AMD EPYC 4565P vs Intel Xeon 654 Comparison

The AMD EPYC 4565P and Intel Xeon 654 are both high-end server processors, yet the benchmark data reveals a clear split in their performance profiles. The AMD part dominates in most general-purpose and single-threaded workloads, while the Intel chip carves out a decisive lead in specific, numerically intensive tasks. This analysis breaks down the head-to-head results, answers common questions, and identifies which processor suits which workload based strictly on the data.

Head-to-Head Benchmarks

The most striking pattern in the head-to-head results is the consistency of the AMD EPYC 4565P's victory in the Cinebench suite. Across all six Cinebench tests (R15, R20, and R23, both single-core and multi-core), the AMD processor wins with an identical 4.3% margin. For example, in Cinebench R23 multi-core, the EPYC 4565P scores 54,405 against the Xeon 654's 52,150. The single-core results are equally one-sided, with the EPYC 4565P posting 7,680 in R23 single-core versus 7,362 for the Intel part. This uniform 4.3% delta suggests a fundamental per-clock or per-core efficiency advantage for the AMD architecture in rendering and 3D modeling tasks.

Beyond Cinebench, the AMD EPYC 4565P extends its lead in several Passmark workloads. The biggest win is in Passmark single-thread, where the AMD part scores 4,712 compared to the Intel Xeon 654's 3,778, a substantial 24.7% advantage. This is the largest single delta in the entire comparison and indicates that the EPYC 4565P is dramatically faster for lightly threaded, latency-sensitive tasks. Data encryption also favors AMD heavily, with the EPYC 4565P scoring 48,268 against 40,675 for the Xeon 654, an 18.7% lead. Integer math is another major AMD win: 250,683 versus 207,745, a 20.7% margin. Even in data compression, the AMD part leads by 5.1% (860,786 vs. 818,902), and in the Passmark multithread test, it wins by 3.5% (63,474 vs. 61,353). In total, the AMD EPYC 4565P wins 13 of the 17 head-to-head benchmarks.

The Intel Xeon 654 does not go down without a fight, however, and its four wins are notable for their magnitude. The most dramatic Intel victory is in Passmark physics, where the Xeon 654 scores 5,596 against the EPYC 4565P's 2,976. That is a 46.8% advantage for Intel, the single largest delta in either direction in the entire dataset. This suggests the Xeon 654 has a vastly superior floating-point or physics simulation capability. The Intel part also wins in Passmark find prime numbers, scoring 390 versus 294, a 24.6% lead. In floating-point math, the Xeon 654 scores 163,093 versus 152,003 for AMD, a 6.8% margin. Finally, the Xeon 654 edges out the EPYC 4565P in random string sorting by a slim 1.8% (82,828 vs. 81,318).

Looking at the broader context, the AMD EPYC 4565P has an average benchmark score of 95,764, which places it in the 96th percentile of all CPUs. Its nearest rival is the AMD Ryzen 9 PRO 9945, which scores 96,083, a mere 0.3% higher. The Intel Xeon 654, by contrast, has an average score of 90,717, also in the 96th percentile, with its closest competitor being the AMD Ryzen AI Max+ 392 at 90,541. This means that despite the Xeon 654's specific wins, its overall average performance is about 5.5% lower than the EPYC 4565P's average (90,717 vs. 95,764).

FAQ

Q: Which processor wins the majority of the head-to-head benchmarks?

A: The AMD EPYC 4565P wins 13 out of 17 benchmarks, while the Intel Xeon 654 wins 4.

Q: What is the biggest performance gap in either direction?

A: The largest gap is in Passmark physics, where the Intel Xeon 654 leads by 46.8% (5,596 vs. 2,976). The largest AMD lead is 24.7% in Passmark single-thread (4,712 vs. 3,778).

Q: How do the two compare in Cinebench R23 multi-core?

A: The AMD EPYC 4565P scores 54,405, which is 4.3% higher than the Intel Xeon 654's 52,150.

Q: Is the Intel Xeon 654 faster in any memory-intensive tasks?

A: Yes, the Xeon 654 wins in Passmark random string sorting by 1.8% (82,828 vs. 81,318), although this is a narrow margin.

Q: Which processor has a higher average benchmark score?

A: The AMD EPYC 4565P has an average benchmark score of 95,764, while the Intel Xeon 654 has an average score of 90,717.

Q: Are both processors in the same performance percentile?

A: Yes, both the AMD EPYC 4565P and the Intel Xeon 654 are in the 96th percentile of all CPUs.

The Verdict

The data presents a clear recommendation for most users: the AMD EPYC 4565P is the superior all-around processor. It wins the vast majority of benchmarks, including all Cinebench tests, and has a higher average score by approximately 5,000 points. Its 24.7% lead in single-thread performance makes it the obvious choice for workloads where a single core's speed is the bottleneck. The 18.7% lead in data encryption and 20.7% lead in integer math further solidify its position for general server tasks, compression, and security-related workloads.

The Intel Xeon 654 is not without its niche, however. The 46.8% lead in Passmark physics is overwhelming and suggests that the Xeon 654 is the preferred option for physics simulations, certain types of scientific computing, or any workload that heavily taxes floating-point physics calculations. The 24.6% lead in prime number finding also points to a strength in specific algorithmic or mathematical workloads. For users whose primary application is physics-based simulation, the Intel part's specialization may outweigh its overall lower average score.

In summary, the AMD EPYC 4565P is the default recommendation for a balanced server or workstation processor, based on its broad benchmark dominance. The Intel Xeon 654 should only be selected if the workload is specifically known to benefit from its physics and prime-number performance advantages. The data does not support choosing the Xeon 654 for general-purpose computing.

Specification Differences

The core counts and clock speeds diverge significantly. The Intel Xeon 654 has 18 cores and 36 threads, while the AMD EPYC 4565P has 16 cores and 32 threads. However, the AMD part operates at a higher frequency: a base clock of 4.30 GHz and a boost clock of 5.70 GHz, compared to the Xeon 654's 3.10 GHz base and 4.80 GHz boost. This clock advantage explains the AMD part's single-thread dominance. Thermal design power also differs, with the Xeon 654 rated at 200 TDP and the EPYC 4565P at 170 TDP.

The memory interface is a major point of difference. The Intel Xeon 654 uses an eight-channel memory bus, providing a theoretical memory bandwidth of 409.6 GB/s. The AMD EPYC 4565P uses a dual-channel bus with 89.6 GB/s bandwidth. This massive memory bandwidth advantage for Intel likely contributes to its wins in physics and floating-point math. PCIe lane counts also differ: the Xeon 654 offers 128 Gen 5 lanes, while the EPYC 4565P provides 24 Gen 5 lanes. The Intel part also has an unlocked multiplier, whereas the AMD part is locked. Launch MSRP for the AMD EPYC 4565P is $589, and the Intel Xeon 654 is $1199.

Architecture Differences

The two processors are built on fundamentally different designs. The AMD EPYC 4565P uses the Zen 5 architecture, codenamed Grado, and is fabricated on a 4 nm process by TSMC. The Intel Xeon 654 uses the Granite Rapids architecture, also its codename, and is fabricated on a 5 nm process by Intel. The AMD chip packages 16,630 million transistors across a die size of 2x 70.6 mm². The Intel die is much larger at 2x 598 mm², but the transistor count is not listed in the data.

Cache hierarchies are also distinct. The AMD EPYC 4565P has 80 KB of L1 cache and 1 MB of L2 cache per core, with 64 MB of shared L3 cache. The Intel Xeon 654 has a larger per-core L1 cache at 112 KB and L2 cache at 2 MB per core, with a larger shared L3 cache of 72 MB. Both support DDR5 memory and implement ECC memory. The AMD part features integrated Radeon Graphics, while the Intel part has no integrated graphics. The AMD EPYC 4565P is part of the EPYC 4005 series and uses the AMD Socket AM5, while the Intel Xeon 654 is in the Xeon 600 series and uses Intel Socket 4710.

Where Each One Wins

AMD EPYC 4565P Wins: The data shows the AMD part is the clear winner in all Cinebench rendering workloads (R15, R20, R23), both single and multi-threaded. It also dominates in Passmark single-thread, integer math, data encryption, data compression, and the overall multithread test. This makes it the superior choice for general server virtualization, databases, web serving, software compilation, and any workload where encryption or integer arithmetic is prevalent. Its high clock speed also makes it ideal for tasks that cannot leverage many cores, as evidenced by its 24.7% single-thread lead.

Intel Xeon 654 Wins: The Intel Xeon 654's victories are concentrated in four specific areas: Passmark physics (46.8% lead), Passmark find prime numbers (24.6% lead), Passmark floating-point math (6.8% lead), and Passmark random string sorting (1.8% lead). These are computationally intensive, often floating-point-heavy tasks. The eight-channel memory interface with 409.6 GB/s bandwidth is likely the key enabler for the physics and floating-point wins. Therefore, the Xeon 654 is the better choice for scientific simulations, physics engines, financial modeling that relies on floating-point calculations, and other high-bandwidth memory workloads. Its 128 PCIe Gen 5 lanes also make it more suitable for systems requiring massive I/O expansion, such as multiple GPUs or high-speed network cards, although this is not directly benchmarked.