The AMD Epyc 9555 is built on a 4nm process node and carries a Thermal Design Power rating of 360W, reflecting the sustained performance demands of its enterprise-class design. It fully supports 64-bit computing and integrates PCIe 5.0 connectivity for high-bandwidth peripheral and storage communication. The processor does not include integrated graphics, which is typical for a dedicated server CPU of this class.
The Epyc 9555 runs 64 cores at a base clock of 3.2 GHz, supporting 128 threads in total, with a turbo clock speed that reaches up to 4.4 GHz under boosted conditions. Its cache hierarchy consists of 5120 KB of L1, 64 MB of L2 at 1 MB per core, and a substantial 256 MB of L3 cache allocated at 4 MB per core, providing ample fast-access memory across the core count. The processor uses a clock multiplier of 32, though the multiplier is locked and cannot be adjusted.
The Epyc 9555 supports DDR5 memory across 12 memory channels, with a maximum RAM speed of 6000 MHz and a peak memory bandwidth of 576 GB/s. It can address up to 9000 GB of total system memory, making it well-suited for workloads that require large in-memory datasets. ECC memory support is included, providing hardware-level error detection and correction to maintain data integrity in server environments.
The Epyc 9555 supports multithreading, allowing each physical core to handle two threads simultaneously for improved throughput across parallel workloads. It includes the NX bit for hardware-enforced memory protection, helping to guard against certain classes of malicious code execution. On the instruction set side, the processor supports a broad range of extensions including AVX2, FMA3, and AES, alongside MMX, F16C, AVX, SSE 4.1, and SSE 4.2, covering vectorized math, floating-point operations, and hardware-accelerated encryption.
In PassMark testing, the Epyc 9555 achieves a multi-threaded score of 133,253, reflecting its capacity to distribute workloads across its full core and thread count. Its single-threaded PassMark result stands at 3,759, indicating the per-core performance available for tasks that do not scale across multiple threads.