Intel Core Ultra 5 235HX vs Intel Xeon 6505P Comparison

Where Each One Wins

The benchmark data splits these two processors into surprisingly distinct roles. The Intel Xeon 6505P, a server/workstation part built on Granite Rapids, claims five wins across the 17 head-to-head tests. The Intel Core Ultra 5 235HX, a mobile chip on Arrow Lake-HX, takes the remaining twelve. That is not a clean sweep for either side; it is a functional division of labor.

The Xeon 6505P dominates in data compression (480,368 vs. 426,417, a 12.7% lead), integer math (120,456 vs. 97,177, a 24% lead), extended instructions (37,515 vs. 34,808, a 7.8% lead), physics (2,991 vs. 2,550, a 17.3% lead), and random string sorting (52,372 vs. 50,929, a 2.8% lead). These are classic server workload patterns: heavy integer throughput, compression algorithms, and instruction-set extensions that reward wide, multi-threaded execution with large caches. The Xeon's 48 MB shared L3 cache and 24 threads clearly benefit these tasks.

The Core Ultra 5 235HX wins everywhere else, including every Cinebench test. In Cinebench R23 multi-core, it scores 34,731 versus 32,687, a 5.9% edge. In single-core R23, it scores 4,903 versus 4,614, also 5.9% ahead. The mobile chip also wins PassMark's single-thread test by a massive 31.9% (4,683 vs. 3,187), floating-point math by 28.4% (129,819 vs. 92,992), data encryption by 25.2% (32,697 vs. 24,458), and find prime numbers by 39.9% (361 vs. 217). These results point to a processor with higher clock speeds and better per-core efficiency, not raw core count.

The pattern is clear: the Xeon wins where data volume and integer-heavy parallelism matter; the Ultra 5 wins where clock speed, single-thread agility, and floating-point throughput dominate. The Ultra 5's 5.10 GHz boost clock versus the Xeon's 4.10 GHz boost clock explains much of this, but the Xeon's 24 threads versus the Ultra 5's 14 threads (with no hyperthreading) tells the other side of the story.

The Verdict

Choose the Intel Xeon 6505P if the workload is compression, integer math, physics simulation, or extended instruction sets. The data shows a 24% lead in integer math and a 17.3% lead in physics, both substantial margins. The Xeon also holds a 12.7% advantage in data compression, which is critical for database, archival, and analytics tasks. Its 150 W TDP and eight-channel memory bus (409.6 GB/s bandwidth) are built for sustained, memory-hungry server processes, not bursty consumer tasks.

Choose the Intel Core Ultra 5 235HX for almost everything else, particularly if single-thread performance or floating-point work matters. A 31.9% lead in PassMark single-thread and a 28.4% lead in floating-point math are decisive. The Ultra 5 also wins all six Cinebench tests, including both multi-core and single-core variants, meaning it is the better all-rounder for rendering, encoding, and general productivity. Its 55 W TDP and 3 nm TSMC process make it far more power-efficient, and its unlocked multiplier allows tuning, though the database does not quantify overclocking gains.

For a user who needs one chip for mixed workloads, the Ultra 5 is the safer pick due to its broader win count (12 vs. 5). For a dedicated server role where integer throughput and data compression are paramount, the Xeon is the clear specialist. The average benchmark scores tell a similar story: the Xeon averages 53,701, the Ultra 5 averages 52,073, a difference of only 1.6%. Both sit at the 91st percentile among all CPUs, so neither is a weak performer.

Head-to-Head Benchmarks

The most striking result is the PassMark single-thread test. The Ultra 5 scores 4,683, the Xeon scores 3,187, a 31.9% gap. That is the largest delta in either direction. It reflects the Ultra 5's 5.10 GHz boost clock versus the Xeon's 4.10 GHz, and it shows how far mobile architectures have come in per-core speed.

The second-largest win is also for the Ultra 5: find prime numbers, 361 vs. 217, a 39.9% margin. This is a heavily single-threaded, integer-iteration test, and the Xeon's lower clock speed hurts it badly. Floating-point math follows at 28.4% (129,819 vs. 92,992), and data encryption at 25.2% (32,697 vs. 24,458). Encryption often benefits from dedicated instructions and high clock speeds, which the Ultra 5 clearly has.

The Xeon's biggest win is integer math at 24% (120,456 vs. 97,177). This is unusual because the Ultra 5 has higher clocks, but the Xeon has 24 threads versus 14, and its 2 MB per-core L2 cache plus 48 MB shared L3 likely helps sustain integer-heavy parallel workloads. Physics shows a 17.3% Xeon lead (2,991 vs. 2,550), and data compression a 12.7% lead (480,368 vs. 426,417). Extended instructions are closer, 7.8% in favor of the Xeon (37,515 vs. 34,808), and random string sorting is nearly even at 2.8% (52,372 vs. 50,929).

All six Cinebench tests go to the Ultra 5 with a consistent 5.9% margin in both multi-core and single-core. This consistency suggests a fundamental clock-per-core advantage, not a workload-specific quirk. The PassMark multithread test also favors the Ultra 5, but only by 5.9% (40,849 vs. 38,456), which is surprising given the Xeon's extra threads. The Xeon's higher thread count cannot overcome the Ultra 5's faster cores in this mixed workload.

FAQ

Q: Which processor wins more benchmarks overall?

A: The Intel Core Ultra 5 235HX wins 12 of the 17 head-to-head tests. The Intel Xeon 6505P wins 5.

Q: Is the Xeon 6505P better at any major workload?

A: Yes. It leads by 24% in PassMark integer math, 17.3% in physics, 12.7% in data compression, 7.8% in extended instructions, and 2.8% in random string sorting.

Q: How large is the single-thread performance gap?

A: The Core Ultra 5 235HX scores 4,683 in PassMark single-thread versus 3,187 for the Xeon, a 31.9% advantage. The Ultra 5 also leads in Cinebench R23 single-core by 5.9% (4,903 vs. 4,614).

Q: Do the two chips have the same memory bandwidth?

A: No. The Xeon 6505P uses an eight-channel memory bus with 409.6 GB/s bandwidth. The Core Ultra 5 235HX uses a dual-channel bus with 102.4 GB/s bandwidth. Both support DDR5.

Q: Which chip has more threads?

A: The Xeon 6505P has 12 cores and 24 threads. The Core Ultra 5 235HX has 14 cores and 14 threads, meaning no hyperthreading on the Ultra 5.

Q: Are both CPUs in the same performance percentile?

A: Yes. Both are listed at the 91st percentile among all CPUs. The Xeon has a slightly higher average benchmark score at 53,701 versus 52,073 for the Ultra 5.

Architecture Differences

The two chips come from different Intel foundry and design philosophies. The Xeon 6505P uses the Granite Rapids architecture, built on Intel's 5 nm process, and targets the server/workstation segment. The Core Ultra 5 235HX uses the Arrow Lake architecture, built on TSMC's 3 nm process, and targets mobile. The process node difference is significant: 5 nm versus 3 nm, which explains the Ultra 5's higher clock speeds at lower power.

Core counts differ in an interesting way. The Xeon has 12 cores and 24 threads, while the Ultra 5 has 14 cores but only 14 threads. This means the Xeon relies on hyperthreading to fill its execution pipeline, while the Ultra 5 runs one thread per core. The Xeon's base clock is 2.20 GHz, its boost is 4.10 GHz. The Ultra 5 starts at 2.90 GHz and boosts to 5.10 GHz. That 1.00 GHz boost advantage is the likely driver of the Ultra 5's single-thread wins.

Cache hierarchies are also distinct. The Xeon has 112 KB of L1 per core, 2 MB of L2 per core, and 48 MB of shared L3. The Ultra 5 has 192 KB of L1 per core, 3 MB of L2 per core, and 24 MB of shared L3. The Xeon's larger L3 is a clear asset for the compression and integer workloads it wins. The Ultra 5's larger per-core L1 and L2 suggest a design focused on fast single-thread access.

Memory support diverges sharply. The Xeon uses an eight-channel memory bus with 409.6 GB/s bandwidth and supports ECC memory. The Ultra 5 uses a dual-channel bus with 102.4 GB/s bandwidth and does not support ECC. PCIe lanes also differ: the Xeon has 88 Gen 5 lanes, the Ultra 5 has 20 Gen 5 lanes. The Xeon has no integrated graphics; the Ultra 5 includes Arc Xe-LPG Graphics with 48 execution units.

The Xeon is a socketed part (Intel Socket 4710), while the Ultra 5 is a BGA 2114 soldered mobile chip. The Ultra 5 has an unlocked multiplier; the Xeon does not. The Xeon's launch MSRP is $563; the Ultra 5 has no recorded launch MSRP in the database.

Specification Differences

The following fields differ between the two processors. The Xeon 6505P has 12 cores and 24 threads; the Ultra 5 235HX has 14 cores and 14 threads. Base clocks are 2.20 GHz versus 2.90 GHz, and boost clocks are 4.10 GHz versus 5.10 GHz. TDP is 150 W for the Xeon versus 55 W for the Ultra 5. Sockets are Intel Socket 4710 versus Intel BGA 2114.

Process nodes differ: 5 nm (Intel) versus 3 nm (TSMC). The Ultra 5 has a recorded transistor count of 17,800 million and a die size of 243 mm²; the Xeon has no recorded transistor or die size data. L1 cache is 112 KB per core versus 192 KB per core. L2 is 2 MB per core versus 3 MB per core. L3 is 48 MB shared versus 24 MB shared.

Memory bus is eight-channel versus dual-channel. Memory bandwidth is 409.6 GB/s versus 102.4 GB/s. ECC support is true for the Xeon, false for the Ultra 5. PCIe lanes are 88 versus 20, both Gen 5. Integrated graphics: the Xeon has none, the Ultra 5 has Arc Xe-LPG Graphics 48EU. Market segment is server/workstation versus mobile. Release dates are 2025-02-23 versus 2025-01-12. The launch MSRP for the Xeon is $563; the Ultra 5 has none recorded. The multiplier is locked for the Xeon, unlocked for the Ultra 5. Part numbers are SRVU7 versus SRVFL.