AMD EPYC 9335 vs Intel Xeon 6745P Comparison

The Verdict

The data splits cleanly by workload. The Intel Xeon 6745P dominates the benchmark record, winning 10 of 11 head-to-head comparisons, with particularly large margins in physics simulation, prime number finding, and multithreaded throughput. The AMD EPYC 9335 wins exactly one test, integer math, by a narrow 2.8% margin. If your workloads are general server tasks, database operations, encryption, or floating-point math, the Intel Xeon 6745P is the clear choice from the measurements. The AMD EPYC 9335 only makes sense if integer math performance is your absolute priority, and even then, the margin is slim enough that platform-level considerations would likely decide the purchase. The Intel part also posts a 26.2% higher PassMark multithread score (84210 vs 65811) and a 26.3% higher single-thread score (3450 vs 2732), which reinforces its overall lead. Neither processor is unlocked, and both target the server/workstation segment, but the EPYC 9335 carries a launch MSRP of $3178, while the Xeon 6745P carries a launch MSRP of $5250.

Architecture Differences

The two processors represent fundamentally different design approaches. The AMD EPYC 9335 uses the Zen 5 architecture, codenamed Turin, built on a 4 nm process at TSMC. It packs 32 cores and 64 threads, with a base clock of 3.00 GHz and a boost clock of 4.40 GHz. The chip integrates 33,260 million transistors across a 4x 70.6 mm² die configuration, and it runs on AMD Socket SP5. Cache distribution is per-core: 80 KB of L1 and 1 MB of L2 per core, with a shared 128 MB L3 pool.

The Intel Xeon 6745P uses the Granite Rapids architecture, built on a 5 nm process at Intel. It also has 32 cores and 64 threads, with a slightly higher base clock of 3.10 GHz but a slightly lower boost clock of 4.30 GHz. The die is 2x 598 mm², and it runs on Intel Socket 4710. Cache is more generous per core: 112 KB of L1 and 2 MB of L2 per core, plus a massive 336 MB shared L3. That 336 MB L3 is 2.6 times larger than the AMD chip's 128 MB L3.

Memory architecture differs significantly. The AMD EPYC 9335 supports twelve-channel DDR5 with a theoretical bandwidth of 576.0 GB/s. The Intel Xeon 6745P supports eight-channel DDR5 with 409.6 GB/s. The AMD part also offers more PCIe lanes: 128 Gen 5 lanes versus 88 Gen 5 lanes on the Intel chip. Both support ECC memory, and neither has integrated graphics. The AMD part was released on October 9, 2024, while the Intel part launched later, on February 23, 2025.

Where Each One Wins

The Intel Xeon 6745P wins across nearly every measurable category in the database. Its largest win is in PassMark physics, where it scores 6144 versus 1905, a 69% advantage. That result suggests the Intel architecture handles complex physical simulations far more efficiently. Prime number finding shows a 50.1% lead (681 vs 340), indicating stronger integer-heavy algorithmic work per clock. Multithreaded performance is 21.8% higher (84210 vs 65811), which is the aggregate measure most buyers would care about for heavily parallel server loads. Single-thread performance is 20.8% higher (3450 vs 2732), which means the Intel chip also feels snappier on lightly threaded tasks.

The Intel part also leads in floating-point math by 14.7% (267438 vs 228123), data compression by 11.1% (1352801 vs 1203096), random string sorting by 12.7% (133528 vs 116608), encryption by 5.3% (66665 vs 63159), and extended instructions by 2.4% (108326 vs 105706). The only test where the AMD EPYC 9335 comes out on top is PassMark integer math, scoring 346291 versus 336926, a 2.8% edge. That is a narrow victory, and it does not offset the Intel part's dominance elsewhere.

FAQ

Q: Which processor is faster for single-threaded workloads?

A: The Intel Xeon 6745P has a 20.8% higher PassMark single-thread score (3450 vs 2732), making it the stronger choice for lightly threaded tasks.

Q: How do the two compare in multithreaded performance?

A: The Intel Xeon 6745P leads by 21.8% in PassMark multithread (84210 vs 65811), despite both having 32 cores and 64 threads.

Q: Which chip has more cache?

A: The Intel Xeon 6745P has 336 MB of shared L3 cache, while the AMD EPYC 9335 has 128 MB. The Intel chip also has larger per-core L1 (112 KB vs 80 KB) and L2 (2 MB vs 1 MB) caches.

Q: What is the memory bandwidth difference?

A: The AMD EPYC 9335 supports twelve-channel DDR5 with 576.0 GB/s of theoretical bandwidth, while the Intel Xeon 6745P supports eight-channel DDR5 with 409.6 GB/s. The AMD part offers 40.6% more bandwidth on paper.

Q: Which processor has more PCIe lanes?

A: The AMD EPYC 9335 has 128 Gen 5 lanes, while the Intel Xeon 6745P has 88 Gen 5 lanes. This matters for systems with many GPUs or NVMe drives.

Q: Is there any workload where the AMD EPYC 9335 wins?

A: Yes, the AMD part wins PassMark integer math by 2.8% (346291 vs 336926). It also has a higher boost clock (4.40 GHz vs 4.30 GHz) and a lower TDP (210 W vs 300 W).

Head-to-Head Benchmarks

The PassMark physics test shows the most dramatic gap. The Intel Xeon 6745P scores 6144, while the AMD EPYC 9335 scores 1905. That is a 69% difference, the largest margin in the entire comparison. For any workload that relies on physics simulation, collision detection, or rigid-body dynamics, the Intel part is in a different class.

Prime number finding also heavily favors Intel. The Xeon 6745P scores 681 versus 340, a 50.1% lead. This test is often used as a proxy for general integer arithmetic and algorithm efficiency, so the result suggests the Intel core design handles tight loops better than the AMD Zen 5 core.

Multithreaded throughput follows the same pattern. The Intel part scores 84210, which is 21.8% higher than the AMD's 65811. Since both chips have identical core and thread counts, this delta comes purely from architectural efficiency, clock rates, and cache behavior. The Intel chip's larger L2 and L3 caches likely play a role here.

Single-thread performance is also a clear Intel win. The Xeon 6745P scores 3450 versus 2732, a 20.8% advantage. This is notable because the AMD chip has a higher boost clock (4.40 GHz vs 4.30 GHz), yet the Intel part still wins by a wide margin. The Intel core architecture appears to extract more instructions per clock despite the lower peak frequency.

Floating-point math favors Intel by 14.7% (267438 vs 228123), and data compression favors Intel by 11.1% (1352801 vs 1203096). Random string sorting shows a 12.7% Intel lead (133528 vs 116608). Encryption is closer, with Intel ahead by 5.3% (66665 vs 63159). Extended instructions show the smallest Intel advantage at 2.4% (108326 vs 105706).

The single AMD victory comes in integer math. The EPYC 9335 scores 346291 versus the Xeon's 336926, a 2.8% margin. This is a meaningful result for workloads like database indexing, hash computation, or financial modeling that rely heavily on integer operations. However, the margin is small enough that it would not justify choosing the AMD part on this basis alone.

Specification Differences

The two processors differ in several key specification fields. The AMD EPYC 9335 is built on a 4 nm TSMC process, while the Intel Xeon 6745P uses a 5 nm Intel process. The AMD chip has a base clock of 3.00 GHz and a boost clock of 4.40 GHz; the Intel chip has a base clock of 3.10 GHz and a boost clock of 4.30 GHz. TDP is a major difference: the AMD part is rated at 210 W, while the Intel part is rated at 300 W.

Socket types are incompatible: the AMD chip uses AMD Socket SP5, the Intel chip uses Intel Socket 4710. Memory channels differ, with the AMD part supporting twelve channels and the Intel part supporting eight. Theoretical memory bandwidth is 576.0 GB/s for AMD versus 409.6 GB/s for Intel. PCIe lane counts are 128 Gen 5 lanes for AMD versus 88 Gen 5 lanes for Intel.

Cache sizes are substantially different. The AMD EPYC 9335 has 80 KB of L1 and 1 MB of L2 per core, with 128 MB of shared L3. The Intel Xeon 6745P has 112 KB of L1 and 2 MB of L2 per core, with 336 MB of shared L3. The transistor count is only listed for the AMD part at 33,260 million; the Intel die is listed as 2x 598 mm². Both processors support DDR5 memory with ECC, and neither includes integrated graphics. The AMD part was released on October 9, 2024, while the Intel part was released on February 23, 2025.