AMD EPYC 7443P vs AMD EPYC 9135 Comparison

The AMD EPYC 9135 and AMD EPYC 7443P are both server processors from AMD, but they represent different generations and design philosophies. The 9135 is built on the newer Zen 5 architecture, while the 7443P uses the older Zen 3 design. Benchmark data shows a close overall contest, with the 9135 winning 11 of 17 head-to-head tests, yet the 7443P secures decisive victories in several specialized workloads.

Head-to-Head Benchmarks

The most striking pattern in the head-to-head data is the split between single-threaded and multi-threaded performance. The EPYC 9135 dominates in single-core tests with a 26.3% lead in PassMark single-thread scoring, recording 3672 against the 7443P’s 2907. This advantage carries through Cinebench R23 single-core, where the 9135 scores 6936 versus 6837, a 1.4% edge. The same 1.5% margin appears across Cinebench R15 single-core (699 vs 689) and R20 single-core (2913 vs 2871). These consistent single-thread wins indicate a substantial per-core performance uplift from the newer architecture.

In multi-core Cinebench tests, the 9135 maintains a narrower but uniform lead. Cinebench R23 multi-core shows 49136 for the 9135 against 48433 for the 7443P, a 1.5% difference. R20 multi-core shows 20637 versus 20341, and R15 multi-core shows 4952 versus 4881, both also at 1.5%. PassMark multithread follows the same trend, with the 9135 scoring 57808 to the 7443P’s 56981, another 1.5% victory. This is notable because the 7443P has 24 cores and 48 threads, while the 9135 has only 16 cores and 32 threads; the 9135 still manages to outpace its rival in these heavily threaded workloads.

The 9135’s largest win outside of single-core comes in PassMark physics, where it scores 6096 against 4748, a 28.4% margin. It also leads in PassMark extended instructions, scoring 54795 versus 48213, a 13.7% advantage. These results suggest that the Zen 5 architecture brings meaningful gains in instruction-heavy and physics simulation tasks.

However, the 7443P fights back with substantial wins in several PassMark sub-tests. The most dramatic is data encryption, where the 7443P scores 57263 against the 9135’s 40941, a 28.5% lead. Data compression also favors the 7443P, with 820859 versus 737167, a 10.2% edge. Integer math shows the 7443P ahead at 232632 versus 204258, a 12.2% margin, and floating-point math gives it a smaller 3.7% win (129932 vs 125125). Prime number finding favors the 7443P by 27.1% (410 vs 299), and random string sorting shows a 3.5% lead (95581 vs 92226). These six wins for the 7443P highlight areas where its higher core count and larger cache overcome the architectural deficit.

Architecture Differences

The two processors are built on fundamentally different designs. The EPYC 9135 uses the Zen 5 architecture, codenamed Turin, fabricated on a 4 nm process by TSMC. The EPYC 7443P relies on the Zen 3 architecture, codenamed Milan, on a 7 nm process, also from TSMC. This process node difference contributes to the 9135’s higher clock speeds: it runs at 3.65 GHz base and 4.30 GHz boost, compared to the 7443P’s 2.85 GHz base and 4.00 GHz boost.

Core and thread counts differ significantly. The 9135 has 16 cores and 32 threads, while the 7443P has 24 cores and 48 threads. Despite having 50% more cores, the 7443P cannot translate that into an overall benchmark win, as the 9135’s superior per-core performance compensates.

Cache layouts also diverge. The 9135 provides 80 KB of L1 cache per core, 1 MB of L2 per core, and 64 MB of shared L3 cache. The 7443P offers 64 KB of L1 per core, 512 KB of L2 per core, but doubles the L3 to 128 MB shared. This larger L3 cache on the 7443P likely explains its strong showing in data compression and encryption workloads, where large working sets benefit from on-die storage.

Memory support marks a generational split. The 9135 uses DDR5 memory with a twelve-channel memory bus, delivering a memory bandwidth of 576.0 GB/s. The 7443P uses DDR4 with an eight-channel bus, providing 204.8 GB/s of bandwidth. Both support ECC memory. The 9135’s nearly threefold bandwidth advantage is substantial for memory-bound server tasks.

PCIe connectivity also differs. The 9135 offers PCIe Gen 5 with 128 lanes (CPU only), while the 7443P provides PCIe Gen 4 with the same 128 lanes. The newer PCIe standard on the 9135 doubles the per-lane bandwidth, beneficial for high-speed storage and accelerators.

Sockets and generations differ as well. The 9135 uses AMD Socket SP5 and belongs to the EPYC (Zen 5 (Turin)) generation, released on 2024-10-09. The 7443P uses AMD Socket SP3 and is part of the EPYC (Zen 3 (Milan)) generation, released on 2021-03-14. Both have a TDP of 200 watts and are currently in active production.

FAQ

Q: Which processor has a higher average benchmark score?

A: The AMD EPYC 9135 has an average benchmark score of 83020, while the AMD EPYC 7443P scores 81661. This places the 9135 1.7% ahead of the 7443P in the nearestRivals data.

Q: How does the core count compare between the two?

A: The EPYC 7443P has 24 cores and 48 threads, whereas the EPYC 9135 has 16 cores and 32 threads. The 7443P offers 8 more cores and 16 more threads.

Q: What is the single-thread performance difference?

A: The EPYC 9135 leads by 26.3% in PassMark single-thread testing, scoring 3672 against the 7443P’s 2907. In Cinebench R23 single-core, the lead narrows to 1.4% (6936 vs 6837).

Q: Which processor is better for data encryption tasks?

A: The EPYC 7443P is significantly better, scoring 57263 in PassMark data encryption versus 40941 for the 9135, a 28.5% advantage. Its larger L3 cache and higher core count likely contribute to this result.

Q: Do both processors support ECC memory?

A: Yes, both the EPYC 9135 and the EPYC 7443P have ECC memory support. However, the 9135 uses DDR5 with a twelve-channel bus, while the 7443P uses DDR4 with an eight-channel bus.

Q: What is the memory bandwidth difference?

A: The EPYC 9135 provides 576.0 GB/s of memory bandwidth, while the EPYC 7443P provides 204.8 GB/s. This is a substantial difference favoring the 9135.

The Verdict

The benchmark data presents a clear picture for different usage scenarios. The EPYC 9135 is the better choice for workloads that depend on single-thread performance, high memory bandwidth, or modern instruction sets. Its 26.3% lead in PassMark single-thread and 13.7% edge in extended instructions make it well-suited for databases, virtualization hosts, and applications where per-core licensing costs matter. The 28.4% win in PassMark physics also indicates strength in simulation or scientific computing tasks that rely on floating-point throughput per core.

The EPYC 7443P, despite being older, remains competitive in specific multi-threaded, data-heavy workloads. Its 28.5% lead in data encryption and 10.2% advantage in data compression point to excellence in cryptographic operations, backup/archival systems, and compression pipelines. The 12.2% win in integer math and 27.1% lead in prime number finding suggest it handles integer-heavy batch processing well. The 7443P’s 128 MB L3 cache, double that of the 9135, appears to be the deciding factor in these cache-sensitive tasks.

Overall, the 9135 wins the majority of head-to-head tests at 11 versus 6, and it holds the higher average benchmark score. That said, the 7443P’s wins are often larger in magnitude, with three victories exceeding 10% (data encryption at 28.5%, prime numbers at 27.1%, integer math at 12.2%), while the 9135’s wins are mostly small (1.5% in most Cinebench tests) except for single-thread (26.3%) and physics (28.4%). For general-purpose server duties, the 9135 provides a more balanced and modern foundation, but for specialized data transformation workloads, the 7443P remains a formidable option.

Specification Differences

The following specifications differ between the two processors:

• Cores: 16 (9135) vs 24 (7443P)
• Threads: 32 (9135) vs 48 (7443P)
• Base Clock: 3.65 GHz (9135) vs 2.85 GHz (7443P)
• Boost Clock: 4.30 GHz (9135) vs 4.00 GHz (7443P)
• Socket: AMD Socket SP5 (9135) vs AMD Socket SP3 (7443P)
• Architecture: Zen 5 (9135) vs Zen 3 (7443P)
• Codename: Turin (9135) vs Milan (7443P)
• Process Node: 4 nm (9135) vs 7 nm (7443P)
• Die Size: 2x 70.6 mm² (9135) vs 4x 81 mm² (7443P)
• L1 Cache: 80 KB per core (9135) vs 64 KB per core (7443P)
• L2 Cache: 1 MB per core (9135) vs 512 KB per core (7443P)
• L3 Cache: 64 MB shared (9135) vs 128 MB shared (7443P)
• Memory Support: DDR5 (9135) vs DDR4 (7443P)
• Memory Bus: Twelve-channel (9135) vs Eight-channel (7443P)
• Memory Bandwidth: 576.0 GB/s (9135) vs 204.8 GB/s (7443P)
• PCIe: Gen 5, 128 Lanes (9135) vs Gen 4, 128 Lanes (7443P)
• Release Date: 2024-10-09 (9135) vs 2021-03-14 (7443P)
• Launch MSRP: $1214 (9135) vs $1337 (7443P)
• Part Number: 100-000001150 (9135) vs 100-000000342100-100000342WOF (7443P)

Where Each One Wins

The EPYC 9135 wins in all Cinebench tests across R15, R20, and R23 versions, both single-core and multi-core. It also takes PassMark multithread, physics, extended instructions, and single-thread tests. This set of victories makes it the preferred processor for general compute, rendering, and any workload that benefits from a modern, high-clock design with fast memory access. The 9135’s 576.0 GB/s memory bandwidth is particularly valuable for large in-memory databases or high-performance computing clusters that move substantial data volumes.

The EPYC 7443P wins in PassMark data compression, data encryption, find prime numbers, floating-point math, integer math, and random string sorting. These are workloads where the 7443P’s 24 cores and 48 threads provide parallel throughput, and its 128 MB L3 cache reduces memory latency for repeated data access. Encryption and compression servers, scientific computing with integer-heavy algorithms, and text processing pipelines would all benefit from the 7443P’s strengths. For users running these specific applications, the 7443P offers clear performance advantages despite being built on an older architecture.