Intel Core Ultra 5 235A vs Intel Core Ultra X9 388H Comparison

The Verdict

The benchmark data presents a clear split between these two Intel processors. The Intel Core Ultra 5 235A (Arrow Lake-S, desktop) wins 15 of the 17 recorded head-to-head tests, while the Intel Core Ultra X9 388H (Panther Lake-H, mobile) takes only 2. The 235A posts a 72.6% advantage in Cinebench R23 multi-core (32633 vs 18911) and a 109.4% lead in R23 single-core (4607 vs 2200.5). The 388H counters with wins in PassMark integer math (90882 vs 88626, a 2.5% edge) and PassMark physics (3226 vs 2437, a 24.5% lead). The 235A also holds a 90th percentile ranking among all CPUs versus the 388H's 88th percentile, and its average benchmark score of 48201 is 8.4% higher than the 388H's 44466. For desktop workloads demanding sustained multi-threaded performance, the data points squarely to the 235A. The 388H's narrow wins in physics and integer math, combined with its 25W TDP and mobile BGA socket, indicate it is positioned for a different role, one where power envelope and platform matter more than raw Cinebench output.

FAQ

Q: Which processor is faster in multi-core rendering?

A: The Intel Core Ultra 5 235A leads decisively. In Cinebench R23 multi-core, it scores 32633 versus the 388H's 18911, a delta of 72.6%. The gap narrows in R20 (13705 vs 13101, a 4.6% edge for the 235A) and R15 (3289 vs 2955, an 11.3% edge).

Q: Which chip wins in single-core tests?

A: The 235A wins every single-core benchmark recorded. Cinebench R23 single-core shows 4607 versus 2200.5, a 109.4% advantage. R20 single-core is 1934 vs 1849 (4.6%), R15 single-core is 464 vs 309.5 (49.9%), and PassMark single-thread is 4557 vs 4280 (6.5%).

Q: Does the Core Ultra X9 388H win any tests?

A: Yes, two. PassMark integer math favors the 388H (90882 vs 88626, a 2.5% margin), and PassMark physics shows a substantial 24.5% lead for the 388H (3226 vs 2437).

Q: How do their memory bandwidth figures compare?

A: The 388H supports LPDDR5X memory with a recorded bandwidth of 153.6 GB/s, while the 235A uses DDR5 with 102.4 GB/s. The 388H's memory bandwidth is 50% higher on paper, though benchmark results in data-heavy tasks like PassMark data compression still favor the 235A (393800 vs 361763, an 8.9% edge).

Q: What are the core and thread counts?

A: The 235A has 14 cores and 14 threads. The 388H has 16 cores and 16 threads. Despite having 2 more cores and threads, the 388H trails in most multi-threaded benchmarks, including PassMark multithread (36811 vs 38392, a 4.3% deficit).

Q: Which chip has a higher boost clock?

A: The 388H boosts to 5.10 GHz, slightly ahead of the 235A's 5.00 GHz. Yet the 235A still wins the single-thread PassMark test by 6.5%, indicating architectural efficiency matters more than peak frequency in this comparison.

Architecture Differences

The two processors come from different Intel design generations and foundries. The Core Ultra 5 235A belongs to the Core Ultra Series 2, uses the Arrow Lake architecture on the Arrow Lake-S codename, and is fabricated by TSMC on a 3 nm process. It carries 17,800 million transistors on a 243 mm² die. The Core Ultra X9 388H belongs to Core Ultra Series 3, uses the Panther Lake architecture with the Panther Lake-H codename, and is also on a 3 nm node, but fabricated by Intel's own foundry. The database lists no transistor count or die size for the 388H.

Cache organization differs primarily in L3. Both chips share the same per-core L1 (192 KB) and L2 (3 MB) allocations. The 235A has 24 MB of shared L3 cache, while the 388H has 18 MB. That 6 MB difference may explain part of the 235A's lead in memory-sensitive workloads like PassMark random string sorting (49489 vs 44010, a 12.4% edge).

The integrated graphics also diverge significantly. The 235A carries Arc Xe-LPG Graphics with 24 execution units, while the 388H features an Arc B390. The 388H's graphics solution is newer, but no graphics benchmarks appear in the recorded dataset. The PCIe configurations differ as well: the 235A provides Gen 5 with 20 CPU lanes, while the 388H offers Gen 5 with only 4 CPU lanes, reflecting the desktop versus mobile platform split.

The 235A targets the desktop segment with an Intel Socket 1851, whereas the 388H uses the mobile Intel BGA 2540 socket. The 235A's production status is Active, as is the 388H's. The 235A was released on 2025-07-28, and the 388H on 2026-01-04, making the 388H the more recent part by about five months. The 235A has a launch MSRP of $269; the database lists no launch MSRP for the 388H.

Specification Differences

The recorded specification sheet shows several direct contrasts. Core count: 14 for the 235A versus 16 for the 388H. Thread count: 14 versus 16. Base clock: 3.40 GHz for the 235A versus 2.10 GHz for the 388H. Boost clock: 5.00 GHz versus 5.10 GHz. TDP: 65W for the 235A versus 25W for the 388H, a 40W gap that reflects the desktop versus mobile power envelope.

Socket: Intel Socket 1851 for the 235A, Intel BGA 2540 for the 388H. Memory support: DDR5 for the 235A, LPDDR5X for the 388H. Memory bandwidth: 102.4 GB/s versus 153.6 GB/s. PCIe: Gen 5 with 20 lanes (CPU only) for the 235A, Gen 5 with 4 lanes (CPU only) for the 388H. L3 cache: 24 MB shared versus 18 MB shared. Integrated graphics: Arc Xe-LPG 24EU versus Arc B390. Foundry: TSMC versus Intel. Part numbers: SRWPN versus SA4QWQ9EK. Generation labels: Ultra 5 (Arrow Lake) versus Ultra X9 (Panther Lake-H). Market segment: Desktop versus Mobile. Release date: 2025-07-28 versus 2026-01-04. The 235A has a launch MSRP of $269; the 388H has none listed.

Both chips share several traits: dual-channel memory bus, no ECC memory support, locked multipliers, L1 of 192 KB per core, L2 of 3 MB per core, and a 3 nm process node.

Head-to-Head Benchmarks

The Cinebench suite delivers the most lopsided results. In R23 multi-core, the 235A scores 32633 against the 388H's 18911, a 72.6% margin. R23 single-core is even starker: 4607 versus 2200.5, a 109.4% lead. R15 single-core shows a 49.9% gap (464 vs 309.5), while R15 multi-core is 11.3% (3289 vs 2955). R20 results are closer: multi-core 13705 versus 13101 (4.6%) and single-core 1934 versus 1849 (4.6%).

PassMark tests mostly follow the same pattern. Data compression favors the 235A by 8.9% (393800 vs 361763). Data encryption is 5.8% (30136 vs 28490). Extended instructions: 5.6% (31625 vs 29943). Find prime numbers: 9.5% (392 vs 358). Floating point math: 5.5% (118778 vs 112550). Multithread: 4.3% (38392 vs 36811). Random string sorting: 12.4% (49489 vs 44010). Single-thread: 6.5% (4557 vs 4280).

The 388H's two wins stand out precisely because they break the pattern. PassMark integer math shows 90882 for the 388H versus 88626 for the 235A, a 2.5% margin. PassMark physics delivers its strongest result for the 388H: 3226 versus 2437, a 24.5% advantage. That physics score is notable because it is the largest single win for the 388H in any test, and it comes in a workload that often tracks memory latency and core scheduling rather than raw throughput.

The overall win tally is 15 for the 235A and 2 for the 388H. The average benchmark scores reinforce this: 48201 for the 235A versus 44466 for the 388H. The 235A's nearest rivals include the AMD Ryzen AI Max PRO 380 (avg 48171, delta 0.1%), Intel Core Ultra 5 245HX (48287, -0.2%), AMD EPYC 4345P (48470, -0.6%), and AMD Ryzen 9 3900 (47918, 0.6%). The 388H's nearest rivals include the AMD Ryzen 5 7500X3D (44573, -0.2%), Intel Core i9-13950HX (44342, 0.3%), AMD Ryzen AI Max 385 (44309, 0.4%), and Intel Core i5-13600 (44240, 0.5%).

Where Each One Wins

The Core Ultra 5 235A wins in nearly every recorded workload category. Its Cinebench dominance across R15, R20, and R23 in both single-core and multi-core indicates strong architectural efficiency per clock and robust multi-threaded scaling under sustained load. PassMark data compression, encryption, extended instructions, prime number finding, floating point math, multithread, random string sorting, and single-thread tests all go to the 235A. The 24 MB L3 cache likely supports its wins in random string sorting and data compression, where larger working sets benefit from more cache. The 65W TDP gives it headroom to maintain boost clocks during long render or compilation runs.

The Core Ultra X9 388H wins in PassMark integer math and PassMark physics. The physics result, at 24.5% ahead, suggests the 388H handles certain latency-sensitive or instruction-mix-specific workloads more efficiently. Its 16 cores and 2.10 GHz base clock with 5.10 GHz boost may help in physics calculations that parallelize across many threads without heavy memory traffic. The 388H's 153.6 GB/s memory bandwidth, while higher than the 235A's 102.4 GB/s, does not translate into wins in memory-heavy PassMark tests, which all favor the 235A. The 25W TDP makes the 388H a low-power part, and that power envelope likely explains its placement in the mobile segment with a BGA socket.

The data suggests the 235A is the stronger general-purpose compute part, particularly for desktop users running Cinebench-style workloads, data encryption, or floating point math. The 388H is a mobile processor that trades raw multi-core output for lower power consumption and a different platform fit. Its wins in physics and integer math are narrow in one case and substantial in the other, but they do not compensate for the 72.6% multi-core deficit in R23 or the 8.4% gap in average benchmark score. The percentile rankings, 90th versus 88th, align with the overall win tally. For any workload that resembles the recorded Cinebench or PassMark tests, the 235A is the data-backed choice. For a low-power mobile part that still competes in physics and integer tasks, the 388H has a defined, if narrow, niche.