus 409.6 GB/s over eight channels for the Xeon. That is a 40.6% bandwidth advantage, consistent with the multi-core performance gaps. The Intel part has unlocked multiplier support, which is unusual for server hardware, while the EPYC is locked. That may appeal to enthusiasts, but the benchmark data does not show any overclocking results.

Architecture Differences

The AMD EPYC 9655P uses the Zen 5 architecture, codenamed Turin, built on a 4 nm process at TSMC. It packs 96 cores and 192 threads. The transistor count is 99,780 million across a die layout of 12x 70.6 mm². Cache is organized as 80 KB L1 per core, 1 MB L2 per core, and 384 MB shared L3. It uses the AMD Socket SP5.

The Intel Xeon 696X uses the Granite Rapids architecture, built on a 5 nm process at Intel. It has 64 cores and 128 threads. The die size is 2x 598 mm², with no transistor count recorded. Cache is 112 KB L1 per core, 2 MB L2 per core, and 336 MB shared L3. It uses the Intel Socket 4710.

The memory controllers differ significantly. AMD runs twelve DDR5 channels with 576.0 GB/s bandwidth. Intel runs eight DDR5 channels with 409.6 GB/s. Both support ECC memory. Both offer PCIe Gen 5 with 128 lanes (CPU only). Neither has integrated graphics.

The process node difference matters: 4 nm TSMC versus 5 nm Intel. The AMD chip fits more cores (96 vs 64) and more threads (192 vs 128) while using a larger shared L3 (384 MB vs 336 MB). The Intel chip has larger per-core L1 and L2 (112 KB and 2 MB versus 80 KB and 1 MB), which helps with per-thread locality, but the aggregate cache advantage goes to AMD.

Release dates differ: the EPYC 9655P launched on 2024-10-09, while the Xeon 696X launched on 2026-02-01. The Intel part is newer by over a year, yet it still trails in every recorded benchmark. The EPYC part number is 100-000001522; the Intel part number is SRWQ7.

FAQ

Q: Which CPU has more cores and threads?

A: The AMD EPYC 9655P has 96 cores and 192 threads. The Intel Xeon 696X has 64 cores and 128 threads. That is 50% more cores and 50% more threads for AMD.

Q: Is the Intel Xeon 696X faster in single-threaded workloads?

A: No. The PassMark single-thread score is 3,842 for the EPYC 9655P and 3,742 for the Xeon 696X, a 2.9% advantage for AMD. Despite the Intel part having a higher boost clock of 4.80 GHz versus 4.50 GHz, the recorded single-thread performance still favors AMD.

Q: What is the biggest performance gap between these two?

A: The PassMark physics test shows the largest delta at 664.3%. The EPYC 9655P scores 25,847 versus 3,382 for the Xeon. The smallest gap is in single-thread at 2.9%.

Q: How do memory bandwidth and channels compare?

A: The EPYC 9655P uses twelve-channel DDR5 with 576.0 GB/s. The Xeon 696X uses eight-channel DDR5 with 409.6 GB/s. AMD has 40.6% more bandwidth.

Q: Are both CPUs unlocked for overclocking?

A: No. The Intel Xeon 696X has an unlocked multiplier, while the AMD EPYC 9655P is locked. This is a rare feature on a server part.

Q: Which CPU has a higher TDP?

A: The AMD EPYC 9655P is rated at 400 watts. The Intel Xeon 696X is rated at 350 watts. The AMD part consumes more power, which may require stronger cooling.

The Verdict

The data points to a clear choice for raw compute performance: the AMD EPYC 9655P. It wins every single recorded benchmark, with margins ranging from 2.9% in single-thread to 664.3% in physics. For multi-core rendering, data compression, encryption, integer math, floating point, and string sorting, the AMD part is anywhere from 33.3% to 150.5% ahead. Its average benchmark score is 397,773 versus 286,102, placing it at the 100th percentile of all CPUs versus the 99th for Intel.

The Intel Xeon 696X does have advantages in the spec sheet: a higher boost clock (4.80 GHz), a lower TDP (350 W), an unlocked multiplier, and a larger per-core L1 and L2 cache. But none of those translate into a benchmark win in the recorded data. The newer release date (2026-02-01 versus 2024-10-09) does not help it overcome the architectural lead of Zen 5.

Who should pick the AMD EPYC 9655P? Anyone running multi-threaded server workloads: database servers, virtualization hosts, scientific computing, encryption-heavy services, or any environment where the 96 cores and 192 threads can be fully utilized. The 384 MB shared L3 and 576.0 GB/s memory bandwidth are substantial assets for large datasets.

Who should consider the Intel Xeon 696X? Those who need the unlocked multiplier for overclocking, which is rare in server CPUs. Also, those with strict power budgets, since the 350 W TDP is lower than the 400 W of the AMD part. But be aware that the performance gap in the benchmarks is massive, and the lower power draw does not compensate for a 52.8% multi-core deficit.

For most buyers, the AMD EPYC 9655P is the data-backed choice. The Intel part makes sense only if the unlocked multiplier or lower TDP are decisive for your specific deployment.

Specification Differences

| Specification | AMD EPYC 9655P | Intel Xeon 696X |

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

| Cores | 96 | 64 |

| Threads | 192 | 128 |

| Base Clock | 2.60 GHz | 2.40 GHz |

| Boost Clock | 4.50 GHz | 4.80 GHz |

| TDP | 400 W | 350 W |

| Socket | AMD Socket SP5 | Intel Socket 4710 |

| Architecture | Zen 5 | Granite Rapids |

| Codename | Turin | Granite Rapids |

| Process Node | 4 nm | 5 nm |

| Foundry | TSMC | Intel |

| Die Size | 12x 70.6 mm² | 2x 598 mm² |

| L1 Cache | 80 KB (per core) | 112 KB (per core) |

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

| L3 Cache | 384 MB (shared) | 336 MB (shared) |

| Memory Bus | Twelve-channel | Eight-channel |

| Memory Bandwidth | 576.0 GB/s | 409.6 GB/s |

| Unlocked Multiplier | No | Yes |

| Part Number | 100-000001522 | SRWQ7 |

| Release Date | 2024-10-09 | 2026-02-01 |

| Launch MSRP | $10811 | $5599 |

The launch MSRP is stated once for reference. The AMD part costs more, but the benchmark data shows it delivers substantially higher performance across all recorded tests.