AMD EPYC 9965 vs Intel Xeon 6781P Comparison

# AMD EPYC 9965 vs Intel Xeon 6781P

The AMD EPYC 9965 and Intel Xeon 6781P occupy different tiers of the server CPU market, with the EPYC 9965 delivering overwhelming multi-threaded dominance while the Xeon 6781P offers a narrower but notable advantage in one specific workload. The database records 14 head-to-head benchmarks between these two processors, with the EPYC 9965 winning 13 and the Xeon 6781P winning 1. The EPYC 9965 sits at the 100th percentile among all CPUs in the database, while the Xeon 6781P sits at the 99th percentile, indicating both are elite performers but with a clear performance gap between them.

Where Each One Wins

The AMD EPYC 9965 is the clear winner in virtually every multi-threaded and throughput-oriented workload. Its largest victories come in integer math, where it scores 229.3% higher than the Xeon 6781P, and data encryption, where it leads by 191.3%. These are substantial margins that reflect the EPYC 9965's massive core count advantage of 192 cores versus 80 cores. The EPYC 9965 also wins decisively in data compression (132.6% ahead), random string sorting (135.7% ahead), and floating-point math (127.3% ahead). In the Cinebench suite, the EPYC 9965 maintains a consistent 36.3% advantage across R15, R20, and R23 multi-core tests.

The Intel Xeon 6781P's single win comes in the PassMark find prime numbers test, where it scores 1687 against the EPYC 9965's 1208, a 28.4% advantage. This is a specialized workload that measures raw integer throughput in a specific algorithmic pattern, and the Xeon's higher boost clock of 3.80 GHz versus 3.70 GHz likely contributes to this result. However, this single victory does little to offset the EPYC 9965's dominance in the broader benchmark suite.

Single-thread performance is nearly a tie. The EPYC 9965 scores 3176 in PassMark single-thread, just 0.8% ahead of the Xeon 6781P's 3152. This suggests that for lightly threaded workloads, the two processors perform essentially identically, making the EPYC 9965's massive multi-core advantage the deciding factor for almost any server workload.

Architecture Differences

The two processors represent fundamentally different design philosophies. The AMD EPYC 9965 uses the Zen 5 architecture under the codename "Turin," built on TSMC's 3 nm process node. It is part of the EPYC 9005 series and features 192 cores and 384 threads. The Intel Xeon 6781P uses the Granite Rapids architecture, built on Intel's 5 nm process node, and features 80 cores and 160 threads. The EPYC 9965's process node advantage allows AMD to pack more than twice the core count into a single socket.

Cache configurations differ significantly. The EPYC 9965 has 80 KB of L1 cache per core, 1 MB of L2 cache per core, and 384 MB of shared L3 cache. The Xeon 6781P has 112 KB of L1 per core, 2 MB of L2 per core, and 336 MB of shared L3 cache. While the Xeon has more cache per core, the EPYC 9965's larger total L3 cache (384 MB versus 336 MB) provides more aggregate cache capacity for data-intensive workloads.

Memory architecture also diverges. The EPYC 9965 supports DDR5 with a twelve-channel memory bus and delivers 576.0 GB/s of memory bandwidth. The Xeon 6781P also supports DDR5 but uses an eight-channel memory bus, providing 409.6 GB/s of memory bandwidth. The EPYC 9965's 40.6% higher memory bandwidth is a critical advantage for workloads that are sensitive to memory throughput.

PCIe connectivity favors Intel in lane count. The Xeon 6781P provides 136 Gen 5 lanes (CPU only), while the EPYC 9965 provides 128 Gen 5 lanes (CPU only). Both support ECC memory, and neither has integrated graphics. The Xeon 6781P uses a dual-die design with 2x 598 mm² die size, while the EPYC 9965's die size is not recorded in the database.

FAQ

Q: Which processor has more cores and threads?

A: The AMD EPYC 9965 has 192 cores and 384 threads, while the Intel Xeon 6781P has 80 cores and 160 threads. The EPYC 9965 provides 112 more cores and 224 more threads.

Q: What is the single-thread performance difference?

A: The EPYC 9965 scores 3176 in PassMark single-thread, and the Xeon 6781P scores 3152. The EPYC 9965 leads by 0.8%, which is effectively a tie for single-threaded workloads.

Q: Which processor has higher memory bandwidth?

A: The EPYC 9965 provides 576.0 GB/s over a twelve-channel DDR5 memory bus. The Xeon 6781P provides 409.6 GB/s over an eight-channel DDR5 bus. The EPYC 9965 has 166.4 GB/s more bandwidth.

Q: What is the boost clock difference?

A: The Xeon 6781P boosts to 3.80 GHz, while the EPYC 9965 boosts to 3.70 GHz. The Xeon has a 0.10 GHz higher boost clock, which contributes to its win in the prime numbers test.

Q: How do the processors compare in Cinebench multi-core performance?

A: The EPYC 9965 scores 136661 in Cinebench R23 multi-core, 57397 in R20, and 13775 in R15. The Xeon 6781P scores 100254, 42106, and 10105 respectively. The EPYC 9965 leads by 36.3% in each test.

Q: Which processor has a higher overall benchmark average?

A: The EPYC 9965 has an average benchmark score of 620487, while the Xeon 6781P has an average of 315524. The EPYC 9965's average is nearly double the Xeon's average.

Specification Differences

The EPYC 9965 and Xeon 6781P differ in nearly every specification category. The EPYC 9965 has 192 cores and 384 threads, while the Xeon 6781P has 80 cores and 160 threads. Base clocks are 2.25 GHz for the EPYC 9965 and 2.00 GHz for the Xeon 6781P, while boost clocks are 3.70 GHz and 3.80 GHz respectively. The EPYC 9965 has a TDP of 500 watts, while the Xeon 6781P has a TDP of 350 watts.

The EPYC 9965 uses AMD Socket SP5, while the Xeon 6781P uses Intel Socket 4710. The EPYC 9965 is built on a 3 nm process at TSMC, while the Xeon 6781P uses Intel's 5 nm process. The EPYC 9965 has 80 KB of L1 and 1 MB of L2 per core, versus 112 KB of L1 and 2 MB of L2 per core for the Xeon 6781P. L3 cache totals 384 MB for the EPYC 9965 and 336 MB for the Xeon 6781P.

Memory bandwidth differs substantially: 576.0 GB/s for the EPYC 9965 over twelve channels, versus 409.6 GB/s for the Xeon 6781P over eight channels. PCIe Gen 5 lane counts are 128 for the EPYC 9965 and 136 for the Xeon 6781P. Release dates are October 2024 for the EPYC 9965 and February 2025 for the Xeon 6781P. The launch MSRP is $14813 for the EPYC 9965 and $8960 for the Xeon 6781P.

Head-to-Head Benchmarks

The largest margin in the head-to-head results is in PassMark integer math, where the EPYC 9965 scores 1926069 against the Xeon 6781P's 584834, a 229.3% advantage. This is the single most lopsided result in the comparison and reflects the EPYC 9965's 192-core architecture handling integer operations far more efficiently than the Xeon's 80-core design.

Data encryption shows the second-largest gap. The EPYC 9965 scores 348449, while the Xeon 6781P scores 119623, a 191.3% difference. This workload benefits heavily from the EPYC 9965's core count and memory bandwidth. Random string sorting follows at 135.7% (633030 versus 268573), and data compression is 132.6% ahead (5679990 versus 2441690). Floating-point math shows a 127.3% advantage for the EPYC 9965 (1153453 versus 507406).

Extended instructions favor the EPYC 9965 by 92.6% (383298 versus 199048). The Cinebench multi-core tests show a consistent 36.3% advantage across all three versions: R15 (13775 versus 10105), R20 (57397 versus 42106), and R23 (136661 versus 100254). PassMark multithread shows a 36.1% advantage (160542 versus 117946), and physics shows a narrower 5.4% edge (18707 versus 17753).

The Xeon 6781P's only victory is in find prime numbers, where it scores 1687 against 1208, a 28.4% advantage. This result is notable because it is the sole test where the Xeon outperforms the EPYC 9965, and it highlights the Xeon's higher boost clock and per-core cache configuration. Single-thread performance is essentially tied, with the EPYC 9965 at 3176 and the Xeon 6781P at 3152, a 0.8% difference.

The Verdict

The data clearly indicates that the AMD EPYC 9965 is the superior processor for almost any multi-threaded server workload. Its 192 cores and 384 threads deliver massive advantages in integer math, encryption, compression, and floating-point operations, with margins ranging from 36.1% to 229.3% across the benchmark suite. For workloads that can utilize many cores, the EPYC 9965 is the obvious choice.

The Intel Xeon 6781P is the better option only for the specific prime numbers workload, where its 28.4% advantage demonstrates that its higher boost clock and larger per-core cache can matter in certain algorithmic patterns. However, this single win does not compensate for the Xeon's deficits in every other benchmark category.

For single-threaded applications, the two processors are effectively equivalent, with the EPYC 9965 leading by just 0.8%. This means that workloads that are not heavily threaded will not see a meaningful difference between the two. However, since the EPYC 9965 matches the Xeon in single-thread performance while vastly exceeding it in multi-threaded performance, the EPYC 9965 provides a more versatile overall package.

The Xeon 6781P's advantages are limited to its higher boost clock, slightly more PCIe lanes (136 versus 128), and larger per-core cache (2 MB L2 versus 1 MB L2). These factors contribute to its prime numbers win but do not translate into broader benchmark success. The EPYC 9965's advantages in core count, memory bandwidth, and L3 cache capacity are far more impactful across the recorded benchmark suite.

Organizations deploying for high-throughput computing, virtualization, database workloads, or content rendering should favor the EPYC 9965 based on the recorded data. The Xeon 6781P may be appropriate for specialized workloads that resemble the prime numbers test pattern, but the database shows no other scenario where it outperforms the EPYC 9965.