AMD EPYC 9965 vs Intel Xeon 6774P Comparison

FAQ

Q: Which processor has more cores and threads?

A: The AMD EPYC 9965 has 192 cores and 384 threads, while the Intel Xeon 6774P has 64 cores and 128 threads. The AMD part offers three times the core count and three times the thread count of the Intel part.

Q: How do the two processors compare in single-threaded performance?

A: In the PassMark single-thread test, the AMD EPYC 9965 scores 3176 versus 3047 for the Intel Xeon 6774P, a 4.2% advantage for AMD. The Cinebench R23 single-core result is not directly listed in the head-to-head table, but the AMD part records 19293 in that test from its own benchmark set.

Q: Which processor wins in multi-threaded workloads?

A: The AMD EPYC 9965 wins all recorded multi-threaded benchmarks. In Cinebench R23 multi-core, it scores 136661 versus 95836 for the Intel Xeon 6774P, a 42.6% lead. The PassMark multithread score is 160542 for AMD versus 112749 for Intel, a 42.4% advantage.

Q: What is the difference in memory bandwidth?

A: The AMD EPYC 9965 supports twelve-channel memory with a bandwidth of 576.0 GB/s, while the Intel Xeon 6774P supports eight-channel memory with a bandwidth of 409.6 GB/s. Both support DDR5 and ECC memory.

Q: Are there any benchmarks where the Intel Xeon 6774P wins?

A: Yes, in the PassMark find prime numbers test, the Intel Xeon 6774P scores 1264 against 1208 for the AMD EPYC 9965, a 4.4% margin. This is the only head-to-head benchmark win for Intel out of 14 recorded comparisons.

Q: How do the two processors rank among all CPUs in the database?

A: The AMD EPYC 9965 sits at the 100th percentile among all CPUs, while the Intel Xeon 6774P sits at the 99th percentile. The AMD part's average benchmark score is 620487, compared to 300372 for the Intel part.

Architecture Differences

The AMD EPYC 9965 belongs to the EPYC 9005 series and uses the Zen 5 architecture under the codename Turin. It is manufactured on a 3 nm process by TSMC. The Intel Xeon 6774P uses the Granite Rapids architecture, also codenamed Granite Rapids, and is manufactured on a 5 nm process by Intel itself. The process node difference is significant: 3 nm versus 5 nm gives the AMD part a density and efficiency advantage on paper, though the actual performance relationship is established by the benchmark data.

The core configurations diverge sharply. The AMD EPYC 9965 packs 192 cores and 384 threads, while the Intel Xeon 6774P offers 64 cores and 128 threads. This 3:1 ratio in both cores and threads dictates the multi-threaded benchmark outcomes. The cache hierarchy also differs: AMD provides 80 KB of L1 and 1 MB of L2 per core, with 384 MB of shared L3. Intel provides 112 KB of L1 and 2 MB of L2 per core, with 336 MB of shared L3. The larger per-core L1 and L2 caches on the Intel part do not compensate for the overall core deficit in heavily parallel workloads, as the results show.

Memory architecture reinforces the AMD advantage. The EPYC 9965 runs a twelve-channel DDR5 memory bus with 576.0 GB/s of bandwidth, while the Xeon 6774P runs an eight-channel DDR5 bus with 409.6 GB/s. The difference in aggregate bandwidth is substantial and appears in memory-sensitive workloads. Both parts support ECC memory and target the server/workstation segment. PCIe connectivity is close: the AMD part provides Gen 5 with 128 lanes from the CPU, while the Intel part provides Gen 5 with 136 lanes from the CPU, a slight edge for Intel in raw lane count.

The sockets differ completely, which matters for platform planning. AMD uses Socket SP5, while Intel uses Socket 4710. The reported die size for Intel is 2x 598 mm², while no die size is recorded for AMD. The AMD part has a higher TDP at 500 watts versus 350 watts for Intel, and the AMD launch MSRP is $14813 compared to $6760 for Intel. The release dates are also distinct: the EPYC 9965 launched on 2024-10-09, while the Xeon 6774P launched on 2025-05-21.

Head-to-Head Benchmarks

The data records 14 head-to-head benchmark comparisons, and the AMD EPYC 9965 wins 13 of them. The single Intel win is narrow and isolated. The overall pattern is one of overwhelming AMD dominance in throughput, with a small but consistent single-thread edge.

Starting with the Cinebench suite, the AMD part leads by the same margin across all three versions. In Cinebench R15 multi-core, AMD scores 13775 against 9660 for Intel, a 42.6% delta. In R20 multi-core, AMD scores 57397 against 40251, again 42.6%. In R23 multi-core, AMD scores 136661 against 95836, still 42.6%. The consistency of this delta across three Cinebench versions indicates a stable architectural advantage in multi-threaded rendering workloads, likely driven by the 3:1 core ratio.

The PassMark suite shows even larger gaps in several tests. Integer math is the biggest win for AMD: 1926069 versus 593440, a 224.6% delta. Data encryption follows at 348449 versus 115025, a 202.9% delta. Floating-point math shows 1153453 versus 465314, a 147.9% delta. Data compression is 5679990 versus 2309868, a 145.9% delta. Random string sorting is 633030 versus 262417, a 141.2% delta. Extended instructions are 383298 versus 177273, a 116.2% delta. These are massive margins, all above 100%, indicating that the AMD part has more than double the throughput in several important server workloads.

The PassMark multithread score is 160542 for AMD versus 112749 for Intel, a 42.4% delta. This is closer to the Cinebench margins than to the integer or encryption margins. The PassMark physics test shows a smaller win for AMD: 18707 versus 16023, a 16.8% delta. Single-threaded performance is also in AMD's favor, but narrowly: 3176 versus 3047, a 4.2% delta in the PassMark single-thread test.

The only Intel victory is in the PassMark find prime numbers test, where Intel scores 1264 against 1208 for AMD, a 4.4% margin. This is a small, specialized workload that favors Intel's per-core characteristics. It does not change the overall picture, but it does show that Intel retains a niche advantage in at least one arithmetic pattern.

Specification Differences

The two processors differ across nearly every major specification category. The AMD EPYC 9965 has 192 cores and 384 threads, while the Intel Xeon 6774P has 64 cores and 128 threads. Base clocks are 2.25 GHz for AMD and 2.50 GHz for Intel, with boost clocks of 3.70 GHz and 3.90 GHz respectively. The Intel part has a higher clock speed on both ends, but the core count gap overwhelms this in multi-threaded results.

TDP differs by 150 watts: the AMD part is rated at 500 watts, the Intel part at 350 watts. The process node is 3 nm for AMD (TSMC) and 5 nm for Intel (Intel foundry). The AMD part uses Socket SP5, while the Intel part uses Socket 4710. Memory channels are twelve for AMD and eight for Intel, with corresponding bandwidth ratings of 576.0 GB/s and 409.6 GB/s. PCIe lane counts are 128 for AMD and 136 for Intel, both Gen 5. Cache configurations differ as described above, with AMD offering more L3 (384 MB versus 336 MB) and Intel offering more L1 and L2 per core. The launch MSRP is $14813 for AMD and $6760 for Intel. Release dates are 2024-10-09 for AMD and 2025-05-21 for Intel.

Where Each One Wins

The AMD EPYC 9965 wins in every major throughput category recorded. It leads in Cinebench R15, R20, and R23 multi-core by 42.6% each. It leads in PassMark multithread by 42.4%. It leads in integer math by 224.6%, in data encryption by 202.9%, in floating-point math by 147.9%, in data compression by 145.9%, in random string sorting by 141.2%, and in extended instructions by 116.2%. It also wins the PassMark physics test by 16.8% and the single-thread test by 4.2%. For workloads that scale with core count, thread count, memory bandwidth, or cache capacity, the AMD part is the clear choice based on the recorded data.

The Intel Xeon 6774P wins only one recorded test: PassMark find prime numbers, with a 4.4% margin. This suggests that Intel retains a slight edge in certain integer-heavy, low-cache-footprint arithmetic loops that do not benefit from additional cores. The Intel part also has a higher boost clock (3.90 GHz versus 3.70 GHz) and a higher base clock (2.50 GHz versus 2.25 GHz), which may explain its single win in that specific test despite the overall single-thread deficit in the PassMark single-thread benchmark.

For memory bandwidth, the AMD part offers 576.0 GB/s versus 409.6 GB/s for Intel, so workloads that are bandwidth-limited will favor AMD. For PCIe lane count, the Intel part offers 136 lanes versus 128 for AMD, a minor advantage for systems that need maximum expansion or I/O connectivity. The Intel part also has a lower TDP at 350 watts versus 500 watts, which may matter for power-constrained deployments, though the benchmark data does not include power efficiency measurements.

The Verdict

The benchmark data is unambiguous: the AMD EPYC 9965 is the superior processor for nearly all recorded workloads. It wins 13 of 14 head-to-head comparisons, with margins ranging from 4.2% in single-threaded performance to 224.6% in integer math. The 3:1 core and thread advantage, combined with twelve-channel memory at 576.0 GB/s, produces dominant results in every multi-threaded test. The 100th percentile ranking among all CPUs, against the Intel part's 99th percentile, confirms the overall standing.

The Intel Xeon 6774P is the better choice only in the narrow case of the find prime numbers workload, where it wins by 4.4%. It also offers more PCIe lanes (136 versus 128) and a lower TDP (350 watts versus 500 watts), which may be relevant for specific platform requirements. Its launch MSRP of $6760 is lower than the AMD part's $14813, but pricing considerations are not part of this analysis.

For high-throughput server and workstation deployments where multi-threaded performance, memory bandwidth, and cache capacity drive results, the AMD EPYC 9965 is the data-supported pick. For workloads that match the Intel part's single niche win, or for systems that prioritize the Intel platform's specific I/O and power profile, the Xeon 6774P remains a viable option, but the recorded benchmarks show it trailing the AMD part by wide margins in the vast majority of tests.