AMD EPYC 8024P

The AMD EPYC 8024P is a server/workstation processor in the EPYC 8004 series, built around the Zen 4c architecture with the codename Siena. It combines 8 cores and 16 threads with a 2.40 GHz base clock, a 3.00 GHz boost clock, and a 90 W TDP. Fabricated by TSMC on a 5 nm process, the chip contains 8,875 million transistors in a 73 mm² die. Its average benchmark score is 26,555, placing it in the 83rd percentile of all CPUs in the database, and its launch MSRP is $409.

Who Should Consider It

The EPYC 8024P is an 8-core, 16-thread server/workstation part, so it is best matched to workloads that can use all of those threads without demanding enormous core counts. The standout benchmark in its own data is PassMark data compression at 232,242, which suggests this processor is particularly useful for file servers, backup targets, and data pipeline jobs that spend time compressing or decompressing streams. Integer math also looks strong at 62,128, making general database processing and integer-heavy application logic a reasonable fit. Floating-point math is more moderate at 34,757, so scientific simulation and heavy floating-point analysis are less clearly the intended use case.

Data encryption scores 15,809 and extended instructions score 14,251, which are not exceptional relative to the rest of the chip’s results. That points to acceptable but not leading performance for encryption-heavy or advanced-instruction workloads. The PassMark random string sorting result of 34,613 gives another angle: sorting and reordering tasks should perform solidly. Meanwhile, the find prime numbers score of 109 and the physics score of 1,905 are low in the dataset, suggesting the part is not optimized for highly serial latency-sensitive code or physics-style calculations. In short, the EPYC 8024P is for users who need reliable 16-thread throughput in a server socket, especially for compression, integer math, and data-management workloads.

Power and Thermals

The EPYC 8024P sits in a 90 W TDP class. For a server/workstation processor, that is a modest power envelope, and the cooling implication is straightforward: a compact or dense server chassis with a modest heatsink and reasonable airflow should be sufficient. Liquid cooling is not suggested by the data, and the 90 W figure keeps this part out of the high-power server tier.

The thermal picture is informed by the manufacturing details. The chip uses TSMC’s 5 nm process and packs 8,875 million transistors into a 73 mm² die. That high density helps explain how 16 threads can be delivered inside a 90 W envelope. Memory support is six-channel DDR5 at 230.4 GB/s, which adds platform bandwidth without changing the processor’s TDP classification. There is no integrated graphics listed in the data, so any system using this processor must have its own display output or rely on a separate graphics solution for local console use.

Benchmark Performance

Cinebench R23 results show a multi-core score of 17,472 and a single-core score of 2,466. In Cinebench R20, the chip scores 7,338 multi-core and 1,035 single-core. In Cinebench R15, it scores 1,761 multi-core and 248 single-core. PassMark multi-thread performance is 20,556, and PassMark single-thread performance is 2,371. These numbers combine into an average benchmark score of 26,555, which places the EPYC 8024P in the 83rd percentile of all CPUs.

The nearest rival grouping is strikingly tight. The Intel Core i9-11900K posts an average score of 26,546 with a deltaPct of 0, meaning the EPYC 8024P is effectively tied with it in aggregate benchmark terms. The AMD Ryzen 7 5800X3D averages 26,574, which is 0.1% higher. The AMD Ryzen 5 7600 averages 26,617, 0.2% higher. The Intel Core i9-12900HK averages 26,672, 0.4% higher. All four rivals sit within 0.4% of the EPYC 8024P in average score. That is a notable result: a server-socket EPYC part holding its aggregate benchmark position against a set of non-EPYC CPUs. The workload mix will matter more than the aggregate average when choosing among them.

How It Compares

Intel Core i9-11900K — This is the nearest rival by average score, at 26,546. The deltaPct of 0 puts the two processors at the same aggregate benchmark level. The data shows no meaningful winner between them in average performance.

AMD Ryzen 7 5800X3D — The Ryzen 7 5800X3D averages 26,574, which is 0.1% above the EPYC 8024P. The gap is small enough that most real workloads would not reliably separate them based on aggregate score alone.

AMD Ryzen 5 7600 — The Ryzen 5 7600 averages 26,617, 0.2% higher than the EPYC 8024P. Again, this is within run-to-run variation for many benchmark suites, so the aggregate comparison does not strongly favor one over the other.

Intel Core i9-12900HK — The Core i9-12900HK has the highest average score in the nearest rival group at 26,672, a 0.4% edge over the EPYC 8024P. Even the largest gap in this cluster is small, reinforcing that the EPYC 8024P occupies the same aggregate performance neighborhood as these four CPUs.

FAQ

Q: What socket does the EPYC 8024P use?

A: It uses AMD Socket SP6, and the part number is 100-000001136.

Q: Does it support ECC memory?

A: Yes, ECC memory support is enabled. The memory interface is six-channel DDR5 with a bandwidth of 230.4 GB/s.

Q: How much cache does the processor have?

A: Each core has 64 KB of L1 cache and 1 MB of L2 cache. There is also 32 MB of shared L3 cache.

Q: Is the multiplier unlocked?

A: No, the multiplier unlocked field is false, so overclocking is not an enabled feature.

Q: When was it released, and is it still in production?

A: The release date is 2023-09-17, and the production status is Active.

Q: What PCIe capability does it provide?

A: It provides PCIe Gen 5 with 96 lanes from the CPU only.

Platform and Compatibility

The EPYC 8024P is built for AMD Socket SP6 and belongs to the EPYC 8004 series. The architecture is Zen 4c with the codename Siena, and the processor is manufactured by TSMC on a 5 nm process. Memory support is DDR5 over a six-channel bus, with ECC enabled and a memory bandwidth of 230.4 GB/s. PCIe Gen 5 is available with 96 lanes from the CPU only, which is a large amount of expansion capacity for storage controllers, network cards, and other server peripherals.

The integrated graphics field is not populated in the data, so this processor does not carry an integrated GPU. The production status is Active, meaning the part is still a current product. The upgrade path is defined by Socket SP6 and the EPYC 8004 series; the database only includes this specific part’s benchmarks, but any processor upgrade would need to be compatible with the same socket and platform generation.

Single-Thread vs Multi-Thread Behavior

The EPYC 8024P has 8 cores and 16 threads, and its benchmark split reflects that design. Cinebench R23 single-core is 2,466, while multi-core is 17,472. PassMark single-thread is 2,371, while multi-thread is 20,556. The multi-core results are substantially higher than the single-thread results, which is exactly what should be expected from a 16-thread server/workstation part.

This split means the processor favors parallel workloads. Integer math at 62,128, data compression at 232,242, and random string sorting at 34,613 all benefit from multi-threaded execution. The single-thread side is more modest: the find prime numbers result of 109 is low, and the encryption score of 15,809 and extended instructions score of 14,251 are not standout numbers. Software that can spread work across 16 threads will get far more from the EPYC 8024P than software limited to one or two threads. The multi-thread scores are where this processor makes its case.