AMD EPYC 9655P vs Intel Xeon 6980P Comparison

Head-to-Head Benchmarks

The benchmark data paints an unusually one-sided picture. Across all seventeen recorded head-to-head tests, the AMD EPYC 9655P takes the win, and in most cases by a margin that is not close. The Intel Xeon 6980P does not record a single victory in the database, which makes this comparison more about the scale of AMD's lead than about any competitive back-and-forth.

The most dramatic gap appears in PassMark physics. The EPYC 9655P scores 25,847 against the Xeon's 3,350, a delta of 671.6%. That is not a typical performance gap between two server-class parts; it suggests a fundamental difference in how the two architectures handle the workload. Prime number finding shows a similar story, with AMD ahead by 203.8%, scoring 1,686 versus 555.

Across the Cinebench suite, the results are remarkably consistent. In Cinebench R15 multi-core, AMD scores 13,744 against Intel's 6,367, a 115.9% advantage. Single-core R15 shows 1,940 versus 898, a 116% delta. R20 multi-core lands at 57,268 versus 26,533 (115.8%), and R20 single-core at 8,085 versus 3,745 (115.9%). R23 repeats the pattern: 136,354 versus 63,175 in multi-core (115.8%), and 19,250 versus 8,918 in single-core (115.9%). The consistency of these deltas across every Cinebench iteration is striking; it indicates a stable architectural advantage rather than workload-specific luck.

PassMark multi-thread mirrors Cinebench almost exactly. The EPYC 9655P scores 160,490, the Xeon 6980P scores 74,324, a 115.9% lead. Single-thread PassMark shows 3,849 versus 1,681, a 129% gap. Integer math favors AMD by 92.2% (1,225,251 versus 637,476), while floating-point math shows a 42.7% lead (715,866 versus 501,720). Data compression is 47.4% faster on AMD (3,486,158 versus 2,364,519), and data encryption is 75.7% ahead (220,074 versus 125,246). Random string sorting shows an 87.6% delta (451,824 versus 240,792).

The closest contest is PassMark extended instructions, where AMD leads by just 7.4% (230,609 versus 214,794). That is the only benchmark in the entire comparison where the two processors are within shouting distance. It suggests that for certain specialized instruction-heavy workloads, the Intel part is far more competitive, even if it still loses.

FAQ

Q: Which processor has the higher average benchmark score?

A: The AMD EPYC 9655P has an average benchmark score of 397,773, while the Intel Xeon 6980P averages 251,516. AMD also sits at the 100th percentile among all CPUs, versus the 99th percentile for Intel.

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

A: The EPYC 9655P leads in Cinebench R23 single-core with 19,250 versus 8,918, a 115.9% advantage. PassMark single-thread shows 3,849 versus 1,681, a 129% lead.

Q: What is the closest benchmark between the two?

A: PassMark extended instructions is the narrowest gap. The EPYC 9655P scores 230,609 and the Xeon 6980P scores 214,794, a 7.4% delta.

Q: Does the Intel Xeon 6980P win any benchmark?

A: No. Out of seventeen head-to-head tests recorded in the database, the EPYC 9655P wins all seventeen. The Xeon 6980P records zero wins.

Q: How do the two processors compare in memory bandwidth?

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

Q: Which processor has more cores?

A: The Intel Xeon 6980P has 128 cores and 256 threads. The AMD EPYC 9655P has 96 cores and 192 threads.

Architecture Differences

The two processors come from fundamentally different design philosophies. The AMD EPYC 9655P is built on Zen 5 architecture, codenamed Turin, and belongs to the EPYC 9005 series. It uses a 4 nm process manufactured by TSMC. The chip is composed of 12 dies, each measuring 70.6 mm², for a combined die area that is relatively small. The transistor count is listed at 99,780 million.

The Intel Xeon 6980P uses Granite Rapids architecture, part of the Xeon 6 family (Granite Rapids-AP). It is built on Intel's 5 nm process. The physical design is very different: three dies, each measuring 598 mm², for a much larger total die footprint. The database does not list a transistor count for the Intel part.

Cache hierarchies also diverge significantly. AMD provides 80 KB of L1 cache per core, 1 MB of L2 per core, and 384 MB of shared L3 cache. Intel provides 112 KB of L1 per core, 2 MB of L2 per core, and a larger 504 MB of shared L3. The larger per-core caches on the Intel side do not translate into benchmark wins, but they do represent a different approach to feeding the cores.

The memory controllers are similar in channel count, both twelve-channel DDR5, but the rated bandwidth differs. Intel lists 614.4 GB/s, AMD lists 576.0 GB/s. PCIe support also differs: AMD offers Gen 5 with 128 lanes (CPU only), while Intel offers Gen 5 with 96 lanes (CPU only). Both support ECC memory and neither includes integrated graphics.

Clock speeds tell another part of the story. The AMD EPYC 9655P has a base clock of 2.60 GHz and a boost clock of 4.50 GHz. The Intel Xeon 6980P has a base clock of 2.00 GHz and a boost clock of 3.90 GHz. AMD's clocks are higher across the board, which helps explain the single-core results despite the core count disadvantage.

Specification Differences

The AMD EPYC 9655P and Intel Xeon 6980P differ on nearly every major specification field. Core count favors Intel: 128 versus 96. Thread count follows, 256 versus 192. The TDP also favors Intel in the sense of being higher, 500 W versus 400 W, though that is a power cost rather than a performance benefit.

Clock speeds favor AMD. Base clock is 2.60 GHz versus 2.00 GHz, and boost clock is 4.50 GHz versus 3.90 GHz. The process node favors AMD as well, 4 nm from TSMC versus 5 nm from Intel. The socket is completely different: AMD uses Socket SP5, Intel uses Socket 7529, so there is no platform compatibility between the two.

Memory bandwidth favors Intel at 614.4 GB/s versus 576.0 GB/s. PCIe lanes favor AMD at 128 versus 96. Cache favors Intel in total L3 capacity, 504 MB versus 384 MB, and in per-core L1 and L2 sizes. The release dates are close, with Intel launching on 2024-09-23 and AMD on 2024-10-09. The launch MSRP for the AMD EPYC 9655P is $10811, and for the Intel Xeon 6980P it is $12460. Both are active production parts with locked multipliers, and neither has integrated graphics.

Where Each One Wins

The AMD EPYC 9655P wins everywhere in the recorded benchmarks, so the use-case split is less about which processor wins a given task and more about which tasks are closer than others. If the workload is heavy on extended instructions, the AMD lead shrinks to 7.4%, making the Intel part a more reasonable choice for specialized instruction-heavy code where the platform and core count might matter more.

For floating-point math, AMD leads by 42.7%, which is substantial but smaller than the margins seen in integer math (92.2%) or data compression (47.4%). Workloads that are integer-bound, encryption-heavy, or compression-heavy will see the largest practical gains from the EPYC 9655P. The data encryption test shows a 75.7% lead, and random string sorting shows 87.6%, both strong indicators for database and storage-related tasks.

The physics test is the outlier, with a 671.6% delta. Any simulation or physics-heavy workload would see an enormous advantage on the AMD side. Prime number finding, at 203.8%, reinforces that pattern for mathematical and number-theoretic workloads.

The Intel Xeon 6980P does offer more cores (128 versus 96) and more L3 cache (504 MB versus 384 MB), plus higher memory bandwidth. For workloads that scale with raw core count and cache capacity, and that do not rely on the specific instruction paths where AMD dominates, the Intel part could still be the right choice. But the benchmark data does not show any test where that theoretical advantage becomes a measured win.

The Verdict

The data is unambiguous. The AMD EPYC 9655P wins all seventeen head-to-head benchmarks, with deltas ranging from 7.4% to 671.6%. The average benchmark score of 397,773 versus 251,516 puts AMD roughly 58% higher overall, and the 100th percentile ranking versus 99th confirms that this is not a near-miss comparison.

For buyers optimizing for raw performance in the measured workloads, the EPYC 9655P is the clear pick. It delivers higher clocks, a smaller process node, and better results in every recorded test, while drawing 100 W less power (400 W versus 500 W). The Intel Xeon 6980P has more cores, more cache, and higher memory bandwidth, but none of those advantages show up as a win in the database.

The narrow extended instructions margin is the only hint that the Xeon 6980P has any competitive territory. For users running workloads that are dominated by extended instruction sets, the Intel part deserves a closer look, especially given its larger core count and cache. But for general server and workstation use, the recorded data points squarely at the AMD EPYC 9655P. The launch MSRP is $10811 for AMD and $12460 for Intel, though the performance gap is far larger than the price gap.