AMD EPYC 9475F vs AMD EPYC 9655 Comparison

FAQ

Q: Which processor delivers the higher average benchmark score?

A: The AMD EPYC 9655 posts an average benchmark score of 373,479, while the AMD EPYC 9475F reaches 350,933. The 9655 holds a 6.4% advantage over the 9475F in this metric, as listed in the nearestRivals data.

Q: How do the two chips compare in single-threaded performance?

A: The AMD EPYC 9655 scores 3,847 in the PassMark single-thread test, versus 3,779 for the AMD EPYC 9475F. That is a 1.8% edge for the 9655, a narrow margin that indicates near-parity in lightly threaded workloads.

Q: What is the core and thread count difference?

A: The AMD EPYC 9655 offers 96 cores and 192 threads. The AMD EPYC 9475F provides 48 cores and 96 threads. The 9655 has exactly double the core and thread count of the 9475F.

Q: Which part has the higher boost clock?

A: The AMD EPYC 9475F has a boost clock of 4.80 GHz, which is higher than the 4.50 GHz boost clock of the AMD EPYC 9655. The base clocks are 3.65 GHz for the 9475F and 2.60 GHz for the 9655.

Q: Do both processors use the same socket and memory configuration?

A: Yes. Both the AMD EPYC 9655 and AMD EPYC 9475F use AMD Socket SP5, support DDR5 memory, feature a twelve-channel memory bus, and deliver 576.0 GB/s of memory bandwidth. Both also support ECC memory.

Q: What is the L3 cache capacity on each chip?

A: The AMD EPYC 9655 has 384 MB of shared L3 cache. The AMD EPYC 9475F has 256 MB of shared L3 cache. The 9655 provides 128 MB more L3 cache.

Architecture Differences

The AMD EPYC 9655 and AMD EPYC 9475F are both members of the EPYC 9005 series, built on the Zen 5 architecture with the codename Turin. Both processors are manufactured by AMD on a 4 nm process node at TSMC. They share the same socket (AMD Socket SP5), the same memory support (DDR5 with a twelve-channel bus), and the same PCIe configuration (Gen 5, 128 lanes from the CPU). Integrated graphics are not applicable on either part.

The fundamental architectural divergence lies in the silicon configuration. The AMD EPYC 9655 is built from 12 compute dies, each measuring 70.6 mm², for a total die size of 12x 70.6 mm². The transistor count is 99,780 million. In contrast, the AMD EPYC 9475F uses 8 compute dies of the same 70.6 mm² size, resulting in a total die size of 8x 70.6 mm² and a transistor count of 66,520 million. This means the 9655 packs roughly 33,260 million more transistors, which aligns with its doubled core count.

Cache hierarchy differs accordingly. Both chips share the same per-core L1 cache (80 KB per core) and per-core L2 cache (1 MB per core). The shared L3 cache, however, scales with the number of dies: the 9655 has 384 MB shared L3, while the 9475F has 256 MB shared L3. The extra 128 MB of L3 on the 9655 is a direct consequence of the additional compute dies.

Clock speeds favor the 9475F. The 9475F has a base clock of 3.65 GHz and a boost clock of 4.80 GHz. The 9655 has a base clock of 2.60 GHz and a boost clock of 4.50 GHz. The 9475F offers a 1.05 GHz higher base clock and a 0.30 GHz higher boost clock. Both parts are rated at a 400 TDP and are not multiplier unlocked.

Both processors were released on the same date, 2024-10-09, and are currently in active production. The market segment for both is Server/Workstation.

Head-to-Head Benchmarks

The head-to-head benchmark data shows a decisive sweep: the AMD EPYC 9655 wins all 11 recorded comparisons, with zero wins for the AMD EPYC 9475F. The margins vary widely by workload type.

The largest delta is in PassMark integer math, where the 9655 scores 1,139,161 against 605,696 for the 9475F. That is an 88.1% advantage, nearly doubling the integer throughput. This is the single biggest gap in the dataset.

Data encryption shows the second-largest margin. The 9655 scores 210,555 versus 116,648 for the 9475F, an 80.5% lead. Random string sorting follows closely: 439,682 for the 9655 versus 253,936 for the 9475F, a 73.1% difference. Floating-point math also skews heavily toward the 9655, with a score of 662,958 versus 406,524, a 63.1% edge.

Physics calculations favor the 9655 by 57.8%, with scores of 25,947 versus 16,443. Data compression shows a 51.7% gap: 3,271,896 for the 9655 versus 2,156,305 for the 9475F. The multithread benchmark gives the 9655 a 27.5% advantage, scoring 156,110 versus 122,476.

Extended instructions are closer but still favor the 9655: 203,285 versus 173,169, a 17.4% delta. The narrowest margins appear in find prime numbers (1,598 versus 1,507, a 6.0% difference) and single-thread performance (3,847 versus 3,779, a 1.8% difference).

The pattern is consistent: the 9655 wins by large margins in heavily parallel workloads (integer math, encryption, floating point, sorting, compression) and by smaller margins in single-threaded or lightly threaded tasks. The 9475F's higher boost clock does not translate into a single-thread win; the 9655 still leads that test, albeit narrowly.

Specification Differences

The two processors share many specifications but differ in the following fields:

• Cores: 96 (9655) versus 48 (9475F)
• Threads: 192 (9655) versus 96 (9475F)
• Base clock: 2.60 GHz (9655) versus 3.65 GHz (9475F)
• Boost clock: 4.50 GHz (9655) versus 4.80 GHz (9475F)
• Transistors: 99,780 million (9655) versus 66,520 million (9475F)
• Die size: 12x 70.6 mm² (9655) versus 8x 70.6 mm² (9475F)
• L3 cache: 384 MB shared (9655) versus 256 MB shared (9475F)
• Launch MSRP: $11852 (9655) versus $7592 (9475F)

Identical specifications include the socket (SP5), architecture (Zen 5, Turin), process node (4 nm, TSMC), TDP (400), L1 cache (80 KB per core), L2 cache (1 MB per core), memory support (DDR5, twelve-channel, 576.0 GB/s), ECC support, PCIe (Gen 5, 128 lanes), integrated graphics (N/A), market segment (Server/Workstation), production status (Active), release date (2024-10-09), and multiplier unlocked status (false).

The Verdict

The benchmark data presents a clear hierarchy. The AMD EPYC 9655 outperforms the AMD EPYC 9475F in every recorded test. The average benchmark score difference is 6.4% in favor of the 9655, but the head-to-head deltas range from 1.8% to 88.1%. The 9655 is the superior processor for throughput-oriented workloads, particularly those that scale with core count and cache capacity.

The AMD EPYC 9475F is not without merit. Its higher base and boost clocks (3.65 GHz and 4.80 GHz) suggest a design optimized for frequency-sensitive tasks, and its single-thread score of 3,779 is only 1.8% behind the 9655. However, in the data provided, the 9475F does not win a single benchmark. Its lower core count (48 versus 96) and smaller L3 cache (256 MB versus 384 MB) put it at a structural disadvantage in multi-threaded scenarios.

The nearest-rival data places the 9655 ahead of the 9475F by 6.4%, with the 9475F also trailing the AMD EPYC 9754 (-3.7%), the Intel Xeon 6960P (-3.9%), and the AMD EPYC 9535 (-7.5%). The 9655, meanwhile, leads the AMD EPYC 9535 in the rival list by 1.6% in the opposite direction (the 9535 has a higher avg score of 379,408, but the 9655 trails by only 1.6%). The data shows the 9655 is competitive at the top of the stack, while the 9475F sits slightly below several other EPYC and Xeon parts in average score.

For buyers prioritizing raw multi-core throughput, the 9655 is the clear choice. For those who need maximum clock speed per core and are willing to sacrifice core count, the 9475F offers a more frequency-focused profile, but the data does not show a workload where that translates into a win.

Where Each One Wins

AMD EPYC 9655 wins in: All 11 head-to-head benchmarks. The largest margins are in integer math (88.1% ahead), data encryption (80.5% ahead), and random string sorting (73.1% ahead). It also leads decisively in floating-point math (63.1%), physics (57.8%), data compression (51.7%), and multithread (27.5%). Even in the closest tests, extended instructions (17.4%), find prime numbers (6.0%), and single-thread (1.8%), the 9655 prevails. This makes it the preferred part for database workloads, scientific computing, virtualization, and any application that can utilize 96 cores and 384 MB of L3 cache.

AMD EPYC 9475F wins in: No recorded benchmark. The data shows zero wins for the 9475F across all 11 comparisons. Its strongest relative showing is in single-thread performance, where it trails by just 1.8%, and find prime numbers, where it trails by 6.0%. The 9475F's higher boost clock (4.80 GHz versus 4.50 GHz) does not overcome the 9655's architectural advantages in any test. Its 48 cores and 256 MB L3 cache are still substantial, but against the 9655's 96 cores and 384 MB L3, the 9475F is consistently behind. The 9475F may be a reasonable choice for lightly threaded workloads where its clock speed could theoretically help, but the available data does not support a win in any such scenario.