AMD EPYC 8534P vs AMD Ryzen 9 PRO 9965 Comparison

Head-to-Head Benchmarks

The head-to-head results show a decisive overall victory for the AMD EPYC 8534P, which wins 8 of the 11 recorded benchmarks. The largest margin comes in data encryption, where the EPYC 8534P scores 121,728 against the Ryzen 9 PRO 9965's 44,920, a 171% advantage. This is followed closely by integer math, with the EPYC posting 514,526 versus 243,280, a 111.5% lead. Data compression also favors the EPYC heavily: 1,791,742 compared to 908,293, a 97.3% gap. These three workloads show the EPYC's dominance in throughput-oriented tasks that scale with core count and memory bandwidth.

The Ryzen 9 PRO 9965 does take the single-threaded tests decisively. In passmark_single_thread, the Ryzen scores 4,682 against the EPYC's 2,441, a 47.9% advantage in the other direction. The Ryzen also wins in finding prime numbers, scoring 364 versus 278, a 23.6% edge. These results reflect the Ryzen's substantially higher clock speeds and newer core design. However, these two wins are isolated to workloads that do not scale with core count.

The remaining benchmarks are all EPYC wins but with varying margins. Floating point math shows an 80.1% lead for the EPYC (289,443 versus 160,746). Extended instructions are 58.5% ahead (112,860 versus 71,210), and random string sorting is 36.4% ahead (129,479 versus 94,915). The multithread score is closer, with the EPYC ahead by 7.9% (71,900 versus 66,655), and physics runs 12.6% higher for the EPYC (3,667 versus 3,256). Notably, the EPYC wins even in the aggregate passmark_multithread, which suggests that its 64 cores can keep it ahead despite the Ryzen's higher single-core efficiency.

Where Each One Wins

The EPYC 8534P is the clear winner for any workload that can leverage its 64 cores and 128 threads. The data shows it excels in data compression, encryption, and integer math, all of which are parallelizable. The EPYC also holds a significant edge in floating-point math, which is critical for scientific computing and rendering. Its 128 MB L3 cache, combined with six-channel memory support, provides 230.4 GB/s of bandwidth, which directly feeds these compute-heavy operations. The EPYC's 80.1% advantage in floating point math, for instance, makes it the obvious choice for numerical simulation or financial modeling.

The Ryzen 9 PRO 9965 wins in single-threaded performance and prime number finding. With a boost clock of 5.50 GHz and a 4 nm process, it offers a 47.9% faster single-thread score. This makes it preferable for legacy applications, certain database queries, or any code that cannot utilize more than a few cores. The Ryzen also has a 38.6% advantage in prime number generation (364 versus 278), a task that is often latency-bound. For an interactive workstation where the user may run a mix of single-threaded and lightly threaded tasks, the Ryzen's responsiveness would be better, though its 89.6 GB/s memory bandwidth is a limitation for memory-intensive work.

For mixed workloads, the EPYC still dominates. The passmark_physics score of 3,667 versus 3,256 shows a 12.6% EPYC lead, and even the multithread score, which is close at 7.9%, still favors the EPYC. The Ryzen's only other win, in the duplicate single_thread test, does not change the overall picture. The EPYC's 8 wins out of 11 benchmarks are not just in raw quantity; they are in categories that matter for server and workstation processing.

Architecture Differences

The two processors come from different architectural generations and serve different segments. The EPYC 8534P is built on the Zen 4c architecture (Siena) on a 5 nm process, with a transistor count of 35,500 million and a die size of 4x 73 mm². It has 64 cores and 128 threads, with a base clock of 2.30 GHz and a boost clock of 3.10 GHz. The Ryzen 9 PRO 9965 uses the newer Zen 5 architecture (Granite Ridge) on a 4 nm process, with 16,630 million transistors across 2x 70.6 mm² dies. It offers 16 cores and 32 threads, running at a much higher base clock of 4.30 GHz and boost of 5.50 GHz. The EPYC's 200 W TDP is 30 W higher than the Ryzen's 170 W, but it delivers four times the cores.

Cache hierarchies also differ. The EPYC provides 64 KB of L1 per core and 1 MB of L2 per core, with a shared 128 MB L3 cache. The Ryzen, with its newer design, has 80 KB L1 per core and 1 MB L2 per core, but only 64 MB of L3. This means the EPYC has double the L3 cache, which is critical for large data sets. Memory support is another major separator: the EPYC uses six-channel DDR5, yielding 230.4 GB/s, while the Ryzen uses dual-channel DDR5, yielding only 89.6 GB/s. This bandwidth difference explains the EPYC's huge wins in compression and integer math.

PCIe connectivity also shows the EPYC's server orientation. The EPYC 8534P provides 96 PCIe Gen 5 lanes (CPU only), while the Ryzen 9 PRO 9965 provides 24 lanes. This allows the EPYC to support more GPUs and NVMe drives, which is essential for a workstation or server. The Ryzen does have integrated Radeon Graphics, while the EPYC has none, giving the Ryzen the edge for a desktop-like system without a discrete GPU. The Ryzen also has a 4 nm process, which is newer, but the EPYC's 5 nm process is still efficient for its core count.

The Verdict

The benchmark data clearly dictates the choice. The AMD EPYC 8534P is the superior processor for any workload where the parallel processing is possible. It wins 8 of 11 benchmarks, and the margins are often huge: 171% in encryption, 111.5% in integer math, and 97.3% in data compression. Its 64 cores, 128 threads, and 128 MB L3 cache, combined with six-channel memory at 230.4 GB/s, make it a far more capable compute platform. The 96 PCIe Gen5 lanes also make it a better host for high-bandwidth expansion. The EPYC is the right pick for a server or workstation running virtualization, large data analytics, or computational science.

The AMD Ryzen 9 PRO 9965 is the better choice for a workstation where single-threaded performance is the primary need. Its 47.9% lead in single-thread, 38.6% lead in prime numbers, and much higher 4.30 GHz base clock make it snappier for interactive use. It also has a lower TDP (170 W versus 200 W) and includes integrated graphics, allowing for a simpler system build. However, the Ryzen's 16 cores are dwarfed by the EPYC's 64, and its dual-channel memory bandwidth of 89.6 GB/s is a bottleneck. The data shows the Ryzen is a fine workstation, but the EPYC is the superior part for the server segment.

The Ryzen's percentileVsAllCpus is 98th, and the EPYC is also 98th, so both are high-end. The EPYC's average benchmark score is 185,092, versus 145,728 for the Ryzen. This 27% gap in the average score is a more holistic measure. The EPYC is also within 1.3% of the AMD Ryzen Threadripper PRO 9975WX and 2.9% above the AMD EPYC 7763, whereas the Ryzen 9 PRO 9965 sits within 1% of several rivals, including the Xeon w9-3575X. The EPYC is 4.3% below the Intel Xeon 678X, but still ahead of the Ryzen's competition. The data does not support the Ryzen for heavy multi-threaded duties.

FAQ

Q: Which processor wins the most benchmark tests?

A: The AMD EPYC 8534P wins 8 out of 11 benchmarks, while the AMD Ryzen 9 PRO 9965 wins 3.

Q: How large is the single-threaded performance difference?

A: The Ryzen 9 PRO 9965 scores 4,682 in passmark_single_thread, which is 47.9% higher than the EPYC 8534P's 2,441.

Q: What is the difference in memory bandwidth?

A: The EPYC 8534P has six-channel memory with a bandwidth of 230.4 GB/s, while the Ryzen 9 PRO 9965 has dual-channel memory with 89.6 GB/s.

Q: Which CPU has more PCIe lanes?

A: The EPYC 8534P offers 96 PCIe Gen 5 lanes (CPU only), whereas the Ryzen 9 PRO 9965 has 24 lanes.

Q: Does the EPYC 8534P have integrated graphics?

A: No, the EPYC 8534P has no integrated graphics, but the Ryzen 9 PRO 9965 includes Radeon Graphics.

Q: What is the average benchmark score for each?

A: The EPYC 8534P has an average benchmark score of 185,092, and the Ryzen 9 PRO 9965 has an average of 145,728.