AMD EPYC 9754 vs Intel Xeon 6747P Comparison

The AMD EPYC 9754 and Intel Xeon 6747P represent two distinct approaches to server processing, and the benchmark data shows a clear split in their respective strengths. The AMD EPYC 9754, with its 128 cores, dominates in throughput-oriented tasks that scale with core count, while the Intel Xeon 6747P, with fewer but faster cores, takes the lead in single-threaded and latency-sensitive workloads. The database records 8 wins for the Intel Xeon 6747P and 6 for the AMD EPYC 9754 across the 14 head-to-head benchmarks, but the magnitude of those wins tells a more nuanced story than the raw win count suggests.

Where Each One Wins

The AMD EPYC 9754 is the clear choice for workloads that can fully utilize its massive 256-thread capacity. Its wins come in categories that reward raw parallel processing. The most dramatic victory is in data encryption, where the AMD part scores 155.4% higher than the Intel Xeon 6747P. Integer math shows a 119.2% advantage for AMD, and data compression shows a 94.1% lead. These are not marginal differences; they represent more than double the throughput in some cases. The EPYC 9754 also leads in floating-point math by 60.7%, extended instructions by 57.4%, and random string sorting by 69.9%. For any server workload that involves bulk data transformation, compression, or cryptographic operations, the AMD EPYC 9754 is the superior processor according to the recorded data.

The Intel Xeon 6747P wins in a different category: tasks that depend on single-core efficiency and low-latency execution. The most notable win is in Pass find prime numbers, where Intel leads by 47.5%. This is a workload that responds directly to clock speed and memory latency rather than core count. The Xeon 6747P also wins in Pass physics by 34.4%, which typically measures per-core physics simulation performance. Most importantly, the Intel Xeon 6747P wins all three Cinebench multicore tests (R15, R20, R23) by a consistent 2.9% margin, despite having 80 fewer cores. This indicates that the Intel architecture is more efficient per core in rendering-style workloads. The single-thread benchmark shows Intel leading by 28.1%, which confirms its advantage in any application that cannot scale beyond a single thread.

Architecture Differences

The fundamental architectural split is between AMD's Zen 4c (Bergamo) design and Intel's Granite Rapids. The AMD EPYC 9754 uses a 5 nm process from TSMC, while the Intel Xeon 6747P uses a 5 nm process from Intel, so the process node is nominally the same, but the implementation differs. The AMD chip is built on a chiplet design with 8 separate dies, each measuring 73 mm², for a total die area of 584 mm². The Intel Xeon 6747P uses two dies, each measuring 598 mm², which is a much larger monolithic block per die.

The core counts are radically different: 128 cores for the AMD versus 48 for Intel, with 256 threads versus 96 threads. The AMD EPYC 9754 has 64 KB of L1 cache and 1 MB of L2 cache per core, while the Intel has 112 KB of L1 and 2 MB of L2 per core. The L3 cache tells a different story: AMD has 256 MB shared, which is larger in total, but Intel has 288 MB shared, which is even larger. Memory support differs as well: AMD uses a twelve-channel DDR5 memory bus with a peak bandwidth of 460.8 GB/s, while Intel uses an eight-channel DDR5 bus with 409.6 GB/s. The AMD EPYC 9754 provides 128 PCIe Gen 5 lanes (CPU only), while the Intel Xeon 6747P provides 88 lanes. Both support ECC memory, but the AMD has no integrated graphics, while the Intel lists integrated graphics as "N/A".

The cache hierarchy is a key differentiator. AMD's larger L2 and L3 caches are distributed across 128 cores, so each core gets less dedicated cache. Intel's larger L1 and L2 per core, combined with a larger L2 total, suggest a design aimed at single-thread performance. The AMD EPYC 9754's 71,000 million transistors spread across 8 small dies contrasts with Intel's two large dies, which likely explains the thermal and power differences despite both being 5 nm.

FAQ

Q: Which processor has more cores, and how does that affect performance?

A: The AMD EPYC 9754 has 128 cores and 256 threads, while the Intel Xeon 6747P has 48 cores and 96 threads. The AMD's core advantage leads to massive wins in parallel workloads like data compression (94.1% lead) and integer math (119.2% lead), but the Intel's fewer cores still win Cinebench multicore tests by 2.9%, showing a core-for-core efficiency advantage.

Q: Which chip is better for single-threaded applications?

A: The Intel Xeon 6747P is clearly better for single-threaded tasks. It wins the Passmark single-thread benchmark by 28.1%, scoring 3236 versus 2328. Its higher base clock of 2.70 GHz and boost clock of 3.90 GHz, compared to 2.25 GHz and 3.10 GHz on the AMD, directly support this result.

Q: How do they compare in memory bandwidth?

A: The AMD EPYC 9754 has a twelve-channel memory bus and a peak bandwidth of 460.8 GB/s, which is 51.2 GB/s higher than the Intel Xeon 6747P's 409.6 GB/s. However, the Intel's eight-channel bus in practice still delivers competitive results in memory-sensitive benchmarks like prime number calculation and Cinebench.

Q: What is the total cache capacity difference?

A: The AMD EPYC 9754 has 256 MB of shared L3 cache, while the Intel Xeon 6747P has 288 MB of shared L3 cache. The Intel also has 2 MB of L2 per core versus 1 MB on the AMD, and 112 KB of L1 per core versus 64 KB on the AMD. The larger cache per core on the Intel contributes to its single-thread advantage.

Q: Which processor has more PCIe lanes?

A: The AMD EPYC 9754 provides 128 PCIe Gen 5 lanes (CPU only), whereas the Intel Xeon 6747P provides 88 PCIe Gen 5 lanes (CPU only). This gives the AMD a 40-lane advantage for expansion cards, storage, or accelerators.

Q: How do the overall benchmark averages compare?

A: The AMD EPYC 9754 has an average benchmark score of 364,371, placing it in the 100th percentile of all CPUs, while the Intel Xeon 6747P has an average score of 238,263, placing it in the 99th percentile. The AMD's average is 52.9% higher, but this average is heavily weighted by the AMD's dominance in parallel tasks.

Specification Differences

The key differences between the two processors are stark and directly explain their benchmark profiles.

• Cores: AMD EPYC 9754 has 128 cores, Intel Xeon 6747P has 48 cores.
• Threads: AMD has 256 threads, Intel has 96 threads.
• Base Clock: AMD runs at 2.25 GHz, Intel at 2.70 GHz.
• Boost Clock: AMD boosts to 3.10 GHz, Intel to 3.90 GHz.
• Cache (L1 per core): AMD has 64 KB, Intel has 112 KB.
• Cache (L2 per core): AMD has 1 MB, Intel has 2 MB.
• Cache (L3 total): AMD has 256 MB, Intel has 288 MB.
• Memory Bus: AMD is twelve-channel, Intel is eight-channel.
• Memory Bandwidth: AMD reaches 460.8 GB/s, Intel reaches 409.6 GB/s.
• PCIe Lanes: AMD has 128 Gen 5 lanes, Intel has 88 Gen 5 lanes.
• Die Size: AMD uses 8x 73 mm² dies (total 584 mm²), Intel uses 2x 598 mm² dies (total 1196 mm²).
• Transistor Count: AMD lists 71,000 million transistors, Intel does not have a recorded figure.
• Socket: AMD uses Socket SP5, Intel uses Socket 4710.
• Release Date: AMD launched on June 12, 2023, Intel on February 23, 2025.

The AMD EPYC 9754 has a TDP of 360 watts, while the Intel Xeon 6747P has a TDP of 330 watts. The Intel's higher clocks and larger L2 cache per core contradict its lower core count, but the AMD's massive core count gives it a raw throughput advantage. The AMD has 40 more PCIe lanes and 51.2 GB/s more memory bandwidth, but the Intel has a larger L3 cache and more cache per core.

Head-to-Head Benchmarks

The 14 head-to-head benchmarks show a striking pattern: the Intel Xeon 6747P wins all Cinebench multicore tests, but the AMD EPYC 9754 wins all Passmark parallel tests. The Cinebench results are consistent: the Intel wins R15 multicore (8712 vs 8460, a 2.9% lead), R20 multicore (36301 vs 35254, a 2.9% lead), and R23 multicore (86432 vs 83939, a 2.9% lead). These are significant because they show that even in a highly threaded rendering workload, the Intel can outperform a processor with 80 more cores.

The Passmark results are where the AMD dominates. The most extreme win is in data encryption, where the AMD scores 231,891 versus 90,789, a 155.4% advantage. Integer math shows an even larger absolute gap: 1,026,896 for AMD versus 468,518 for Intel, a 119.2% difference. Data compression follows with 3,558,043 for AMD versus 1,833,378 for Intel, a 94.1% lead. The AMD also wins in floating-point math (588,187 vs 365,904, 60.7% higher), extended instructions (224,322 vs 142,557, 57.4% higher), and random string sorting (306,481 vs 180,382, 69.9% higher).

The Intel wins in three specific tests. The Passmark find prime numbers score is 1151 for Intel versus 604 for AMD, a 47.5% lead. This is a classic single-thread test. Passmark physics shows Intel at 13,398 versus AMD at 8,793, a 34.4% advantage. And in the Passmark single-thread and singlethread tests, both of which score 3236 for Intel and 2328 for AMD, the Intel leads by 28.1%. The Passmark multithread test is close, with Intel winning by 2.9% (101,685 versus 98,752).

The data indicates that the AMD EPYC 9754 is a throughput monster for data-heavy workloads, but the Intel Xeon 6747P is a faster processor per core and wins in tasks that require high individual core performance. The 2.9% margin in Cinebench and the 28.1% margin in single-thread are not trivial; they show the Intel is 3.5% faster in clock speed (3.90 GHz vs 3.10 GHz boost) and that this translates into real performance gains in latency-bound software.

The Verdict

Based on the recorded data, the choice between the AMD EPYC 9754 and the Intel Xeon 6747P depends entirely on the workload profile. For server environments that handle large-scale data compression, encryption, or mathematical transformations on massive datasets, the AMD EPYC 9754 is the correct pick. Its 94.1% lead in data compression and 155.4% lead in encryption mean that any database, analytics, or storage workload will see more than double the throughput from the AMD. The AMD also has 128 PCIe lanes and higher memory bandwidth, which are advantageous for systems with many storage drives or accelerators.

For workloads that require high single-thread performance, such as legacy software, some simulation tasks, or applications that are not well-parallelized, the Intel Xeon 6747P is the better choice. The 28.1% single-thread lead and 47.5% prime-number win are clear indicators. The Intel also wins in every Cinebench test, so if rendering or modeling software that runs on a single core is the primary use case, the Intel will be faster despite having 80 fewer cores. The Intel's 288 MB L3 cache and 2 MB L2 per core also suggest it will handle memory-hungry single-threaded software better.

In summary, the database shows a typical trade-off: the AMD EPYC 9754 is a massive parallel processor that prioritizes total throughput, and the Intel Xeon 6747P is a high-clock processor that prioritizes per-core speed. The AMD is the right tool for scale-out, data-density workloads, while the Intel is the right tool for scale-up, latency-critical workloads. Neither is a universal winner; the 6 AMD mentions versus 8 Intel mentions in the head-to-head data does not reflect the magnitude of the AMD's wins, which are often several times larger. The decision should be based on whether your application can use 128 cores effectively. If it can, the AMD will crush the Intel. If it cannot, the Intel will be more responsive.