AMD EPYC 9575F vs AMD EPYC 9655P Comparison

The AMD EPYC 9655P and AMD EPYC 9575F are both Zen 5 Turin parts on the same SP5 platform, but they target different workloads. The 9655P is a 96-core, 192-thread monster that dominates almost every multi-threaded benchmark in this comparison, winning 15 of 17 head-to-head tests. The 9575F, with 64 cores and 128 threads, fights back with a higher boost clock and takes the single-thread PassMark crown, but its overall average benchmark score of 311,774 sits well behind the 9655P’s 397,773. The data is unambiguous: the 9655P is the throughput king, while the 9575F is the pick for latency-sensitive single-thread tasks.

The Verdict

From the benchmark results, the AMD EPYC 9655P is the clear choice for any workload that scales with core count. It leads in Cinebench R15, R20, and R23 across both multi-core and single-core tests, with margins ranging from 6.7% to 6.8%. Its PassMark wins are even more decisive, with leads of 25.2% in physics, 35.2% in data encryption, and 38.8% in prime number finding. The 9655P also holds a 100th percentile ranking among all CPUs, versus the 9575F’s 99th percentile, and its nearest rival is the AMD Ryzen Threadripper PRO 9995WX, which scores 3.5% higher on average; the 9575F’s closest competitor is the AMD EPYC 9734, which is just 0.4% behind it.

The AMD EPYC 9575F is the specialist’s tool. Its only wins in the entire head-to-head are PassMark single-thread and singlethread tests, where it scores 4173 versus 3849, a 7.8% advantage. If your application is bound by single-thread performance and cannot use more than 64 cores, the 9575F’s higher boost clock of 5.00 GHz makes it the right part. But for nearly everything else, the 9655P’s additional 32 cores and 64 threads translate into massive real-world performance advantages. The verdict is simple: choose the 9655P for scale-out compute, choose the 9575F for frequency-hungry single-thread tasks.

Architecture Differences

Both processors are built on the same Zen 5 architecture, codenamed Turin, and are manufactured by TSMC on a 4 nm process node. The 9655P uses 12 compute dies, each 70.6 mm², totaling 99,780 million transistors. The 9575F uses 8 dies of the same size, with 66,520 million transistors. This die count difference explains the core disparity: 96 cores and 192 threads for the 9655P versus 64 cores and 128 threads for the 9575F.

Cache configurations scale with the core counts. Each core on both chips gets 80 KB of L1 and 1 MB of L2. The shared L3 cache, however, differs significantly: the 9655P has 384 MB shared across the chip, while the 9575F has 256 MB. This 128 MB difference is directly tied to the number of active dies. Memory support is identical: both use DDR5 across a twelve-channel bus, delivering 576.0 GB/s of bandwidth, and both support ECC memory.

Clock speeds are where the 9575F pushes back. The 9575F has a base clock of 3.30 GHz and a boost clock of 5.00 GHz, while the 9655P runs at 2.60 GHz base and 4.50 GHz boost. Despite the lower clocks, the 9655P still wins all Cinebench single-core tests, posting 1940 in R15, 8085 in R20, and 19250 in R23 versus the 9575F’s 1817, 7573, and 18033. This suggests the 9655P’s larger L3 cache and higher core count help even in lightly threaded scenarios, though the 9575F’s lead in PassMark single-thread (4173 vs 3849) shows the higher boost clock can win out in certain instruction mixes.

Both CPUs share the same socket (AMD Socket SP5), PCIe Gen 5 with 128 lanes from the CPU, and a 400 W TDP. Neither has integrated graphics, and both are unlocked for overclocking? No, both have multiplierUnlocked set to false. They were released on the same date, 2024-10-09, and carry launch MSRPs of $10811 for the 9655P and $11791 for the 9575F.

Head-to-Head Benchmarks

The Cinebench suite shows a consistent pattern: the 9655P wins every test by nearly identical margins. In R15 multi-core, it scores 13744 against 12876, a 6.7% lead. The single-core R15 test shows 1940 versus 1817, a 6.8% edge. R20 multi-core yields 57268 against 53650 (6.7%), and R20 single-core is 8085 versus 7573 (6.8%). R23 multi-core follows with 136354 versus 127739 (6.7%), and R23 single-core with 19250 versus 18033 (6.7%). These results indicate that the 9655P’s architecture advantages — more cores, more cache — provide a uniform uplift across rendering workloads.

The PassMark suite amplifies the 9655P’s dominance in multi-threaded tasks. Data compression shows the biggest delta at 25.7% (3486158 vs 2773634). Data encryption is even starker: 220074 versus 162818, a 35.2% gap. Extended instructions come in at 16.8% (230609 vs 197484). Find prime numbers delivers a 38.8% win for the 9655P (1686 vs 1215). Floating point math is 36.1% ahead (715866 vs 526003), and integer math is 37.4% ahead (1225251 vs 891817). The multithread test shows a smaller 8.4% lead (160490 vs 147998), while physics is 25.2% ahead (25847 vs 20652). Random string sorting rounds out the wins with a 29.6% margin (451824 vs 348500).

The 9575F’s two wins are in PassMark single-thread and singlethread tests, both scoring 4173 versus 3849, a 7.8% advantage. This is the only area where the higher boost clock of 5.00 GHz translates into a benchmark victory. Notably, the 9655P still wins the Cinebench single-core tests despite the clock deficit, which highlights that benchmark-specific instruction handling matters more than raw frequency in some cases.

FAQ

Q: Which CPU has more cores and threads?

A: The AMD EPYC 9655P has 96 cores and 192 threads. The AMD EPYC 9575F has 64 cores and 128 threads.

Q: What are the boost clock speeds of these processors?

A: The AMD EPYC 9655P has a boost clock of 4.50 GHz. The AMD EPYC 9575F has a higher boost clock of 5.00 GHz.

Q: How much L3 cache does each processor have?

A: The AMD EPYC 9655P has 384 MB of shared L3 cache. The AMD EPYC 9575F has 256 MB of shared L3 cache.

Q: Which CPU wins in PassMark single-thread performance?

A: The AMD EPYC 9575F wins, scoring 4173 compared to the AMD EPYC 9655P’s 3849, a 7.8% advantage.

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

A: The largest margin is in PassMark find prime numbers, where the AMD EPYC 9655P scores 1686 versus 1215, a 38.8% lead.

Q: Do both CPUs support the same memory configuration?

A: Yes, both support DDR5 on a twelve-channel bus with 576.0 GB/s bandwidth, and both support ECC memory.

Where Each One Wins

The AMD EPYC 9655P is the overwhelming winner for multi-threaded and parallel workloads. Its 96 cores and 192 threads deliver decisive leads in every Cinebench multi-core test and in 13 of 15 PassMark tests. The data shows particularly strong performance in compute-heavy tasks: integer math (37.4% ahead), floating point math (36.1%), and data encryption (35.2%). For database workloads, the 9655P’s advantage in random string sorting (29.6%) and data compression (25.7%) makes it the superior choice. Its 384 MB of L3 cache likely contributes to these wins, providing more on-die data for multi-threaded access patterns. The 9655P also wins all Cinebench single-core tests, despite the 9575F’s higher boost clock, which means it is not a slouch even in lightly threaded scenarios.

The AMD EPYC 9575F has exactly one arena where it wins: PassMark single-thread performance. Its 5.00 GHz boost clock delivers a 7.8% lead over the 9655P in that specific test. This makes it the pick for applications that are strictly serial and cannot leverage more than 64 cores. However, the 9575F’s overall average benchmark score of 311,774 is 27.6% lower than the 9655P’s 397,773, and its 99th percentile ranking versus 100th percentile shows it is a step below in general compute capability. For users running high-frequency trading, legacy single-threaded simulation code, or any workload where a single core’s speed is the bottleneck, the 9575F’s clock advantage is the deciding factor. For everyone else, the 9655P’s core and cache advantages make it the default recommendation from the data.