AMD EPYC 9555P vs AMD EPYC 9965 Comparison

Head-to-Head Benchmarks

The benchmark data shows a decisive overall victory for the AMD EPYC 9965, which wins 15 of the 17 recorded head-to-head comparisons. The 9965’s strongest advantage appears in PassMark integer math, where it scores 1,926,069 against the 9555P’s 787,106, a 144.7% lead. Floating point math tells a similar story: 1,153,453 versus 486,407, a 137.1% edge. These are not small margins; they represent workloads where the 9965 effectively doubles or more the output of the 9555P.

Data compression and encryption workloads also heavily favor the 9965. In PassMark data compression, the 9965 scores 5,679,990 against 2,639,400, a 115.2% advantage. Data encryption shows a 134% delta, with scores of 348,449 versus 148,896. Random string sorting, another memory and cache sensitive test, goes to the 9965 by 125.8%, scoring 633,030 versus 280,398. Extended instructions, which often reflect AVX-style throughput, show a 100.6% delta (383,298 versus 191,082). These results indicate that the 9965 is not merely faster in raw core count; it scales almost linearly in workloads that depend on parallel execution and cache bandwidth.

The Cinebench suite paints a more moderated picture. In Cinebench R23 multicore, the 9965 scores 136,661 against 115,186 for the 9555P, an 18.6% delta. R20 multicore shows 57,397 versus 48,378, also 18.6%. R15 multicore yields 13,775 versus 11,610, an 18.6% delta. Interestingly, single-core Cinebench results are also close but favor the 9965: R23 single-core is 19,293 versus 16,261 (18.6%), R20 is 8,103 versus 6,829 (18.7%), and R15 is 1,944 versus 1,638 (18.7%). This is notable because the 9555P has a higher boost clock (4.40 GHz versus 3.70 GHz), yet the 9965 still wins single-threaded Cinebench tests by nearly 19%. The database suggests the 9965’s Zen 5c architecture at 3 nm delivers better instructions per clock in these specific workloads.

The 9555P’s only victories come in PassMark single-thread tests. It scores 3,410 versus the 9965’s 3,176, a 6.9% advantage. This appears twice in the data (as passmark_single_thread and passmark_singlethread), but both record the same scores. The 9555P also wins the PassMark physics test by a smaller margin, but that is not listed as a win in the head-to-head table; the wins table shows 15 for the 9965 and 2 for the 9555P. The two wins correspond to the two single-thread entries. So in purely single-threaded, clock-sensitive scenarios, the 9555P’s higher boost clock pays off, but everywhere else the 9965 dominates.

PassMark multithread shows a 29.9% delta (160,542 versus 123,576), and find prime numbers shows a modest 13.2% delta (1,208 versus 1,067). The overall average benchmark score for the 9965 is 620,487, versus 287,066 for the 9555P. The 9965 sits at the 100th percentile among all CPUs in the database, while the 9555P is at the 99th percentile. Among its nearest rivals, the 9965 is 18.5% ahead of the AMD EPYC 9845 and 22.7% ahead of the AMD EPYC 9755. The 9555P, by contrast, is only 0.3% ahead of the Intel Xeon 696X and 0.6% ahead of the AMD EPYC 9565, making it part of a tightly packed cluster at the top.

Architecture Differences

The two processors share the same Zen 5 architecture family and both use the AMD Socket SP5, but they diverge significantly in implementation. The 9965 is built on a 3 nm process node at TSMC, while the 9555P uses a 4 nm node. The 9965 is codenamed Turin and belongs to the Zen 5c variant, which prioritizes core density. The 9555P is also codenamed Turin but uses the standard Zen 5 core design. This explains the core count difference: the 9965 packs 192 cores and 384 threads, while the 9555P offers 64 cores and 128 threads.

Cache configurations differ substantially. Both have 80 KB of L1 per core and 1 MB of L2 per core, but the shared L3 cache is 384 MB on the 9965 versus 256 MB on the 9555P. That extra 128 MB of L3 is critical for workloads that repeatedly access large datasets, which helps explain the massive deltas in data compression and random string sorting. The 9555P has a listed transistor count of 66,520 million and a die size of 8x 70.6 mm², while the 9965 does not have those figures recorded in the database. The 9965’s base clock is 2.25 GHz and boost clock is 3.70 GHz, whereas the 9555P runs at 3.20 GHz base and 4.40 GHz boost. The lower clocks on the 9965 are the trade-off for fitting 192 cores within the 500 W TDP envelope, compared to 360 W for the 9555P.

Both processors support DDR5 memory with a twelve-channel bus and 576.0 GB/s of memory bandwidth, and both support ECC memory. PCIe connectivity is identical: Gen 5 with 128 lanes (CPU only). Neither has integrated graphics. Both are listed as active production parts, released on the same date, and neither has an unlocked multiplier. The launch MSRP for the 9965 is $14813, and for the 9555P it is $7983. The generation field notes the 9965 as EPYC (Zen 5c (Turin)) and the 9555P as EPYC (Zen 5 (Turin)), highlighting that the core design philosophy differs even though both are Turin-based.

The 9965’s 3 nm process allows for a higher density of cores on the same socket, which is why it can offer three times the core count of the 9555P. However, the 9555P’s larger process node and lower core count permit higher clock speeds. The database shows that in single-threaded PassMark tests, the 9555P’s 4.40 GHz boost wins out, but in Cinebench single-core tests, the 9965’s Zen 5c cores still outperform despite the clock deficit. This suggests that Zen 5c on 3 nm has a meaningful IPC advantage in certain workloads, or that the Cinebench single-core test is more sensitive to memory latency and cache size than raw clock speed.

The Verdict

The data points to a clear split. If the workload is heavily multithreaded, the AMD EPYC 9965 is the superior part. It wins every multicore benchmark in the database, often by margins exceeding 100% in PassMark integer, floating point, compression, encryption, and extended instructions. For database servers, scientific computing, virtualization hosts running many VMs, or any workload that can use 192 cores, the 9965 is the obvious choice. Its 100th percentile ranking and average benchmark score of 620,487 versus 287,066 for the 9555P reflect a processor that sits in a league of its own. The nearest rival to the 9965, the AMD EPYC 9845, is still 18.5% behind on average score.

The AMD EPYC 9555P should be selected when single-threaded performance matters more, or when the software licensing model penalizes high core counts. It wins the PassMark single-thread test by 6.9% and offers a higher boost clock. The 9555P is also part of a much tighter competitive field: its nearest rival, the Intel Xeon 696X, is only 0.3% behind, and the AMD EPYC 9565 is 0.6% behind. This means the 9555P does not have the same overwhelming lead over its peers as the 9965 does. But for workloads that are lightly threaded or that require the highest possible clock speed on a single core, the 9555P is the better fit. It also has a lower TDP at 360 W versus 500 W, which can matter in dense server deployments with cooling constraints.

Neither processor is a good choice for a desktop or small workstation; both are clearly server parts with twelve-channel memory and 128 PCIe Gen 5 lanes. The 9965’s 384 MB of L3 cache is a massive advantage for in-memory analytics or large caching layers. The 9555P’s 256 MB is still substantial but is 33% smaller. The decision ultimately comes down to core count versus clock speed. The 9965 wins 15 of 17 benchmarks; the 9555P wins the other two. If the budget allows for the 9965’s higher launch MSRP, the data says it is the faster processor in almost every scenario. If cost or single-thread performance is the primary constraint, the 9555P is the sensible alternative.

Specification Differences

The two processors differ in several key specification fields. Core count is 192 versus 64, and thread count is 384 versus 128. Base clock is 2.25 GHz for the 9965 versus 3.20 GHz for the 9555P. Boost clock is 3.70 GHz versus 4.40 GHz. TDP is 500 W versus 360 W. Process node is 3 nm versus 4 nm. The 9965 has 384 MB of shared L3 cache, while the 9555P has 256 MB. The 9555P lists a transistor count of 66,520 million and a die size of 8x 70.6 mm²; the 9965 does not have those fields recorded. The generation field differs: Zen 5c (Turin) versus Zen 5 (Turin). Launch MSRP is $14813 versus $7983. All other specifications, including socket, memory support, memory bus, memory bandwidth, ECC support, PCIe lanes, integrated graphics, market segment, production status, release date, and unlocked multiplier status, are identical between the two.

FAQ

Q: Which processor has more cores and threads?

A: The AMD EPYC 9965 has 192 cores and 384 threads, while the AMD EPYC 9555P has 64 cores and 128 threads.

Q: Does the 9555P ever beat the 9965 in any benchmark?

A: Yes, the 9555P wins the PassMark single-thread test with a score of 3,410 versus 3,176 for the 9965, a 6.9% advantage. This appears twice in the data as two separate entries.

Q: How much larger is the L3 cache on the 9965?

A: The 9965 has 384 MB of shared L3 cache, while the 9555P has 256 MB. The per-core L1 and L2 caches are identical at 80 KB and 1 MB respectively.

Q: What is the difference in average benchmark scores?

A: The 9965 has an average benchmark score of 620,487, while the 9555P has an average score of 287,066. The 9965 is at the 100th percentile of all CPUs, and the 9555P is at the 99th percentile.

Q: Do both processors support the same memory and PCIe?

A: Yes, both support DDR5 with a twelve-channel bus, 576.0 GB/s memory bandwidth, ECC memory, and PCIe Gen 5 with 128 lanes (CPU only). They also use the same AMD Socket SP5.

Q: Which processor has a higher boost clock?

A: The 9555P has a boost clock of 4.40 GHz, while the 9965 has a boost clock of 3.70 GHz. The 9555P also has a higher base clock at 3.20 GHz versus 2.25 GHz.

Where Each One Wins

The AMD EPYC 9965 wins in every multicore workload recorded in the database. Its largest margins come in PassMark integer math (144.7% ahead), floating point math (137.1%), data encryption (134%), random string sorting (125.8%), and data compression (115.2%). These are workloads that scale with core count and cache capacity, and the 9965’s 192 cores and 384 MB of L3 cache provide an overwhelming advantage. It also wins all six Cinebench tests, both multicore and single-core, by roughly 18.6% each. For any workload that can use more than 64 cores, the 9965 is the clear winner. The data shows it is also faster in Cinebench single-core tests despite a lower boost clock, meaning its per-core efficiency on 3 nm Zen 5c is superior in those specific benchmarks.

The AMD EPYC 9555P wins only in the PassMark single-thread test. Its 4.40 GHz boost clock gives it a 6.9% edge over the 9965 in that specific metric. This makes it the better choice for workloads that are strictly single-threaded and clock-bound, such as certain legacy applications, license servers, or lightly threaded database queries. The 9555P also has a lower TDP of 360 W versus 500 W, which can be an advantage in power-constrained environments, though the database does not include power efficiency benchmarks. In the context of its nearest rivals, the 9555P is only 0.3% ahead of the Intel Xeon 696X and 0.6% ahead of the AMD EPYC 9565, so it does not have the same dominant position as the 9965, which leads its closest rival by 18.5%. For a use case that prioritizes raw multithreaded throughput, the 9965 is the only logical choice. For a use case that requires the fastest single thread and lower power draw, the 9555P is the pick.