AMD EPYC 9754 vs Intel Xeon 696X Comparison

Where Each One Wins

The AMD EPYC 9754 and Intel Xeon 696X split their head-to-head benchmark wins exactly evenly, with seven victories apiece, but the nature of those wins tells a clear story about workload suitability. The AMD EPYC 9754 dominates in data-centric and computational throughput tasks, while the Intel Xeon 696X takes every rendering and multithreaded synthetic test.

For data processing and encryption, the AMD EPYC 9754 is the clear leader. It delivers a 57.1% advantage in PassMark data compression, scoring 3,558,043 against Intel's 2,264,907. The encryption gap is even more pronounced: 231,891 versus 112,529, a 106.1% delta. This makes the EPYC 9754 the obvious choice for database workloads, secure communications, and any environment where data transformation and protection are the primary bottleneck.

Integer math is another decisive AMD win. The EPYC 9754 posts 1,026,896 in PassMark integer math, a 79.5% lead over the Xeon 696X's 572,072. Floating point math also favors AMD, with 588,187 versus 450,164, a 30.7% advantage. For scientific computing, financial modeling, and engineering simulations, the AMD part's raw arithmetic throughput is substantially higher. The random string sorting test shows a 69.9% lead for AMD (306,481 versus 180,392), reinforcing its strength in sorting and algorithmic workloads. The physics test is the most lopsided result: the EPYC 9754 scores 8,793 versus 3,382, a 160% advantage, indicating superior performance in physics-based simulation code.

On the Intel side, the Xeon 696X wins every Cinebench multicore test by the same 5.9% margin. In Cinebench R15 multicore, Intel scores 8,994 against AMD's 8,460; in R20 it is 37,475 versus 35,254; and in R23 it is 89,227 versus 83,939. These results show Intel's higher per-core clock performance translates into better raw rendering throughput when the workload scales across all threads. The PassMark multithread score also goes to Intel (104,974 versus 98,752, a 5.9% lead), confirming this trend.

Single-thread performance is decisively Intel's territory. The Xeon 696X scores 3,742 in PassMark single-thread, while the EPYC 9754 manages only 2,328, a 37.8% gap. This gives Intel a substantial advantage in latency-sensitive tasks, legacy software with poor thread scaling, and applications where a single core's responsiveness determines overall performance. The prime number finding test also favors Intel (853 versus 604, a 29.2% lead), suggesting an edge in integer-heavy sequential logic.

The Verdict

The data points to two different buyers for these processors. The AMD EPYC 9754 is the better choice for throughput-oriented data center workloads that stress memory bandwidth, encryption, data movement, and parallel arithmetic. Its 128 cores and 256 threads provide the raw parallelism needed for cloud-native services, big data analytics, data compression pipelines, and scientific computing, and the 100th percentile rank versus all CPUs confirms its top-tier position in the database's aggregate metric.

The Intel Xeon 696X, with its 64 cores and 128 threads, is the better pick for workloads that require maximum single-thread performance and where higher clock speeds (base 2.40 GHz, boost 4.80 GHz) matter more than core count. Its 5.9% across-the-board Cinebench lead and 37.8% single-thread advantage make it a stronger candidate for high-frequency trading, database query processing that runs single-threaded, and large-scale rendering where per-thread speed is critical. The Xeon 696X also has a 99th percentile rank, only one point below the AMD part.

For a user who must choose based on measured performance, the decision is straightforward: if the workload is dominated by compression, encryption, or integer math at high thread counts, the AMD EPYC 9754 wins by margins from 30% to 160%. If the workload is dominated by rendering (Cinebench) or requires fast single-thread response, the Intel Xeon 696X wins by 5.9% and 37.8%, respectively. The Intel part also has a lower launch MSRP ($5,599 versus $11,900), but the database's average benchmark score for the AMD part is higher (364,371 versus 286,102), indicating the EPYC 9754 delivers more aggregate performance per dollar, despite being more expensive.

Head-to-Head Benchmarks

The biggest win for the AMD EPYC 9754 is in PassMark physics, where it scores 8,793 against the Intel Xeon 696X's 3,382, an 160% advantage. This is the largest single delta in any test and indicates the EPYC's 128 cores provide a massive throughput advantage in physics simulation. In encryption, the EPYC 9754 doubles Intel's score, 231,891 versus 112,529, a 106.1% lead. For data centers handling encrypted traffic or secure storage, this is a decisive metric.

Data compression is another AMD blowout: 3,558,043 versus 2,264,907, a 57.1% margin. Integer math follows with 1,026,896 versus 572,072, a 79.5% lead. Random string sorting shows a 69.9% gap (306,481 versus 180,392), and floating point math a 30.7% lead (588,187 versus 450,164). Extended instructions also favor AMD, 224,322 versus 172,975, a 29.7% margin.

The Intel Xeon 696X wins most convincingly in single-thread performance, posting 3,742 versus 2,328, a 37.8% delta. This is a clear statement about clock speed advantage (4.80 GHz boost versus 3.10 GHz). In Cinebench multicore tests, Intel wins each by 5.9%: R15 scores 8,994 versus 8,460, R20 37,475 versus 35,254, and R23 89,227 versus 83,939. The PassMark multithread test also goes to Intel, 104,974 versus 98,752, a 5.9% margin. The find-prime-numbers test is Intel's only other win, 853 versus 604, a 29.2% gap.

The pattern is consistent: Intel wins where per-thread speed dictates, AMD wins where aggregate throughput is king. The 5.9% Intel delta in all three Cinebench tests is almost certainly a clock speed effect, while AMD's 57% to 160% leads in data-heavy tests reflect the doubling of core count (128 versus 64) and the larger shared L3 cache (256 MB versus 336 MB? The data shows AMD has 256 MB, Intel has 336 MB, but AMD's wins are in data manipulation, not cache capacity).

FAQ

Q: Which processor has more cores?

A: The AMD EPYC 9754 has 128 cores and 256 threads, with a base clock of 2.25 GHz and boost of 3.10 GHz. The Intel Xeon 696X has 64 cores and 128 threads, with a base clock of 2.40 GHz and boost of 4.80 GHz.

Q: How does the single-thread performance compare?

A: The Intel Xeon 696X leads in PassMark single-thread with a score of 3,742, versus the AMD EPYC 9754's 2,328, a 37.8% advantage. The Intel also has a higher boost clock (4.80 GHz versus 3.10 GHz).

Q: Which is better for data compression and encryption?

A: The AMD EPYC 9754 wins decisively. It scores 3,558,043 in data compression versus Intel's 2,264,907 (57.1% lead), and 231,891 in data encryption versus Intel's 112,529 (106.1% lead).

Q: What are the cache configurations?

A: The AMD EPYC 9754 has 64 KB of L1 and 1 MB of L2 per core, plus 256 MB of shared L3. The Intel Xeon 696X has 112 KB of L1 and 2 MB of L2 per core, plus 336 MB of shared L3.

Q: What is the memory bandwidth difference?

A: The AMD EPYC 9754 supports twelve-channel DDR5 memory with a peak bandwidth of 460.8 GB/s. The Intel Xeon 696X supports eight-channel DDR5, with a peak bandwidth of 409.6 GB/s.

Q: Which CPU has a higher aggregate benchmark score?

A: The AMD EPYC 9754 has an average benchmark score of 364,371, whereas the Intel Xeon 696X has an average of 286,102, based on the database's measurement.

Architecture Differences

The AMD EPYC 9754 is built on the Zen 4c architecture, codenamed Bergamo, using a 5 nm process from TSMC. It integrates 71,000 million transistors across a die area of 8x 73 mm², with 128 cores and 256 threads. The Intel Xeon 696X is based on the Granite Rapids architecture, using a 5 nm process from Intel, with a die area of 2x 598 mm², and packs 64 cores and 128 threads. The AMD part has a 360W TDP, the Intel a 350W TDP.

Both processors have 128 PCIe Gen 5 lanes, but the memory support differs: AMD offers a twelve-channel DDR5 bus with 460.8 GB/s bandwidth, while Intel offers an eight-channel bus with 409.6 GB/s. Both support ECC memory, a critical feature for server reliability.

The cache hierarchy is different in size and design. The AMD EPYC 9754 has 64 KB of L1 and 1 MB of L2 per core, plus 256 MB of shared L3, for a total cache of 256 MB. The Intel Xeon 696X has 112 KB of L1 and 2 MB of L2 per core, plus 336 MB of shared L3, for a total of 336 MB. Intel's larger L3 cache is notable, but the AMD's advantage in data compression and encryption suggests the higher bandwidth and core count matter more.

The socket is AMD Socket SP5 for the EPYC, and Intel Socket 4710 for the Xeon. The Intel Xeon 696X has an unlocked multiplier, allowing overclocking, while the AMD EPYC 9754 is locked. The market segment is server/workstation for both. The AMD EPYC 9754 was released in June 2023, while the Intel Xeon 696X has a future release date of February 2026. The Intel part has a part number SRWQ7, the AMD part 100-000001234. The Intel has an integrated graphics "N/A, while the AMD has none listed. The AMD's launch MSRP is $11,900, while the Intel's is $5,599.