AMD EPYC 9745 vs Intel Xeon 696X Comparison

The AMD EPYC 9745 and Intel Xeon 696X are two high-end server processors aimed at different ends of the workload spectrum. The EPYC packs 128 cores and 256 threads, while the Xeon offers 64 cores and 128 threads. Benchmark results show a clear split: AMD dominates nearly every multi-threaded test, while Intel takes single-thread performance. This analysis breaks down where each processor wins, what separates their architectures, and how they compare in head-to-head benchmarks.

Where Each One Wins

The recorded benchmark data shows a decisive split. The AMD EPYC 9745 wins 12 of the 14 head-to-head tests, while the Intel Xeon 696X wins only the two single-thread tests (Passmark single_thread and singlethread). This makes the EPYC the clear choice for heavily parallel workloads such as rendering, data compression, encryption, and scientific simulation. The Xeon, by contrast, is better suited for lightly threaded applications where single-core responsiveness matters more than raw core count.

In multi-threaded Cinebench tests, the EPYC leads by 24.5% in R15, R20, and R23 multicore scores. Its advantage grows in specific Passmark subtests: 73.5% in data compression, 103.9% in data encryption, 114% in integer math, and a staggering 406.3% in physics. These are not marginal gains; they represent fundamental throughput differences. The EPYC also wins in floating-point math (69.1%), extended instructions (62.1%), random string sorting (160%), and find prime numbers (14.8%). The only area where the Xeon pulls ahead is single-thread performance, where it scores 3742 against the EPYC's 2806, a 25% deficit for AMD.

For a server administrator choosing between the two, the decision hinges on the workload. If the application scales across many cores, the EPYC 9745 is overwhelmingly faster. If the workload is single-threaded or latency-sensitive, the Xeon 696X offers a significant advantage.

Architecture Differences

The two processors are built on fundamentally different architectures. The EPYC 9745 uses AMD's Zen 5 core, code-named Turin, fabricated on a 3 nm process at TSMC. The Xeon 696X uses Intel's Granite Rapids core, also code-named Granite Rapids, on Intel's 5 nm process. The process node difference is notable: 3 nm allows AMD to pack more transistors into a smaller area, though the die size for the EPYC is not listed in the database, while the Xeon lists a die size of 2x 598 mm².

Core counts diverge sharply. The EPYC has 128 cores and 256 threads, while the Xeon has 64 cores and 128 threads. This is the primary driver of the EPYC's multi-threaded dominance. Cache configurations also differ. The EPYC has 80 KB of L1 per core, 1 MB of L2 per core, and 256 MB of shared L3. The Xeon has 112 KB of L1 per core, 2 MB of L2 per core, and 336 MB of shared L3. The Xeon's larger per-core caches and total L3 help its single-thread performance, but the EPYC's massive core count compensates in throughput.

Memory support is another differentiator. The EPYC uses a twelve-channel DDR5 interface with a bandwidth of 576.0 GB/s, while the Xeon uses an eight-channel interface with 409.6 GB/s. Both support ECC memory and PCIe Gen 5 with 128 lanes. The EPYC's higher memory bandwidth is essential for feeding its 128 cores, while the Xeon's lower bandwidth is adequate for its 64-core design.

Other architectural differences include the socket (AMD Socket SP5 for the EPYC, Intel Socket 4710 for the Xeon), the base clock (both 2.40 GHz), and the boost clock (3.70 GHz for the EPYC, 4.80 GHz for the Xeon). The Xeon has an unlocked multiplier, while the EPYC does not. The Xeon also has a later release date: February 2026 versus October 2024 for the EPYC.

Head-to-Head Benchmarks

The head-to-head results paint a clear picture. In Cinebench R23 multicore, the EPYC scores 111093 against the Xeon's 89227, a 24.5% lead. The same delta appears in Cinebench R15 and R20 multicore, with scores of 11198 vs 8994 and 46659 vs 37475 respectively. These are consistent, repeatable margins across all three Cinebench versions.

The Passmark suite shows even larger gaps. In integer math, the EPYC scores 1224315 versus 572072, a 114% advantage. Floating-point math sees 761219 against 450164, a 69.1% lead. Data encryption is particularly lopsided: 229447 versus 112529, a 103.9% margin. Physics performance is the most extreme: 17122 versus 3382, a 406.3% difference. Random string sorting also shows a massive gap: 468975 versus 180392, a 160% lead. Data compression follows at 73.5% (3929890 vs 2264907), and extended instructions at 62.1% (280477 vs 172975). Even find prime numbers, a test that often favors high clock speeds, goes to the EPYC by 14.8% (979 vs 853).

The only tests the Xeon wins are the Passmark single-thread and singlethread tests, both scoring 3742 against the EPYC's 2806. That is a 25% deficit for AMD, confirming that the Xeon's higher boost clock (4.80 GHz vs 3.70 GHz) and larger per-core caches give it a meaningful single-core edge.

Specification Differences

The two processors differ in several key specifications. The most obvious is core and thread count: 128 cores / 256 threads for the EPYC, 64 cores / 128 threads for the Xeon. Boost clocks differ as well: 3.70 GHz for AMD, 4.80 GHz for Intel. TDP ratings are 400 W for the EPYC and 350 W for the Xeon. Sockets are incompatible: SP5 for AMD, 4710 for Intel.

Architecture and process node are distinct: Zen 5 (Turin) on 3 nm TSMC for AMD, Granite Rapids on 5 nm Intel for Intel. Cache sizes vary: L1 is 80 KB per core for AMD, 112 KB per core for Intel; L2 is 1 MB per core for AMD, 2 MB per core for Intel; L3 is 256 MB shared for AMD, 336 MB shared for Intel. Memory channels and bandwidth differ: twelve-channel with 576.0 GB/s for AMD, eight-channel with 409.6 GB/s for Intel. Both support DDR5 and ECC, and both have 128 PCIe Gen 5 lanes.

The Xeon has an unlocked multiplier, while the EPYC does not. The Xeon also has a listed die size of 2x 598 mm², while the EPYC's is not recorded. Release dates are October 2024 for the EPYC and February 2026 for the Xeon. Launch MSRP: the EPYC 9745 is $12141, and the Xeon 696X is $5599.

FAQ

Q: Which processor has more cores?

A: The AMD EPYC 9745 has 128 cores and 256 threads, while the Intel Xeon 696X has 64 cores and 128 threads.

Q: Which has higher single-thread performance?

A: The Intel Xeon 696X scores 3742 in Passmark single-thread, while the EPYC 9745 scores 2806, a 25% deficit for AMD.

Q: Which has more L3 cache?

A: The Intel Xeon 696X has 336 MB of shared L3, while the AMD EPYC 9745 has 256 MB of shared L3.

Q: Which supports more memory bandwidth?

A: The AMD EPYC 9745 has a twelve-channel DDR5 interface with 576.0 GB/s, while the Intel Xeon 696X has an eight-channel interface with 409.6 GB/s.

Q: Which is built on a smaller process node?

A: The AMD EPYC 9745 uses a 3 nm process at TSMC, while the Intel Xeon 696X uses a 5 nm process at Intel.

Q: Which has a higher boost clock?

A: The Intel Xeon 696X boosts to 4.80 GHz, while the AMD EPYC 9745 boosts to 3.70 GHz.