AMD EPYC 9335 vs AMD Ryzen Threadripper 9970X Comparison

Head-to-Head Benchmarks

The recorded data shows a clean sweep for the AMD Ryzen Threadripper 9970X across all eleven benchmark tests, with no wins recorded for the AMD EPYC 9335. The most dramatic gap appears in the passmark_physics test, where the Threadripper scores 6835 against the EPYC's 1905, a delta of 258.8%. That is not a marginal lead; it is a fundamental difference in how the two processors handle physics simulation workloads.

The multithreaded test tells a similar story, with the Threadripper posting 107399 versus the EPYC's 65811, a 63.2% advantage. The single-thread test shows the Threadripper at 4530 against the EPYC's 2732, a 65.8% lead. Random string sorting also favors the Threadripper heavily: 191445 versus 116608, a 64.2% delta. These are not narrow victories; the Threadripper is decisively ahead in every measured category.

Looking at compute-heavy integer work, the Threadripper scores 465378 in integer math versus 346291 for the EPYC, a 34.4% edge. Floating point math shows a similar pattern: 309719 against 228123, a 35.8% advantage. Data compression, a common server and workstation workload, favors the Threadripper at 1757998 versus 1203096, a 46.1% delta. Encryption workloads are also faster on the Threadripper: 86765 versus 63159, a 37.4% gap. Extended instruction workloads, which often stress AVX and similar features, show the Threadripper ahead by 34.7% (142342 versus 105706). The prime number test, which is highly sensitive to clock speed and memory latency, shows the Threadripper at 615 versus the EPYC's 340, an 80.9% delta.

In context, the Threadripper's average benchmark score is 279778, placing it in the 99th percentile of all CPUs. Its nearest rivals in the database include the Intel Xeon 6780E at 280438 (a 0.2% delta), the AMD EPYC 9565 at 285471 (a 2% delta), the Intel Xeon 696X at 286102 (a 2.2% delta), and the AMD EPYC 9555P at 287066 (a 2.5% delta). The EPYC 9335, by contrast, averages 194228, also in the 99th percentile, but its nearest rivals are lower-tier parts: the Intel Xeon 6741P at 194901 (0.3% delta), the Intel Xeon 678X at 193477 (0.4% delta), the Intel Xeon 6740E at 187718 (3.5% delta), and the AMD EPYC 8534P at 185092 (4.9% delta). The data indicates the Threadripper sits in a higher performance class overall, even though both are 32-core Zen 5 parts.

Architecture Differences

Both processors are built on Zen 5 architecture and manufactured by TSMC on a 4 nm process node. Both use the same transistor count of 33,260 million and the same die configuration of 4x 70.6 mm². The core and thread counts are identical at 32 cores and 64 threads. The L2 cache is also the same at 1 MB per core, and both have 128 MB of L3 cache, though the EPYC lists its L3 as shared while the Threadripper does not specify that detail.

The first major architectural divergence is clock speed. The Threadripper runs a base clock of 4.00 GHz and a boost clock of 5.40 GHz. The EPYC runs at 3.00 GHz base and 4.40 GHz boost. That 1.0 GHz base clock gap and 1.0 GHz boost gap explain much of the benchmark delta, especially in single-threaded and lightly threaded workloads. The Threadripper also has a higher thermal envelope at 350 watts, which allows those clocks to be sustained, while the EPYC is rated at 210 watts.

The L1 cache differs: the Threadripper has 64 KB per core, while the EPYC has 80 KB per core. That is a notable difference, as a larger L1 cache can improve certain workloads, but the EPYC does not convert that advantage into benchmark wins in the recorded data.

Socket and platform are entirely different. The Threadripper uses AMD Socket sTR5, while the EPYC uses AMD Socket SP5. The Threadripper is a desktop part with a multiplier unlocked, meaning users can adjust clock multipliers. The EPYC is a server/workstation part with a locked multiplier. Memory support is DDR5 for both, but the memory bus width is very different: the Threadripper runs quad-channel memory with a bandwidth of 204.8 GB/s, while the EPYC runs twelve-channel memory with a bandwidth of 576.0 GB/s. That is a massive memory bandwidth advantage for the EPYC, yet it does not translate into higher benchmark scores in the recorded tests.

PCIe connectivity also differs. The Threadripper provides Gen 5 with 80 lanes (CPU only). The EPYC provides Gen 5 with 128 lanes (CPU only). The market segment is split: the Threadripper is classified as Desktop, the EPYC as Server/Workstation. The Threadripper has no integrated graphics, and neither does the EPYC. Both support ECC memory.

Release timing is also different. The Threadripper launched on 2025-07-29, while the EPYC launched on 2024-10-09. Both are listed as Active in production status. The Threadripper carries part number 100-000001594, and the EPYC carries 100-000001149.

FAQ

Q: Which processor has the higher boost clock?

A: The AMD Ryzen Threadripper 9970X boosts to 5.40 GHz, while the AMD EPYC 9335 boosts to 4.40 GHz. That 1.0 GHz difference is reflected in the Threadripper's 65.8% lead in single-thread performance.

Q: Why does the EPYC have a much higher memory bandwidth rating?

A: The EPYC 9335 uses a twelve-channel memory bus, delivering 576.0 GB/s, while the Threadripper 9970X uses a quad-channel bus, delivering 204.8 GB/s. Even with that bandwidth advantage, the EPYC does not win any of the eleven recorded benchmark tests.

Q: Are these processors based on the same architecture?

A: Yes, both use Zen 5 architecture. The Threadripper is codenamed Shimada Peak, and the EPYC is codenamed Turin. Both are built on TSMC's 4 nm process with 33,260 million transistors and a 4x 70.6 mm² die configuration.

Q: Which processor supports more PCIe lanes?

A: The EPYC 9335 supports 128 Gen 5 lanes (CPU only), while the Threadripper 9970X supports 80 Gen 5 lanes (CPU only). The EPYC also has a twelve-channel memory bus compared to the Threadripper's quad-channel.

Q: What is the largest benchmark gap between the two?

A: The passmark_physics test shows the largest gap. The Threadripper scores 6835 versus the EPYC's 1905, a delta of 258.8%. The second-largest gap is in multithread performance at 63.2%.

Q: Is the Threadripper unlocked for overclocking?

A: Yes, the Threadripper 9970X has an unlocked multiplier. The EPYC 9335 does not have an unlocked multiplier.

Specification Differences

| Field | AMD Ryzen Threadripper 9970X | AMD EPYC 9335 |

|-------|-------------------------------|---------------|

| Series | 9000 series | EPYC 9005 series |

| Base Clock | 4.00 GHz | 3.00 GHz |

| Boost Clock | 5.40 GHz | 4.40 GHz |

| TDP | 350 W | 210 W |

| Socket | AMD Socket sTR5 | AMD Socket SP5 |

| Codename | Shimada Peak | Turin |

| Generation | Ryzen Threadripper (Zen 5, Shimada Peak) | EPYC (Zen 5, Turin) |

| L1 Cache | 64 KB (per core) | 80 KB (per core) |

| L3 Cache | 128 MB | 128 MB (shared) |

| Memory Bus | Quad-channel | Twelve-channel |

| Memory Bandwidth | 204.8 GB/s | 576.0 GB/s |

| PCIe | Gen 5, 80 Lanes (CPU only) | Gen 5, 128 Lanes (CPU only) |

| Market Segment | Desktop | Server/Workstation |

| Launch MSRP | $2499 | $3178 |

| Multiplier Unlocked | Yes | No |

| Part Number | 100-000001594 | 100-000001149 |

| Release Date | 2025-07-29 | 2024-10-09 |

Where Each One Wins

The AMD Ryzen Threadripper 9970X wins in every recorded benchmark category. If the task is anything measured by the database, the Threadripper is the faster part. Its advantages are especially pronounced in single-threaded work (65.8% faster), multithreaded work (63.2% faster), and physics simulation (258.8% faster). The prime number test, which often reflects the combination of clock speed and memory latency, shows an 80.9% advantage. For desktop and workstation users who prioritize raw compute throughput, the Threadripper is the clear choice, and its unlocked multiplier offers additional tuning flexibility.

The AMD EPYC 9335 does not win any of the eleven benchmark tests. Its strengths are architectural rather than performance-based in these measurements. It offers a twelve-channel memory bus with 576.0 GB/s of bandwidth, which is 2.8 times the Threadripper's rated bandwidth, and it provides 128 PCIe Gen 5 lanes, which is 48 more than the Threadripper. The EPYC also has a higher L1 cache per core at 80 KB versus 64 KB, and it consumes 140 watts less power by TDP rating. For server deployments where memory capacity, I/O expansion, and power efficiency matter more than single-thread or physics performance, the EPYC 9335 is the appropriate platform. Its locked multiplier is irrelevant in server environments where overclocking is not a consideration.

The data also shows that the Threadripper sits in a higher performance tier relative to its nearest rivals, with an average benchmark score of 279778 compared to the EPYC's 194228. The Threadripper's closest competitor is the Intel Xeon 6780E at a 0.2% delta, while the EPYC's closest competitor is the Intel Xeon 6741P at a 0.3% delta. Both processors are in the 99th percentile of all CPUs, but the Threadripper is operating in a different performance bracket entirely. Users who need the absolute fastest compute in these specific workloads should choose the Threadripper. Users who need massive memory bandwidth, extensive PCIe expansion, and lower power draw in a server form factor should choose the EPYC. The benchmark data does not show any workload where the EPYC outperforms the Threadripper, but the platform-level features of the EPYC address a different set of requirements.