AMD EPYC 9335 vs AMD Ryzen 9 PRO 9965 Comparison

Head-to-Head Benchmarks

The benchmark data presents a fascinating split between these two AMD processors. The EPYC 9335 wins 6 of the 11 recorded tests, while the Ryzen 9 PRO 9965 takes 5, yet the nature of those wins tells a clear story about each chip's design intent.

The EPYC 9335 dominates in raw throughput tasks. Its data compression score of 1,203,096 versus 908,293 for the Ryzen represents a 32.5% advantage, a massive gap for workloads that shuffle large datasets. Data encryption shows an even larger 40.6% lead (63,159 vs 44,920), reflecting the EPYC's server-grade security processing capabilities. Extended instructions, which often indicate cryptography and specialized math operations, see the EPYC ahead by 48.4% (105,706 vs 71,210).

Floating-point math and integer math follow the same pattern. The EPYC 9335 scores 228,123 in floating-point versus 160,746 for the Ryzen, a 41.9% difference, while integer math shows 346,291 versus 243,280, a 42.3% gap. Random string sorting, another throughput-heavy test, favors the EPYC by 22.9% (116,608 vs 94,915). Across these six wins, the EPYC's average advantage is substantial, driven by its doubled core count and twelve-channel memory system.

The Ryzen 9 PRO 9965 counters in single-thread performance and latency-sensitive tests. Its single-thread score of 4,682 crushes the EPYC's 2,732, a 41.6% lead that reflects its higher 5.50 GHz boost clock versus 4.40 GHz. Physics simulation shows the most dramatic reversal: the Ryzen scores 3,256 versus 1,905, a 41.5% advantage, suggesting superior per-core efficiency for sequential calculations. Prime number finding, another per-core test, favors the Ryzen by 6.6% (364 vs 340).

Interestingly, the multithread benchmark is nearly a dead heat. The Ryzen 9 PRO 9965 edges out the EPYC 9335 by just 1.3% (66,655 vs 65,811), despite having half the cores and threads. This suggests that while the EPYC has more parallel resources, the Ryzen's higher clock speeds and newer architecture allow it to nearly match the EPYC in this specific test. The multithread result is anomalous compared to the other throughput tests, where the EPYC's core advantage shines through.

FAQ

Q: Which processor has the higher average benchmark score?

A: The AMD EPYC 9335 records an average benchmark score of 194,228, while the AMD Ryzen 9 PRO 9965 averages 145,728. The EPYC sits in the 99th percentile of all CPUs, while the Ryzen ranks in the 98th percentile.

Q: How significant is the single-thread performance gap?

A: The Ryzen 9 PRO 9965 leads by 41.6% in single-thread scoring (4,682 vs 2,732). This is one of the largest deltas in the entire head-to-head comparison, reflecting the Ryzen's 5.50 GHz boost clock versus the EPYC's 4.40 GHz.

Q: Does the EPYC 9335 win every multithreaded workload?

A: No. While the EPYC wins data compression, encryption, extended instructions, floating-point, integer math, and random string sorting, it loses the multithread test by 1.3% (65,811 vs 66,655). The Ryzen's higher per-core performance nearly compensates for its lower core count.

Q: What does the physics simulation score indicate?

A: The Ryzen 9 PRO 9965 scores 3,256 versus the EPYC's 1,905, a 41.5% advantage. This benchmark typically rewards higher clock speeds and efficient per-core execution, which aligns with the Ryzen's design priorities.

Q: How do these processors compare to their nearest rivals in the database?

A: The EPYC 9335 is nearly tied with the Intel Xeon 6741P (0.3% below) and slightly ahead of the Intel Xeon 678X (0.4% above). The Ryzen 9 PRO 9965 is 0.6% ahead of the AMD EPYC 7643P and 1% ahead of the Intel Xeon w9-3575X, while trailing the AMD EPYC 8434P by 0.8%.

Q: Is the Ryzen 9 PRO 9965 competitive in absolute multithread terms despite fewer cores?

A: Yes, the multithread score of 66,655 essentially matches the EPYC's 65,811. This is remarkable given the EPYC has 32 cores versus the Ryzen's 16, indicating the Ryzen's higher clock speeds and architectural efficiency close the gap in this particular workload.

Architecture Differences

Both processors share the Zen 5 architecture and a 4 nm TSMC manufacturing process, but they diverge significantly in implementation. The EPYC 9335 uses the Turin codename and comes from the EPYC 9005 series, while the Ryzen 9 PRO 9965 uses the Granite Ridge codename from the 9000 series. The EPYC packs 32 cores and 64 threads, exactly double the Ryzen's 16 cores and 32 threads.

The transistor counts reveal the scale difference. The EPYC integrates 33,260 million transistors across a 4x 70.6 mm² die configuration, while the Ryzen uses 16,630 million transistors on a 2x 70.6 mm² setup. Both use identical per-core die sizes, but the EPYC doubles the number of chiplets.

Cache hierarchies follow the same pattern. Both allocate 80 KB of L1 per core and 1 MB of L2 per core, but the EPYC's shared L3 cache totals 128 MB versus the Ryzen's 64 MB. This doubled L3 capacity gives the EPYC more room for large working sets in server workloads.

Memory architecture presents one of the most significant differences. The EPYC 9335 supports twelve-channel DDR5 memory with a bandwidth of 576.0 GB/s, while the Ryzen 9 PRO 9965 uses dual-channel DDR5 at 89.6 GB/s. That is a 6.4x bandwidth advantage for the EPYC, which explains its dominance in data-heavy benchmarks.

PCIe connectivity also differs substantially. The EPYC provides Gen 5 with 128 lanes (CPU only), while the Ryzen offers Gen 5 with 24 lanes. The EPYC targets multi-GPU and high-density storage configurations, while the Ryzen suits standard workstation builds. The EPYC has no integrated graphics, whereas the Ryzen includes Radeon Graphics.

The sockets reflect their different market positions. The EPYC uses AMD Socket SP5, designed for dual-socket-capable server platforms, while the Ryzen uses AMD Socket AM5, the mainstream desktop socket. Both support ECC memory, but the EPYC's twelve-channel design far exceeds the Ryzen's dual-channel capability.

Specification Differences

The core and thread counts present the most obvious divergence: the EPYC 9335 offers 32 cores and 64 threads, while the Ryzen 9 PRO 9965 provides 16 cores and 32 threads. Base clocks differ notably, with the EPYC running at 3.00 GHz and the Ryzen at 4.30 GHz. Boost clocks show a similar pattern: 4.40 GHz for the EPYC versus 5.50 GHz for the Ryzen.

Thermal design power differs by 40 watts, with the EPYC rated at 210 W and the Ryzen at 170 W. This aligns with the EPYC's higher core count and server-oriented power delivery. The EPYC's launch MSRP is $3178, while the Ryzen's launch MSRP is not recorded in the database.

Memory channels and bandwidth represent the most dramatic specification split. The EPYC supports twelve-channel DDR5 with 576.0 GB/s bandwidth, while the Ryzen supports dual-channel DDR5 at 89.6 GB/s. PCIe lanes differ equally starkly: 128 Gen 5 lanes for the EPYC versus 24 Gen 5 lanes for the Ryzen.

Cache configurations diverge in L3 only. Both share the same per-core L1 and L2 sizing, but the EPYC's shared L3 is 128 MB versus the Ryzen's 64 MB. The EPYC uses a 4x 70.6 mm² die layout with 33,260 million transistors, while the Ryzen uses a 2x 70.6 mm² layout with 16,630 million transistors.

Integrated graphics appear only on the Ryzen, which includes Radeon Graphics. The EPYC lists N/A for integrated graphics, relying on discrete GPUs or server management controllers. Both processors have locked multipliers and are marked as Active in production status, with the EPYC releasing on 2024-10-09 and the Ryzen on 2026-06-29.

The Verdict

The data points to two distinct usage profiles. The AMD EPYC 9335 is the choice for throughput-intensive server and workstation workloads where memory bandwidth and core count matter most. Its 32 cores, 64 threads, and 576.0 GB/s twelve-channel memory deliver decisive wins in data compression, encryption, extended instructions, floating-point math, integer math, and random string sorting. The 128 MB L3 cache and 128 PCIe Gen 5 lanes support large-scale virtualization, databases, and high-performance computing clusters.

The AMD Ryzen 9 PRO 9965 excels in single-thread and latency-sensitive tasks. Its 5.50 GHz boost clock and 4.30 GHz base clock produce a 41.6% single-thread advantage over the EPYC, along with a 41.5% lead in physics simulation and a 6.6% edge in prime number finding. The 170 W TDP and dual-channel memory make it suitable for desktop workstations where per-core responsiveness matters more than aggregate throughput.

The multithread result complicates the simple core-count narrative. The Ryzen essentially matches the EPYC in that test (66,655 vs 65,811), suggesting that for workloads that scale well with clock speed rather than raw core count, the Ryzen can hold its own. However, the EPYC's wins in the other five throughput tests are much larger in magnitude, with deltas ranging from 22.9% to 48.4%.

The percentile rankings show both are elite performers, with the EPYC at the 99th percentile and the Ryzen at the 98th. The EPYC's average score of 194,228 places it nearly even with the Intel Xeon 6741P (0.3% difference), while the Ryzen's 145,728 average sits close to the AMD EPYC 8434P (0.8% below) and the Intel Xeon w9-3575X (1% above).

Where Each One Wins

The AMD EPYC 9335 dominates in data-centric and security-focused workloads. Its 32.5% lead in data compression makes it ideal for storage systems, backup pipelines, and database logging. The 40.6% encryption advantage suits VPN gateways, secure transaction processing, and full-disk encryption servers. Extended instructions, with a 48.4% gap, benefit cryptographic hashing and specialized math libraries.

Floating-point and integer math see the EPYC ahead by 41.9% and 42.3% respectively, making it the stronger choice for scientific simulations, financial modeling, and rendering workloads that parallelize across many cores. Random string sorting, with a 22.9% advantage, supports log processing and data indexing at scale. The 128 MB L3 cache and twelve-channel memory bandwidth ensure these workloads have ample data feeding the cores.

The AMD Ryzen 9 PRO 9965 wins where single-thread speed dominates. Its 41.6% single-thread advantage makes it preferable for interactive applications, compile-heavy development workflows, and software that relies on serial execution. The 41.5% physics simulation lead points to game physics, collision detection, and real-time simulation tasks that cannot fully utilize 32 cores.

Prime number finding, with a 6.6% edge for the Ryzen, suggests advantages in certain cryptographic key generation and mathematical workloads that emphasize per-core integer throughput. The near-tie in multithread (1.3% Ryzen advantage) means the Ryzen can handle moderately parallel workloads without falling far behind, making it a versatile option for desktop workstations.

The Ryzen's integrated Radeon Graphics and 24 PCIe Gen 5 lanes make it suitable for single-GPU workstation builds, while the EPYC's 128 lanes and lack of integrated graphics target multi-GPU servers and storage arrays. The EPYC's twelve-channel memory versus the Ryzen's dual-channel reinforces this split: bandwidth-hungry applications belong on the EPYC, while latency-sensitive ones favor the Ryzen.