AMD EPYC 9565 vs Intel Xeon 696X Comparison

# Intel Xeon 696X vs AMD EPYC 9565

The Intel Xeon 696X and AMD EPYC 9565 are both 99th-percentile server processors, but the benchmark data tells a lopsided story. The EPYC 9565 dominates the Xeon 696X in nearly every multi-threaded workload, while the Xeon only manages a narrow single-thread victory. Average benchmark scores are essentially tied — 286,102 for the Intel part versus 285,471 for the AMD part, a delta of just 0.2% — but that near-parity hides the fact that the two chips win in completely different categories. The EPYC 9565 takes 12 of 14 head-to-head benchmarks, with the Xeon 696X winning only the two single-thread tests.

Head-to-Head Benchmarks

The most striking gap appears in Cinebench multi-core testing. Across R15, R20, and R23, the EPYC 9565 beats the Xeon 696X by the same 22.4% margin in every iteration: 11,585 versus 8,994 in R15, 48,273 versus 37,475 in R20, and 114,937 versus 89,227 in R23. That consistency suggests a fundamental throughput advantage rather than a workload-specific quirk.

PassMark's multi-thread score confirms the pattern, with the EPYC 9565 reaching 135,221 against the Xeon's 104,974 — again a 22.4% gap. Integer math shows a 20.3% lead for AMD (717,948 versus 572,072), and floating-point math follows at 18.1% (549,422 versus 450,164). Data encryption favors the EPYC 9565 by 20.7% (141,936 versus 112,529), while data compression shows a smaller but still decisive 12.2% advantage (2,579,631 versus 2,264,907).

The largest single deltas are in physics and prime number finding. PassMark physics scores the EPYC 9565 at 18,036 versus just 3,382 for the Xeon 696X — an 81.2% blowout. Prime number generation is similarly one-sided: 2,422 versus 853, a 64.8% deficit for Intel. Random string sorting also heavily favors AMD at 291,941 versus 180,392, a 38.2% gap. Extended instructions testing gives the EPYC 9565 a 17.5% edge (209,595 versus 172,975).

The Xeon 696X's only wins come in single-thread performance. PassMark single-thread scores show 3,742 for Intel versus 3,696 for AMD, a slim 1.2% margin. This is a real but minor advantage — enough to edge out the EPYC 9565 in lightly threaded scenarios, but nowhere near enough to offset the multi-core deficits. The near-equal average scores (0.2% apart) reflect this trade-off: the Xeon's single-thread win barely registers against AMD's broad multi-core sweep.

Architecture Differences

The two processors take fundamentally different design approaches. The Intel Xeon 696X uses Granite Rapids architecture on a 5 nm process built in-house by Intel, while the AMD EPYC 9565 uses Zen 5 architecture on a 4 nm process from TSMC. The AMD part is the newer release, launching on 2024-10-09 versus the Xeon's 2026-02-01.

Core counts differ substantially. The EPYC 9565 packs 72 cores and 144 threads, while the Xeon 696X offers 64 cores and 128 threads. That 8-core advantage helps explain AMD's multi-thread dominance. Clock speeds tell a different story: the Xeon 696X boosts to 4.80 GHz from a 2.40 GHz base, while the EPYC 9565 boosts to 4.30 GHz from a 3.15 GHz base. The Xeon's higher boost clock likely contributes to its single-thread win, despite the EPYC's higher base frequency.

Cache configurations also diverge. The Xeon 696X has 112 KB of L1 and 2 MB of L2 per core, with 336 MB of shared L3. The EPYC 9565 has 80 KB L1 and 1 MB L2 per core, but a larger 384 MB shared L3. The EPYC's 48 MB extra L3 cache is notable for workloads with large working sets.

Memory architecture favors AMD as well. The EPYC 9565 supports twelve-channel DDR5 with 576.0 GB/s bandwidth, versus eight-channel DDR5 at 409.6 GB/s for the Xeon 696X. Both support ECC memory and offer PCIe Gen 5 with 128 lanes from the CPU. The EPYC 9565 uses AMD Socket SP5, while the Xeon 696X uses Intel Socket 4710. Transistor counts and die layouts differ dramatically: the EPYC 9565 uses 99,780 million transistors across 12 chiplets of 70.6 mm² each, while the Xeon 696X uses a dual-die design with 2x 598 mm².

FAQ

Q: Which processor has more cores and threads?

A: The AMD EPYC 9565 has 72 cores and 144 threads, compared to the Intel Xeon 696X's 64 cores and 128 threads.

Q: Is the Intel Xeon 696X faster in any benchmark?

A: Yes, it wins PassMark single-thread tests with a score of 3,742 versus the EPYC 9565's 3,696, a 1.2% advantage.

Q: How large is the multi-thread performance gap?

A: The EPYC 9565 leads by 22.4% across all three Cinebench multi-core tests (R15, R20, R23) and in PassMark multi-thread, with scores of 11,585, 48,273, 114,937, and 135,221 respectively versus the Xeon's 8,994, 37,475, 89,227, and 104,974.

Q: Which CPU has higher memory bandwidth?

A: The AMD EPYC 9565 offers 576.0 GB/s over a twelve-channel DDR5 bus, while the Intel Xeon 696X provides 409.6 GB/s over eight-channel DDR5.

Q: What are the launch MSRPs?

A: The Intel Xeon 696X has a launch MSRP of $5599, and the AMD EPYC 9565 has a launch MSRP of $10486.

Q: Which CPU has a larger L3 cache?

A: The AMD EPYC 9565 has 384 MB of shared L3 cache, compared to 336 MB on the Intel Xeon 696X.

The Verdict

The data supports a clear conclusion: the AMD EPYC 9565 is the superior multi-threaded processor, winning 12 of 14 head-to-head benchmarks with margins ranging from 12.2% to 81.2%. Its 22.4% Cinebench advantage across all versions, combined with similarly large leads in integer math, floating-point math, encryption, and physics, makes it the obvious choice for heavily parallel workloads. The Xeon 696X's only meaningful win is a 1.2% single-thread margin, which is negligible for most server and workstation tasks. The near-identical average benchmark scores (0.2% difference) might suggest a toss-up, but the distribution of wins tells the real story — AMD dominates where it matters most for compute-heavy applications. The EPYC 9565 also has a 48 MB L3 cache advantage and 166.4 GB/s more memory bandwidth, reinforcing its position for data-intensive work.

Specification Differences

| Specification | Intel Xeon 696X | AMD EPYC 9565 |

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

| Cores | 64 | 72 |

| Threads | 128 | 144 |

| Base Clock | 2.40 GHz | 3.15 GHz |

| Boost Clock | 4.80 GHz | 4.30 GHz |

| TDP | 350 W | 400 W |

| Socket | Intel Socket 4710 | AMD Socket SP5 |

| Architecture | Granite Rapids | Zen 5 |

| Codename | Granite Rapids | Turin |

| Process Node | 5 nm | 4 nm |

| Foundry | Intel | TSMC |

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

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

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

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

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

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

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

| Launch MSRP | $5599 | $10486 |

| Multiplier Unlocked | Yes | No |

Where Each One Wins

The AMD EPYC 9565 is the clear pick for multi-threaded compute workloads. Its 22.4% lead in Cinebench R15, R20, and R23 makes it the better choice for 3D rendering, video encoding, and simulation tasks that scale across cores. The 81.2% physics benchmark advantage and 64.8% prime-number lead point to strength in scientific computing and financial modeling. Data encryption performance (20.7% ahead) suits security-focused workloads, while the 38.2% random string sorting advantage helps with database and data processing tasks. The larger 384 MB L3 cache and 576.0 GB/s memory bandwidth further support these data-heavy use cases.

The Intel Xeon 696X wins only in single-thread performance, with a 1.2% edge in PassMark's single-thread test. This makes it marginally better for lightly threaded applications where per-core speed matters, such as some legacy software or single-threaded scripting workloads. Its higher 4.80 GHz boost clock also gives it headroom for bursty workloads. However, the narrow margin means this advantage is unlikely to be noticeable in practice. For any workload that can utilize more than a couple of threads — which is the norm for server and workstation processors in this class — the EPYC 9565 is the data-backed choice. The Xeon 696X's lower TDP of 350 W versus 400 W may appeal to power-constrained deployments, but the EPYC's performance advantages dominate the comparison.