AMD EPYC 9455P vs AMD EPYC 9555P Comparison

Head-to-Head Benchmarks

The benchmark data shows a clear overall winner in the AMD EPYC 9555P, which takes 13 of the 17 recorded tests. The margin is not uniform, though. In the Cinebench suite, the 9555P leads by exactly 16.1% across all six tests, both multicore and singlecore. That consistency suggests the clock-for-clock performance difference is baked into the silicon, not workload dependent. For example, Cinebench R23 multicore shows 115186 for the 9555P against 99206 for the 9455P, while R23 singlecore shows 16261 versus 14005. The same 16.1% gap repeats in R15 and R20, which is a strong sign that the advantage comes from core count and thread scaling rather than any boost behavior.

The PassMark suite tells a more varied story. The biggest win for the 9555P is in extended instructions, where it scores 191082 versus 137485, a 39% lead. Data compression follows at 36.8% (2639400 versus 1928897), and floating point math is close behind at 36% (486407 versus 357783). Integer math gives the 9555P a 29.8% edge (787106 versus 606239), and data encryption shows a 29% gap (148896 versus 115403). These are heavily parallel workloads, so the extra 16 cores on the 9555P translate directly into large percentage advantages.

Random string sorting goes to the 9555P by 15.5% (280398 versus 242701), and the multithread PassMark test shows only a 5.7% gap (123576 versus 116927), which is surprisingly small given the core difference. The 9455P does win four tests, though. The most notable is PassMark single thread, where the 9455P scores 3745 versus the 9555P’s 3410, an 8.9% advantage. Physics also goes to the 9455P by 10.6% (17315 versus 15474), and find prime numbers gives the 9455P a narrow 3.6% win (1107 versus 1067). These wins are concentrated in latency-sensitive or lightly threaded tasks, which suggests the 9455P’s higher per-core efficiency in specific workloads.

Looking at the rivals listed in the database, the 9555P sits at the 99th percentile of all CPUs, with an average benchmark score of 287066. Its nearest rival is the Intel Xeon 696X at 286102, a mere 0.3% gap, followed by the AMD EPYC 9565 at 285471 (0.6% behind) and the Intel Xeon 6780E at 280438 (2.4% behind). The 9455P also sits at the 99th percentile with an average score of 217854. Its closest competitor is the Intel Xeon w9-3595X at 209881, which the 9455P beats by 3.8%. The Intel Xeon 6747P is actually ahead of the 9455P by 8.6%, while the Intel Xeon 6741P trails by 11.8% and the AMD EPYC 9335 trails by 12.2%.

FAQ

Q: Which CPU wins more benchmarks overall?

A: The AMD EPYC 9555P wins 13 of the 17 head-to-head tests. The AMD EPYC 9455P wins 4 tests.

Q: Is the 9555P always faster in multicore workloads?

A: Yes, in all three Cinebench multicore tests (R15, R20, R23) the 9555P leads by exactly 16.1%. In PassMark multithread, the lead shrinks to 5.7%, but it still wins.

Q: Where does the 9455P outperform the 9555P?

A: The 9455P wins in PassMark single thread (3745 versus 3410, an 8.9% lead), PassMark physics (17315 versus 15474, a 10.6% lead), and PassMark find prime numbers (1107 versus 1067, a 3.6% lead).

Q: How do these CPUs compare to their closest rivals?

A: The 9555P has an average score of 287066, just 0.3% ahead of the Intel Xeon 696X. The 9455P averages 217854, which is 3.8% ahead of the Intel Xeon w9-3595X but 8.6% behind the Intel Xeon 6747P.

Q: Do both CPUs share the same architecture?

A: Yes, both are Zen 5 architecture from the EPYC 9005 series, codenamed Turin, built on TSMC’s 4 nm process with the same 66,520 million transistors and 8x 70.6 mm² die size.

Q: What is the core and thread difference?

A: The 9555P has 64 cores and 128 threads. The 9455P has 48 cores and 96 threads. Both have the same 4.40 GHz boost clock, but the 9555P has a 3.20 GHz base clock versus 3.15 GHz for the 9455P.

Architecture Differences

Both processors come from the same EPYC 9005 series, use the Zen 5 architecture, and carry the Turin codename. They are built on the same TSMC 4 nm process node, with identical transistor counts of 66,520 million and the same die layout of 8x 70.6 mm². The cache hierarchy is also identical: 80 KB of L1 per core, 1 MB of L2 per core, and 256 MB of shared L3. Neither chip has 3D V-Cache, so the L3 capacity is uniform across both.

The architectural similarity means the performance differences come almost entirely from core count and clock behavior rather than from any microarchitectural divergence. Both support DDR5 memory over a twelve-channel bus with 576.0 GB/s bandwidth, both have ECC memory support, and both use AMD Socket SP5. The PCIe implementation is the same: Gen 5 with 128 lanes (CPU only). Neither has integrated graphics, and neither has an unlocked multiplier.

The base clock differs slightly: the 9555P runs at 3.20 GHz while the 9455P runs at 3.15 GHz. The boost clock is identical at 4.40 GHz. This small base clock difference, combined with the 16 extra cores, explains why the 9555P consistently outperforms in heavily threaded benchmarks. However, the fact that the 9455P wins in single-threaded PassMark and physics tests suggests that the 9455P can sustain higher effective clocks in certain lightly threaded scenarios, or that the scheduler handles its 48 cores differently under those specific loads.

The TDP also differs: the 9555P is rated at 360 watts, while the 9455P comes in at 300 watts. This is a notable design distinction, as it means the 9455P delivers most of the performance of the larger chip while drawing less power, at least according to the rated specifications. The release dates are identical (2024-10-09), and both are listed as Active production status. The part numbers differ (100-000001523 for the 9555P, 100-000001563 for the 9455P), but the market segment is the same: Server/Workstation.

Specification Differences

The two CPUs differ in several key fields. Core count: 64 for the 9555P versus 48 for the 9455P. Thread count: 128 versus 96. Base clock: 3.20 GHz versus 3.15 GHz. TDP: 360 watts versus 300 watts. Launch MSRP: the 9555P was $7983, and the 9455P was $4819. Part numbers differ as noted above.

Everything else is identical: boost clock at 4.40 GHz, same socket (SP5), same architecture (Zen 5), same codename (Turin), same generation (EPYC Zen 5 Turin), same process node (4 nm), same foundry (TSMC), same transistor count (66,520 million), same die size (8x 70.6 mm²), same cache per core (80 KB L1, 1 MB L2), same shared L3 (256 MB), same memory support (DDR5), same memory bus (twelve-channel), same memory bandwidth (576.0 GB/s), same ECC support, same PCIe (Gen 5, 128 lanes), no integrated graphics on either, and both have locked multipliers.

The average benchmark score difference is substantial: 287066 for the 9555P versus 217854 for the 9455P, a raw gap of 69212 points. Both sit at the 99th percentile of all CPUs, but the 9555P does so with a much higher absolute score.

Where Each One Wins

The AMD EPYC 9555P is the clear choice for any workload that scales with core count and thread count. The data shows large wins in data compression (36.8% ahead), floating point math (36% ahead), extended instructions (39% ahead), and integer math (29.8% ahead). These are the kinds of tasks found in scientific computing, financial modeling, video encoding, and large-scale database operations. The consistent 16.1% lead across all Cinebench tests also indicates strong rendering and 3D workload performance. For virtualization, where you want as many threads as possible, the 64-core/128-thread configuration gives a clear advantage.

The AMD EPYC 9455P wins in specific niches. PassMark single thread shows it at 3745 versus 3410, an 8.9% advantage, which points to better performance in lightly threaded applications like legacy software, certain database queries that run on a single thread, or development workflows that compile sequentially. The physics test win (17315 versus 15474, a 10.6% lead) suggests an edge in simulation workloads that rely on physics calculations with lower thread counts. The find prime numbers win (1107 versus 1067, a 3.6% lead) is a narrow but real edge in certain algorithmic workloads.

The 9455P also has a lower TDP at 300 watts versus 360 watts, which matters in dense server environments where power and cooling are constrained. For a node that runs many lightly threaded VMs or containers, the 9455P’s wins in single-thread and physics tests could translate to better responsiveness per core, even though the 9555P has more total throughput.

The Verdict

Choose the AMD EPYC 9555P if your primary concern is raw multithreaded performance. The data is unambiguous: it wins 13 out of 17 tests, and its wins are often by 30% to 39% in parallel workloads. It also posts an average benchmark score of 287066, putting it at the 99th percentile and just 0.3% ahead of the Intel Xeon 696X, which is its closest rival. For database servers, high-performance computing, rendering farms, or any environment where you can fully utilize 128 threads, the 9555P is the stronger part.

Choose the AMD EPYC 9455P if your workloads are more mixed or if you need better single-threaded performance in specific tasks. The 9455P wins in PassMark single thread by 8.9%, in physics by 10.6%, and in find prime numbers by 3.6%. It also carries a lower TDP of 300 watts, which can be a decisive factor in power-constrained deployments. Its average benchmark score of 217854 still places it at the 99th percentile, and it beats the Intel Xeon w9-3595X by 3.8%, though it trails the Intel Xeon 6747P by 8.6%. If your application mix includes enough lightly threaded tasks to benefit from the 9455P’s wins, the 48-core chip offers a more balanced profile. For pure throughput, the 9555P is the answer.