AMD EPYC 9655 vs Intel Xeon 6980P Comparison

Where Each One Wins

The benchmark record splits these two server processors into sharply different roles. The AMD EPYC 9655 wins 16 of the 17 head-to-head tests, and the margins are often enormous. The Intel Xeon 6980P takes exactly one test, PassMark extended instructions, where it leads by 5.4%. That single victory points to a narrow specialty, while the AMD part dominates everything else.

The EPYC 9655 is the clear choice for general compute density. In Cinebench R15, R20, and R23, both multi-core and single-core, the AMD processor wins by 110% or more in every case. That is not a marginal advantage; it is a doubling of performance in rendering workloads. The PassMark multithread test also shows a 110% lead, reinforcing that heavily threaded workloads are entirely AMD territory.

The gap extends to integer math, where the EPYC 9655 scores 1,139,161 versus 637,476 for the Xeon, a 78.7% advantage. Floating point math follows the same pattern: 662,958 versus 501,720, a 32.1% lead. Prime number finding is even more lopsided, with the AMD chip scoring 1,598 against 555, a 187.9% difference. Data compression favors AMD by 38.4%, and random string sorting by 82.6%.

The Xeon 6980P, despite having 128 cores and 256 threads, does not translate its core count into wins. Its only benchmark victory, extended instructions, is a specialized workload where the Intel architecture holds a modest edge. The data suggests the Xeon is not a general-purpose performance leader in this comparison; it is a niche player for specific instruction-heavy tasks, while the EPYC 9655 is the versatile workhorse.

Architecture Differences

The two processors come from different design philosophies. The AMD EPYC 9655 is built on the Zen 5 architecture, codenamed Turin, part of the EPYC 9005 series. It uses a 4 nm process from TSMC, with 99,780 million transistors spread across 12 chiplets, each 70.6 mm². The Intel Xeon 6980P uses the Granite Rapids architecture, part of the Xeon 6 family, on Intel's 5 nm process, with three dies of 598 mm² each. Intel does not report a transistor count in the database.

Core counts differ significantly. The EPYC 9655 has 96 cores and 192 threads, while the Xeon 6980P has 128 cores and 256 threads. The Intel part has more raw parallelism available, yet the benchmark results show it cannot convert that into higher scores. The AMD chip compensates with higher clock speeds: a base clock of 2.60 GHz and boost of 4.50 GHz, versus 2.00 GHz base and 3.90 GHz boost for the Xeon.

Cache hierarchies also differ. The EPYC 9655 has 80 KB of L1 per core, 1 MB of L2 per core, and 384 MB of shared L3 cache. The Xeon 6980P has 112 KB of L1 per core, 2 MB of L2 per core, and 504 MB of shared L3. Intel's larger caches do not overcome its clock and architecture disadvantages in the recorded tests.

Memory support is similar in width: both use DDR5 with a twelve-channel bus. The Xeon has higher peak memory bandwidth at 614.4 GB/s, compared to 576.0 GB/s for the AMD part. Both support ECC memory. PCIe connectivity differs: the EPYC 9655 offers Gen 5 with 128 lanes (CPU only), while the Xeon provides Gen 5 with 96 lanes (CPU only). The AMD part has more PCIe lanes, which matters for systems with many accelerators or storage devices.

Power envelopes are also distinct. The EPYC 9655 has a TDP of 400 W, while the Xeon 6980P is rated at 500 W. Intel's processor draws more power while delivering lower benchmark scores in nearly every test. The EPYC 9655 runs on AMD Socket SP5, while the Xeon uses Intel Socket 7529. Both are active production parts, with the Xeon released on 2024-09-23 and the EPYC on 2024-10-09. The EPYC 9655 has a launch MSRP of $11852, and the Xeon 6980P has a launch MSRP of $12460.

Head-to-Head Benchmarks

The Cinebench suite shows the most extreme differences. In Cinebench R15 multi-core, the EPYC 9655 scores 13,373 against 6,367, a 110% lead. Single-core R15 is equally decisive: 1,887 versus 898, also a 110.1% margin. R20 multi-core repeats the pattern at 55,722 versus 26,533, another 110% gap. R20 single-core is 7,866 versus 3,745, again 110%. R23 multi-core delivers 132,672 versus 63,175, and R23 single-core 18,730 versus 8,918, both at 110%. These are not incremental wins; they are consistent doublings across every Cinebench iteration.

PassMark multithread mirrors the Cinebench results. The EPYC 9655 scores 156,110, and the Xeon 6980P scores 74,324, a 110% difference. Single-thread performance is even more pronounced: 3,847 versus 1,681, a 128.9% lead for AMD. This indicates the EPYC 9655 has a substantial per-thread advantage, not just a core-count advantage.

Integer math shows a 78.7% lead for AMD: 1,139,161 versus 637,476. Floating point math is closer but still decisive: 662,958 versus 501,720, a 32.1% win. Data encryption favors AMD by 68.1%, with scores of 210,555 versus 125,246. Data compression goes to AMD by 38.4%, at 3,271,896 versus 2,364,519. Random string sorting is an 82.6% win for AMD, 439,682 versus 240,792.

Prime number finding is the largest single margin in the entire comparison. The EPYC 9655 scores 1,598, while the Xeon 6980P scores 555, a 187.9% advantage. Physics calculations are even more lopsided in percentage terms: 25,947 versus 3,350, a 674.5% lead for AMD. That physics result is the most extreme gap recorded, suggesting the Xeon struggles badly with that particular workload.

The only Intel victory is PassMark extended instructions: 214,794 versus 203,285, a 5.4% edge. This is a narrow win in a specific area, and it does not offset the scale of AMD's dominance elsewhere. The overall average benchmark score reflects this: the EPYC 9655 averages 373,479, while the Xeon 6980P averages 251,516.

The percentile rankings confirm the separation. The EPYC 9655 sits at the 100th percentile among all CPUs in the database, while the Xeon 6980P is at the 99th percentile. The EPYC's nearest rivals include the AMD EPYC 9535, which is 1.6% behind, and the Intel Xeon 6960P, which is 2.3% behind. The Xeon 6980P's nearest rivals include the AMD EPYC 9634, which is 3% ahead, and the AMD EPYC 9684X, which is 5.8% ahead. In other words, the Xeon 6980P is not only behind the EPYC 9655; it is also behind several other server chips in the database.

The Verdict

The data is unambiguous. The AMD EPYC 9655 outperforms the Intel Xeon 6980P in 16 of 17 recorded benchmarks, with margins ranging from 32.1% to 674.5%. The only Intel win is a 5.4% edge in extended instructions, a narrow specialty that does not compensate for the widespread losses.

For workloads that depend on multi-core rendering, the Cinebench results make the choice obvious. The EPYC 9655 doubles the Xeon 6980P's scores in R15, R20, and R23, across both single-core and multi-core tests. For general server tasks, including integer math, floating point math, encryption, compression, and sorting, the AMD processor leads by margins between 32.1% and 82.6%. For prime number finding and physics simulations, the AMD lead is extreme, at 187.9% and 674.5% respectively.

The Intel Xeon 6980P does offer more cores, more threads, larger caches, and higher memory bandwidth on paper. It also has a higher TDP. But the recorded benchmarks show that those specifications do not translate into performance. The EPYC 9655 achieves its results with 96 cores, 192 threads, and 384 MB of L3 cache, using a 4 nm process and higher clock speeds.

The Xeon 6980P is only the right choice for a system that specifically needs its extended instruction performance advantage, or for a workload that scales with its 504 MB of L3 cache and 614.4 GB/s memory bandwidth in ways not captured by these tests. The database record does not show such an advantage. For every measured workload, the EPYC 9655 is the superior processor.

The verdict: the AMD EPYC 9655 is the benchmark leader by a wide margin. The Intel Xeon 6980P is a niche alternative for extended instruction workloads, but the overall data strongly favors AMD for general and multi-threaded server computing.

FAQ

Q: How many benchmarks does the AMD EPYC 9655 win compared to the Intel Xeon 6980P?

A: The EPYC 9655 wins 16 of the 17 head-to-head benchmarks. The Xeon 6980P wins only one, PassMark extended instructions, by 5.4%.

Q: What is the largest performance margin between the two processors?

A: The largest margin is in PassMark physics, where the EPYC 9655 scores 25,947 versus 3,350 for the Xeon 6980P, a 674.5% advantage.

Q: Which processor has more cores and threads?

A: The Intel Xeon 6980P has 128 cores and 256 threads, while the AMD EPYC 9655 has 96 cores and 192 threads. Despite having fewer cores, the EPYC wins nearly all benchmarks.

Q: What are the clock speed differences?

A: The EPYC 9655 has a base clock of 2.60 GHz and a boost clock of 4.50 GHz. The Xeon 6980P has a base clock of 2.00 GHz and a boost clock of 3.90 GHz.

Q: How does memory bandwidth compare?

A: The Xeon 6980P has higher peak memory bandwidth at 614.4 GB/s, compared to 576.0 GB/s for the EPYC 9655. Both support DDR5 with a twelve-channel memory bus and ECC memory.

Q: What is the average benchmark score for each processor?

A: The EPYC 9655 has an average benchmark score of 373,479, while the Xeon 6980P has an average of 251,516. The EPYC also ranks at the 100th percentile among all CPUs, while the Xeon ranks at the 99th.