CPU Comparison

Head-to-Head Benchmarks

The head-to-head data shows a clear split between two very different performance profiles. The AMD EPYC 9575F wins 9 of the 14 benchmark comparisons, while the Intel Xeon 6960P takes 5. The margin of victory tells the real story: AMD's wins are often decisive, while Intel's are frequently narrow.

The most striking result is in single-threaded performance. The EPYC 9575F scores 4,173 in PassMark single-thread testing against the Xeon's 3,287, a 21.2% advantage. This is the largest single-core gap in the entire comparison, and it directly reflects the architectural difference between the two parts. The same pattern appears in Cinebench R23 single-core, where AMD posts 18,033 versus Intel's result, though that test was not run on the Xeon 6960P in this dataset.

Multi-core rendering workloads reinforce AMD's dominance. In Cinebench R15 multi-core, the EPYC 9575F scores 12,876 against the Xeon's 11,194, a 13.1% lead. The identical 13.1% delta repeats in Cinebench R20 (53,650 vs 46,645) and Cinebench R23 (127,739 vs 111,060). This consistency across all three Cinebench versions suggests a stable throughput advantage rather than a workload-specific quirk.

Integer math shows the largest absolute gap. The EPYC 9575F posts 891,817 in PassMark integer math versus 727,750 for the Xeon, an 18.4% deficit for Intel. This is particularly relevant for database workloads, financial calculations, and general server logic. The PassMark multithread score follows the same direction: 147,998 for AMD versus 130,659 for Intel, an 11.7% lead.

However, the Xeon 6960P has its own victories, and some are substantial. In PassMark find prime numbers, Intel scores 1,484 against AMD's 1,215, a 22.1% advantage. This is the single largest Intel win in the entire dataset. PassMark physics shows a similar pattern: 24,937 for Intel versus 20,652 for AMD, a 20.7% lead. These two workloads suggest Intel's architecture handles certain algorithmic patterns more efficiently.

Random string sorting favors Intel by 6.7% (371,795 vs 348,500). Data compression is nearly a tie: Intel's 2,797,724 versus AMD's 2,773,634 gives Intel a 0.9% edge. Floating-point math is even closer, with Intel ahead by just 0.3% (527,473 vs 526,003). Data encryption is essentially a dead heat, with AMD winning by 0.5% (162,818 vs 162,013). Extended instructions also favor AMD by 2.1% (197,484 vs 193,404).

The average benchmark score tells a different story than the head-to-head wins. The Xeon 6960P has an average score of 365,194 across all benchmarks, placing it in the 100th percentile of all CPUs. The EPYC 9575F averages 311,774, which is the 99th percentile. The Xeon's nearest rival is the AMD EPYC 9754, which trails by just 0.2%. The AMD EPYC 9655 beats the Xeon by 2.2%, and the EPYC 9535 beats it by 3.7%. For the EPYC 9575F, its closest rival is the AMD EPYC 9734, which trails by 0.4%; the Intel Xeon 6781P beats it by 1.2%.

The Verdict

The benchmark data indicates two distinct purchasing rationales. The AMD EPYC 9575F is the stronger choice for workloads that depend on single-threaded speed, integer math, and consistent multi-core scaling. Its 21.2% single-thread lead and 18.4% integer math advantage make it particularly well-suited for high-frequency transaction processing, real-time analytics, and applications where per-core performance directly impacts latency-sensitive operations.

The Intel Xeon 6960P, despite winning fewer head-to-head tests, shows its strength in specific algorithmic domains. The 22.1% lead in prime number finding and 20.7% lead in physics simulation indicate that certain computational patterns, likely those involving branch prediction, specific instruction sequences, or memory access patterns, execute more efficiently on Intel's design. For workloads in scientific computing, physics simulation, or cryptographic prime-related operations, the Xeon's architecture may deliver better throughput despite lower raw scores elsewhere.

The average benchmark scores complicate the picture. The Xeon's 365,194 average places it in the 100th percentile and above the EPYC 9575F's 311,774 average. This suggests that across a broader benchmark suite, the Xeon's overall balance of strengths and weaknesses yields a higher composite score. The Xeon also holds its own against its nearest rivals, the EPYC 9754 is nearly tied at 0.2% behind, and only the EPYC 9655 and 9535 clearly surpass it. The EPYC 9575F, by contrast, sits slightly below several competitors, including the Xeon 6781P and both Threadripper PRO 9985WX and Threadripper 9980X.

For buyers prioritizing raw multi-core throughput in rendering, compilation, or data processing, the EPYC 9575F's consistent 13.1% Cinebench lead and 11.7% multithread advantage make a compelling case. For buyers who need the highest possible performance in physics simulation, prime computation, or string sorting, the Xeon 6960P is the data-backed choice. The Xeon also offers more cores (72 versus 64) and more threads (144 versus 128), which may matter for highly parallel workloads that scale linearly with core count.

Architecture Differences

The two processors come from fundamentally different design philosophies. The Intel Xeon 6960P uses the Granite Rapids architecture, built on Intel's 5 nm process node with a die size of 3x 598 mm². The AMD EPYC 9575F uses the Zen 5 architecture, codenamed Turin, built on TSMC's 4 nm process with a die size of 8x 70.6 mm². AMD's design uses 66,520 million transistors across its eight chiplets, while Intel's transistor count is not listed.

Cache hierarchies differ significantly. The Xeon provides 112 KB of L1 cache per core, 2 MB of L2 per core, and a massive 432 MB of shared L3 cache. The EPYC 9575F offers 80 KB of L1 per core, 1 MB of L2 per core, and 256 MB of shared L3. Intel's substantially larger L3 cache, 432 MB versus 256 MB, likely contributes to its strong performance in data compression and string sorting, where large working sets benefit from on-die storage.

The memory controllers also differ. Both support DDR5 with twelve-channel memory buses, but Intel's memory bandwidth is rated at 614.4 GB/s versus AMD's 576.0 GB/s. The Xeon's higher memory bandwidth may explain its narrow wins in floating-point math (0.3%) and data compression (0.9%), where memory throughput often limits performance.

PCIe connectivity diverges as well. The Xeon provides 96 Gen 5 lanes, while the EPYC 9575F offers 128 Gen 5 lanes. AMD's higher lane count gives it more headroom for storage arrays, GPUs, and network interfaces in dense server configurations.

The manufacturing approach differs fundamentally: Intel uses a monolithic design with three large dies, while AMD uses eight smaller chiplets on a 4 nm process. AMD's smaller dies (70.6 mm² each) typically yield better manufacturing efficiency, while Intel's larger dies (598 mm² each) enable tighter integration but with different trade-offs.

Specification Differences

| Specification | Intel Xeon 6960P | AMD EPYC 9575F |

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

| Cores | 72 | 64 |

| Threads | 144 | 128 |

| Base Clock | 2.70 GHz | 3.30 GHz |

| Boost Clock | 3.90 GHz | 5.00 GHz |

| TDP | 500 W | 400 W |

| Socket | Intel Socket 7529 | AMD Socket SP5 |

| Process Node | 5 nm | 4 nm |

| Foundry | Intel | TSMC |

| L1 Cache | 112 KB (per core) | 80 KB (per core) |

| L2 Cache | 2 MB (per core) | 1 MB (per core) |

| L3 Cache | 432 MB (shared) | 256 MB (shared) |

| Memory Bandwidth | 614.4 GB/s | 576.0 GB/s |

| PCIe | Gen 5, 96 Lanes | Gen 5, 128 Lanes |

| Launch MSRP | $9625 | $11791 |

| Release Date | 2024-09-23 | 2024-10-09 |

| Part Number | SRPKX | 100-000001554 |

The Xeon features more cores and threads, a larger L3 cache, and higher memory bandwidth. The EPYC 9575F features higher base and boost clocks, a smaller process node, more PCIe lanes, and a lower TDP. The EPYC 9575F's launch MSRP is $11,791, while the Xeon 6960P's is $9,625. Both are active production parts for the server/workstation market segment, both support ECC memory, and neither has integrated graphics or an unlocked multiplier.

FAQ

Q: Which processor has a higher single-threaded score?

A: The AMD EPYC 9575F scores 4,173 in PassMark single-thread testing, which is 21.2% higher than the Intel Xeon 6960P's 3,287.

Q: How do the two compare in Cinebench R23 multi-core?

A: The EPYC 9575F scores 127,739 versus the Xeon's 111,060, giving AMD a 13.1% advantage across all three Cinebench multi-core versions.

Q: What is the biggest performance gap between them?

A: The largest gap is in PassMark find prime numbers, where the Xeon 6960P leads by 22.1% (1,484 vs 1,215). The second-largest is single-thread performance, where AMD leads by 21.2%.

Q: Which processor has more cores and threads?

A: The Intel Xeon 6960P has 72 cores and 144 threads, while the AMD EPYC 9575F has 64 cores and 128 threads.

Q: What are the memory bandwidth specifications?

A: The Xeon 6960P has 614.4 GB/s memory bandwidth, while the EPYC 9575F has 576.0 GB/s. Both use twelve-channel DDR5 memory.

Q: How do their average benchmark scores compare?

A: The Xeon 6960P averages 365,194 across all benchmarks, placing it in the 100th percentile of all CPUs. The EPYC 9575F averages 311,774, placing it in the 99th percentile.