AMD EPYC 9535 vs AMD EPYC 9754 Comparison

The AMD EPYC 9535 and AMD EPYC 9754 are both flagship-class server processors built for the same SP5 socket, but they represent two very different design philosophies. The 9535 is a 64-core Zen 5 part with blistering single-thread speed, while the 9754 is a 128-core Zen 4c part built for massive parallel throughput. Benchmark results show a clear split: the 9754 wins most multi-threaded integer and compression workloads, while the 9535 dominates single-thread and prime-number tasks. This is not a simple "which is faster" question—it is a question of workload shape and core-count scaling.

Head-to-Head Benchmarks

The most dramatic victory for the AMD EPYC 9754 comes in passmark_data_compression, where it scores 3,558,043 against the 9535's 2,308,822—a 35.1% lead. This is a workload that scales almost perfectly with core count, and the 9754's 128 cores give it a massive structural advantage. Similarly, in passmark_data_encryption, the 9754 posts 231,891 versus 127,372, a 45.1% margin. Encryption is another parallel-friendly task, and the 9754's double core count shines here.

The 9754 also wins passmark_integer_math with 1,026,896 versus 730,281 (28.9% ahead) and passmark_floating_point_math with 588,187 versus 457,047 (22.3% ahead). These are broad synthetic workloads that favor raw thread count. In passmark_physics, the 9754 scores 8,793 versus 3,834—a massive 56.4% advantage, which is the largest percentage win for either chip. The 9754 also takes passmark_extended_instructions (224,322 vs 175,784, 21.6% ahead) and passmark_random_string_sorting (306,481 vs 247,506, 19.2% ahead).

However, the AMD EPYC 9535 fights back in ways that reveal its architectural superiority per core. The single-thread score is a blowout: 3,720 versus 2,328, a 59.8% advantage for the 9535. This is the largest percentage win for either processor, and it reflects the Zen 5 core's substantial IPC gain over Zen 4c. The 9535 also wins passmark_find_prime_numbers decisively, scoring 874 versus 604—a 44.7% lead. This workload is notoriously latency-sensitive and does not scale well with core count, so the 9535's superior per-core performance dominates.

Interestingly, the 9535 wins passmark_multithread with 114,528 versus 98,752, a 16% margin, despite having half the cores. This suggests that the 9535's high clock speeds (4.30 GHz boost versus 3.10 GHz) and stronger per-core performance can overcome core-count deficits in certain mixed workloads. The head-to-head tally is 7 wins for the 9754 and 4 for the 9535, but the wins are not equal in weight—the 9535's single-thread victory is nearly 60% ahead, while the 9754's largest win is 56.4% in physics.

FAQ

Q: Which processor has the higher average benchmark score?

A: The AMD EPYC 9535 has an average benchmark score of 379,408, while the AMD EPYC 9754 sits at 364,371. The 9535 is 4.1% ahead of the 9754 in this metric, and the nearest rival data shows the 9535 is also 1.6% ahead of the EPYC 9655 and 3.9% ahead of the Intel Xeon 6960P.

Q: How does the single-thread performance compare?

A: The 9535 is dramatically faster in single-thread workloads, scoring 3,720 versus 2,328 in passmark_single_thread—a 59.8% advantage. This is the largest percentage gap in any head-to-head benchmark between the two.

Q: Does the 9754's higher core count always translate to more performance?

A: No. While the 9754 wins 7 of 11 head-to-head benchmarks, it loses passmark_multithread (98,752 vs 114,528, a 16% deficit) and passmark_find_prime_numbers (604 vs 874, a 44.7% deficit) despite having twice the cores.

Q: What is the memory bandwidth difference?

A: The 9535 has a memory bandwidth of 576.0 GB/s, while the 9754 is rated at 460.8 GB/s. Both use twelve-channel DDR5 memory, but the 9535's higher bandwidth reflects its newer memory controller design.

Q: Which processor has a higher boost clock?

A: The 9535 boosts to 4.30 GHz, while the 9754 tops out at 3.10 GHz. The 9535 also has a higher base clock at 2.40 GHz versus 2.25 GHz.

Q: Are both processors in the same performance percentile?

A: Yes, both the 9535 and the 9754 are in the 100th percentile versus all CPUs, meaning both rank at the very top of the database. The difference is in workload-specific strengths, not overall tier.

Architecture Differences

The fundamental split is between Zen 5 and Zen 4c. The 9535 uses the Zen 5 architecture with the codename Turin, while the 9754 uses Zen 4c with the codename Bergamo. This is not a minor revision—Zen 5 is a full next-generation core design, whereas Zen 4c is a density-optimized variant of the previous generation. The 9535 is built on a 4 nm process from TSMC, while the 9754 uses a 5 nm process. The 9535 packs 66,520 million transistors across 8x 70.6 mm² dies, while the 9754 has 71,000 million transistors across 8x 73 mm² dies—despite being on an older node, the 9754 has more transistors because it has double the cores.

Cache configuration is another major divergence. The 9535 has 80 KB of L1 cache per core, while the 9754 has 64 KB per core. Both have 1 MB of L2 per core and 256 MB of shared L3 cache, but the per-core L1 difference reflects Zen 5's larger and more capable front-end. The 9535's smaller core count means it has more cache per core effectively, while the 9754's larger core count spreads the same L3 across more threads.

The 9754 is a denser design, which is why it can fit 128 cores in the same socket. Zen 4c sacrifices some per-core performance and clock headroom for density, which is why the 9754's boost clock is 3.10 GHz versus 4.30 GHz for the 9535. The 9535's higher clocks and newer core design are the main reasons it wins single-thread and other latency-bound tasks. The 9754's process node is 5 nm, one generation behind the 9535's 4 nm, but it still manages a higher transistor count due to the sheer core count.

Specification Differences

The most obvious spec gap is core count: the 9535 has 64 cores and 128 threads, while the 9754 has 128 cores and 256 threads. TDP also differs, with the 9535 rated at 300 W and the 9754 at 360 W. Boost clock is a major differentiator—4.30 GHz for the 9535 versus 3.10 GHz for the 9754—and base clock follows the same pattern at 2.40 GHz versus 2.25 GHz. Memory bandwidth is another clear split: 576.0 GB/s for the 9535 versus 460.8 GB/s for the 9754.

Both processors use the same AMD Socket SP5 and support twelve-channel DDR5 memory with ECC. Both also have PCIe Gen 5 with 128 lanes (CPU only). The 9535 has a launch MSRP of $8992, while the 9754 has a launch MSRP of $11900. The 9535 was released on 2024-10-09, while the 9754 came earlier on 2023-06-12. The 9535 is part of the EPYC 9005 series, while the 9754 belongs to the EPYC 9004 series. Neither processor has an unlocked multiplier, and neither has integrated graphics. The 9535's part number is 100-000001147, and the 9754's is 100-000001234.

The Verdict

The data points to a clear workload split. The AMD EPYC 9754 is the choice for workloads that scale with raw thread count and can tolerate lower per-core performance—data compression, encryption, integer math, and physics simulations all see massive wins for the 9754. Its 128 cores deliver 35-56% advantages in these areas, which is exactly what you want for high-density virtualization, big-data processing, or large-scale scientific computing.

The AMD EPYC 9535 is the choice for workloads where per-core performance matters more than core count. Its single-thread lead of 59.8% is enormous, and it also wins prime-number finding by 44.7%. The 9535 even wins passmark_multithread by 16%, which suggests that its high clocks and newer core design can compensate for half the core count in mixed workloads. For databases, high-frequency trading, or any application with latency-sensitive single-thread paths, the 9535 is the better fit.

The 9535 also has a lower TDP (300 W vs 360 W) and higher memory bandwidth (576.0 GB/s vs 460.8 GB/s), which are practical advantages for dense server deployments. However, the 9754's launch MSRP is higher at $11900 versus $8992 for the 9535. If your workloads are massively parallel and your software scales beyond 64 cores, the 9754's 128 cores provide undeniable throughput. If your workloads have serial bottlenecks or you need maximum per-core speed, the 9535 is the superior processor.

Where Each One Wins

AMD EPYC 9754 wins in data-heavy, parallel workloads: data compression (35.1% ahead), data encryption (45.1% ahead), integer math (28.9% ahead), floating-point math (22.3% ahead), physics (56.4% ahead), extended instructions (21.6% ahead), and random string sorting (19.2% ahead). Its 128 cores are the decisive factor in all these tests, and the 256 threads provide ample parallelism for throughput-oriented applications. The 9754 is also the only choice if you need maximum core density in a single socket, as no other processor in the database offers 128 cores.

AMD EPYC 9535 wins in per-core-sensitive workloads: single-thread performance (59.8% ahead), prime-number finding (44.7% ahead), and the mixed passmark_multithread test (16% ahead). Its 4.30 GHz boost clock and Zen 5 architecture give it a decisive edge in latency-bound tasks. The 9535 also has a higher average benchmark score (379,408 vs 364,371), which suggests that on average, across a broad mix of tests, it edges out the 9754. For single-socket servers running a mix of workloads, the 9535's higher average score and lower TDP make it a more balanced choice, while the 9754 is a specialized tool for extreme parallel throughput.