AMD Ryzen Threadripper PRO 5965WX vs Intel Xeon 6731P Comparison

FAQ

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

A: The AMD Ryzen Threadripper PRO 5965WX wins 15 of the 17 recorded head-to-head tests, while the Intel Xeon 6731P wins only 2. The AMD part also holds a higher overall percentile rank, sitting at 97 compared to Intel's 96.

Q: How do their average benchmark scores compare?

A: The AMD Ryzen Threadripper PRO 5965WX has an average benchmark score of 98504, while the Intel Xeon 6731P averages 87756. The AMD processor's nearest rival, the AMD EPYC 4585PX, scores 99324, a delta of -0.8% relative to the Threadripper.

Q: What is the single biggest margin of victory in the head-to-head results?

A: The largest win for the AMD part is in Passmark single-thread testing, where it scores 3337 against Intel's 2107, a delta of 58.4%. For Intel, its biggest win is in Passmark physics, where it scores 7105 against AMD's 4251, a delta of -40.2% (meaning AMD trails by 40.2%).

Q: Are both processors designed for the same market segment?

A: Yes, both are listed as Server/Workstation parts. Both support ECC memory and both have a production status of Active. However, they use different sockets: the AMD uses Socket WRX8, and the Intel uses Socket 4710.

Q: How do their memory specifications differ?

A: The AMD processor supports DDR4 memory with a bandwidth of 204.8 GB/s, while the Intel processor supports DDR5 memory with a bandwidth of 409.6 GB/s. Both use an eight-channel memory bus. The Intel part's bandwidth is exactly double that of the AMD part.

Q: What is the difference in core and thread counts?

A: The Intel Xeon 6731P has 32 cores and 64 threads, while the AMD Ryzen Threadripper PRO 5965WX has 24 cores and 48 threads. Despite having fewer cores, the AMD part wins the vast majority of multi-threaded benchmark tests.

Architecture Differences

The AMD Ryzen Threadripper PRO 5965WX is built on the Zen 3 architecture, codenamed Chagall PRO, and fabricated by TSMC on a 7 nm process. It uses a multi-die design with a die size of 4x 81 mm² and contains 16,600 million transistors. The Intel Xeon 6731P is built on the Granite Rapids architecture (Granite Rapids-SP) and fabricated by Intel on a 5 nm process. Its die is a monolithic 598 mm², and the database does not record a transistor count for it.

Cache layouts differ substantially. The AMD part has 64 KB of L1 per core, 512 KB of L2 per core, and 128 MB of L3. The Intel part has 112 KB of L1 per core, 2 MB of L2 per core, and 144 MB of shared L3. This means the Intel processor has larger per-core L1 and L2 caches, and a larger total L3 pool, though the AMD part's L3 is more than 80% of Intel's capacity despite having fewer cores.

Memory architecture separates the two clearly. AMD uses DDR4 with eight-channel support and 204.8 GB/s of bandwidth. Intel uses DDR5 with eight-channel support and 409.6 GB/s of bandwidth. Both support ECC memory, which aligns with their server/workstation positioning.

PCIe capabilities differ as well. The AMD part offers Gen 4 with 128 lanes (CPU only), while the Intel part offers Gen 5 with 136 lanes (CPU only). The Intel part has a newer PCIe generation and 8 additional lanes.

Clock speeds favor AMD. The AMD processor has a base clock of 3.80 GHz and a boost clock of 4.50 GHz. The Intel processor has a base clock of 2.50 GHz and a boost clock of 4.10 GHz. The AMD part's base clock is 1.3 GHz higher, and its boost clock is 0.4 GHz higher.

Thermal design power favors Intel. The AMD part has a TDP of 280 W, while the Intel part has a TDP of 245 W. This is notable because the Intel part has 8 more cores and 16 more threads, yet is rated for 35 W less.

The AMD part does not have integrated graphics recorded, while the Intel part lists integrated graphics as N/A. Neither processor has an unlocked multiplier. The AMD part was released on 2022-03-07, and the Intel part on 2025-02-23.

The Verdict

The recorded data shows a clear winner for general compute workloads: the AMD Ryzen Threadripper PRO 5965WX. It wins 15 of 17 head-to-head tests, including all Cinebench R15, R20, and R23 tests, both single-core and multi-core. It also wins Passmark integer math by 41.5%, data encryption by 57.4%, and single-thread tests by 58.4%. Its average benchmark score of 98504 is 12.2% higher than Intel's 87756.

Who should pick the AMD part? Anyone whose workload maps to the benchmarks where AMD dominates: Cinebench rendering, integer math, data compression, encryption, and single-threaded responsiveness. The AMD part's higher clock speeds (3.80 base, 4.50 boost) and higher per-core performance drive these wins. Its 97th percentile ranking versus Intel's 96th percentile supports the same conclusion.

Who should pick the Intel part? The data points to two specific areas. First, Passmark physics: Intel scores 7105 against AMD's 4251, a 40.2% advantage. This is the largest Intel win in the entire dataset. Second, Passmark find prime numbers: Intel scores 541 against AMD's 523, a 3.3% edge. These are narrow and workload-specific wins, but they are real.

The Intel part also offers architectural advantages that the benchmarks do not directly capture: DDR5 memory with double the bandwidth (409.6 GB/s versus 204.8 GB/s), PCIe Gen 5 with 136 lanes versus Gen 4 with 128 lanes, and a lower TDP of 245 W versus 280 W. For workloads that are memory-bandwidth-bound or require the latest PCIe connectivity, the Intel part has structural appeal. But for raw compute throughput across the tested metrics, the AMD processor is the stronger choice.

Specification Differences

| Specification | AMD Ryzen Threadripper PRO 5965WX | Intel Xeon 6731P |

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

| Cores | 24 | 32 |

| Threads | 48 | 64 |

| Base clock | 3.80 GHz | 2.50 GHz |

| Boost clock | 4.50 GHz | 4.10 GHz |

| TDP | 280 W | 245 W |

| Socket | AMD Socket WRX8 | Intel Socket 4710 |

| Architecture | Zen 3 | Granite Rapids |

| Process node | 7 nm | 5 nm |

| Foundry | TSMC | Intel |

| Die size | 4x 81 mm² | 598 mm² |

| Transistors | 16,600 million | Not recorded |

| L1 cache | 64 KB per core | 112 KB per core |

| L2 cache | 512 KB per core | 2 MB per core |

| L3 cache | 128 MB | 144 MB (shared) |

| Memory support | DDR4 | DDR5 |

| Memory bandwidth | 204.8 GB/s | 409.6 GB/s |

| PCIe | Gen 4, 128 lanes | Gen 5, 136 lanes |

| Integrated graphics | Not recorded | N/A |

| Release date | 2022-03-07 | 2025-02-23 |

| Launch MSRP | $2399 | $2700 |

Head-to-Head Benchmarks

The Cinebench results are uniform across the board. In R15 multi-core, AMD scores 5685 against Intel's 4522, a 25.7% delta. In R15 single-core, AMD scores 802 against Intel's 638, also 25.7%. The same 25.7% margin repeats in R20 multi-core (23688 vs 18845), R20 single-core (3344 vs 2660), R23 multi-core (56400 vs 44871), and R23 single-core (7962 vs 6334). This consistency suggests a stable per-core advantage for the AMD architecture across all Cinebench workloads, regardless of thread count.

Passmark multithread follows the same pattern: AMD scores 66353, Intel scores 52790, a 25.7% delta. This is the same margin seen in Cinebench, reinforcing that the AMD part's advantage scales evenly from single-threaded to fully multi-threaded tests.

The largest AMD wins are in Passmark single-thread (3337 vs 2107, 58.4% delta) and Passmark data encryption (63113 vs 40087, 57.4% delta). These are massive margins, more than double the typical Cinebench gap. Integer math also shows a large AMD advantage: 281247 vs 198761, a 41.5% delta. Data compression favors AMD by 26.3% (1010067 vs 799474), and random string sorting favors AMD by 12.8% (99287 vs 88019).

Smaller AMD wins include extended instructions (67300 vs 65656, 2.5% delta) and floating point math (157708 vs 157330, 0.2% delta). The floating point margin is essentially negligible, indicating parity in that specific workload.

The Intel wins are concentrated in two areas. Passmark physics is the standout: Intel scores 7105 against AMD's 4251, a 40.2% advantage. This is the only benchmark where Intel leads by a large margin. Passmark find prime numbers is the other Intel win, but it is narrow: 541 vs 523, a 3.3% delta.

Where Each One Wins

The AMD Ryzen Threadripper PRO 5965WX wins in every Cinebench test, both single-core and multi-core, across R15, R20, and R23. It also wins Passmark multithread, single-thread, integer math, data compression, data encryption, extended instructions, floating point math, and random string sorting. This is a broad sweep across rendering, encryption, compression, sorting, and general integer workloads.

The AMD part's wins are not marginal. The median delta across its 15 wins is 25.7%, and several wins exceed 40%. The single-thread and encryption margins (58.4% and 57.4%) are particularly striking for a workstation processor. The data indicates that the AMD part is the better choice for Cinebench-based rendering, Passmark CPU benchmarks, and any workload that relies on integer throughput or single-thread performance.

The Intel Xeon 6731P wins in Passmark physics by a wide margin (7105 vs 4251, 40.2%) and in Passmark find prime numbers by a small margin (541 vs 523, 3.3%). These are distinct workloads. Physics simulations and prime number calculations appear to favor the Intel architecture despite its lower clock speeds. The Intel part also has structural advantages in memory bandwidth (409.6 GB/s DDR5 versus 204.8 GB/s DDR4), PCIe generation (Gen 5 versus Gen 4), and lane count (136 versus 128). For workloads that stress physics computation or require maximum memory throughput and the latest PCIe devices, the Intel part is the data-supported choice.

For everything else in the recorded benchmark set, the AMD part wins, often by substantial margins. The data supports choosing AMD for Cinebench rendering, encryption, compression, integer math, and general multithreaded and single-threaded performance. The Intel part is the pick for physics-heavy workloads and prime number searches, with the additional caveat that its DDR5 and PCIe Gen 5 support may matter more in real-world deployments than these synthetic tests reveal.