AMD EPYC 9355P vs AMD EPYC 9375F Comparison

FAQ

Q: Which processor has the higher average benchmark score, and by how much?

A: The AMD EPYC 9375F has an average benchmark score of 162497, while the AMD EPYC 9355P scores 160853. That puts the 9375F roughly 1% ahead of the 9355P in overall average performance.

Q: How do the two chips compare to the same external rivals?

A: Both sit in the 99th percentile of all CPUs. The 9375F trails the AMD EPYC 7C13 by 3.2% and the AMD Ryzen Threadripper PRO 3995WX by 5.4%, while leading the AMD EPYC 7663 by 0.3%. The 9355P trails the same two rivals by 4.1% and 6.3% respectively, and lags the EPYC 7663 by 0.7%.

Q: Which processor wins more head-to-head benchmark comparisons?

A: The AMD EPYC 9355P wins 11 of the 17 head-to-head tests, while the AMD EPYC 9375F wins 6. However, the 9375F’s wins include some very large margins in specific workloads.

Q: Are there any differences in core count, cache, memory support, or PCIe lanes?

A: No. Both processors feature 32 cores, 64 threads, 80 KB L1 per core, 1 MB L2 per core, 256 MB shared L3, DDR5 memory support, twelve-channel memory bus, 576.0 GB/s memory bandwidth, and Gen 5 PCIe with 128 lanes (CPU only).

Q: What is the biggest single benchmark margin between the two?

A: In the PassMark physics test, the AMD EPYC 9355P scores 14521 versus the 9375F’s 9019 — a 37.9% advantage for the 9355P. This is by far the largest delta in either direction.

Q: Do both processors share the same release date and production status?

A: Yes, both were released on 2024-10-09, are currently marked as Active in production, and belong to the EPYC Zen 5 (Turin) generation.

Architecture Differences

Both processors are built on the same fundamental architecture: AMD’s Zen 5 microarchitecture, codenamed Turin, fabricated on TSMC’s 4 nm process. The transistor count is identical at 66,520 million, and both use the same die configuration of 8x 70.6 mm² chiplets. The socket is the same AMD Socket SP5, and both are unlocked-free server parts.

The core and cache layout is also identical. Each has 32 cores and 64 threads, with 80 KB of L1 per core, 1 MB of L2 per core, and a shared 256 MB L3 cache. Memory support is the same: DDR5 with a twelve-channel bus and 576.0 GB/s of bandwidth, plus ECC support. PCIe connectivity is identical as well, with Gen 5 and 128 lanes available from the CPU.

Where they diverge is in operating frequencies and power envelope. The AMD EPYC 9375F has a base clock of 3.85 GHz and a boost clock of 4.80 GHz, while the AMD EPYC 9355P operates at a base of 3.55 GHz and boosts to 4.40 GHz. The 9375F carries a 320 W TDP, whereas the 9355P is rated at 280 W. This is a 40 W difference in thermal design power, which partly explains why the 9375F is the higher-clocked part.

The 9375F also carries a distinct part number (100-000001197) versus the 9355P (100-000001521), and the launch MSRP differs: $5306 for the 9375F versus $2998 for the 9355P. Neither part has an unlocked multiplier, and neither includes integrated graphics. Both are aimed at the server/workstation segment.

Where Each One Wins

The benchmark split reveals two very different performance profiles. The AMD EPYC 9355P dominates the Cinebench suite, winning all six Cinebench tests — R15, R20, and R23, both multi-core and single-core — by a consistent 1.5% margin. It also takes PassMark’s multithread, integer math, random string sorting, data encryption, and physics tests. The physics win is especially dramatic at 37.9%.

The AMD EPYC 9375F, despite losing the overall head-to-head count, wins in several specialized workloads. It leads in PassMark data compression by 4.5%, extended instructions by 20.1%, find prime numbers by 34.2%, floating point math by 2%, and single-thread performance by 0.6%. These are not marginal wins in every case; the prime number and extended instruction advantages are substantial.

The implication is that the 9355P is the more balanced all-rounder, particularly strong in rendering-style multi-threaded workloads and physics simulations. The 9375F, by contrast, appears tuned for specific computational tasks — prime number finding, extended instruction sets, and floating point math — where its higher clocks translate into outsized gains.

Specification Differences

The two chips are nearly identical on paper, so the notable differences are confined to clocks, power, and price.

• Base Clock: 3.85 GHz (9375F) vs 3.55 GHz (9355P)
• Boost Clock: 4.80 GHz (9375F) vs 4.40 GHz (9355P)
• TDP: 320 W (9375F) vs 280 W (9355P)
• Launch MSRP: $5306 (9375F) vs $2998 (9355P)

Everything else matches: 32 cores, 64 threads, same cache hierarchy, same memory bus and bandwidth, same PCIe configuration, same socket, same process node, same foundry, same transistor count, same die size, same release date, same production status, and same market segment. The only other difference is the part number.

Head-to-Head Benchmarks

The most striking result is in PassMark physics, where the 9355P scores 14521 against the 9375F’s 9019 — a 37.9% advantage. This is an outlier in magnitude and suggests the 9355P’s lower clocks somehow translate into dramatically better physics simulation performance, which is counterintuitive given the 9375F’s higher boost clock.

The 9355P also wins in data encryption by 10.1% (81945 vs 73634), random string sorting by 10.7% (180482 vs 161091), and integer math by 6.4% (414347 vs 387901). In multithread, it holds a 1.5% edge (97255 vs 95768). Across all six Cinebench tests, the 9355P maintains a uniform 1.5% lead, including Cinebench R23 multi-core with 82666 versus 81402.

The 9375F counters with its own set of wins. Find prime numbers shows a 34.2% advantage (1397 vs 1041), and extended instructions deliver a 20.1% lead (128296 vs 106821). Data compression favors the 9375F by 4.5% (1496149 vs 1431968), floating point math by 2% (260392 vs 255177), and single-thread by a slim 0.6% (3762 vs 3741). The 9375F also wins the duplicate PassMark single-thread test by the same 0.6% margin.

Looking at the averages, the 9375F’s overall benchmark score of 162497 is only 1% higher than the 9355P’s 160853. This narrow aggregate gap masks the fact that the two chips win in very different workload categories. The 9355P’s wins are more numerous but often modest (1.5% to 10.7%), while the 9375F’s wins include two blowouts above 20%.

The Verdict

From the data, the AMD EPYC 9355P is the better general-purpose processor. It wins 11 of 17 head-to-head tests, including all Cinebench benchmarks and the physics test by a massive 37.9%. Its average score is only 1% behind the 9375F, yet it does so with a 280 W TDP versus 320 W, and its launch MSRP is $2998 versus $5306 for the 9375F. For most multi-threaded server workloads, the 9355P is the stronger choice.

The AMD EPYC 9375F is the specialist. Its wins in prime number finding (34.2% ahead), extended instructions (20.1% ahead), and data compression (4.5% ahead) point to workloads that benefit from its higher clock speeds. If the task involves heavy floating-point math, extended instruction sets, or compression, the 9375F’s 4.80 GHz boost clock delivers measurable gains. Its single-thread performance is also slightly better at 0.6%.

The 9375F’s 1% average score advantage comes at a significant power cost — 40 W higher TDP — and a much higher launch MSRP. The 9355P achieves nearly identical average performance with lower power draw and a lower price point, yet the 9375F remains compelling for applications that hit its specific strengths. Buyers should match the workload to the chip: general server duties favor the 9355P, while specialized computational tasks justify the 9375F.