Intel Core Ultra 9 285K vs Intel Core Ultra X9 378H Comparison

Head-to-Head Benchmarks

The benchmark data records 17 head-to-head comparisons between the Intel Core Ultra 9 285K and the Intel Core Ultra X9 378H. The desktop part claims 15 wins, while the mobile part takes 2. The margins tell a more nuanced story than the raw win count.

The 285K dominates in multi-threaded workloads. In Cinebench R15 multi-core, it scores 6494 against 3281, a 97.9% advantage. Cinebench R20 multi-core shows 24003 versus 13672, a 75.6% lead. The R23 multi-core test narrows the gap to 30.6%, with the 285K scoring 42522 and the 378H scoring 32553. That closing margin suggests the mobile chip scales better as the workload grows, though it never overtakes.

The 285K also wins decisively across the PassMark suite. Random string sorting shows the largest delta at 112.6%, with scores of 94927 and 44648. Data compression follows at 104.4% (790052 versus 386591). Extended instructions deliver a 98.9% edge (62277 against 31315), and floating point math lands at 95.9% (224324 versus 114500). Data encryption shows a 93.5% gap (57745 versus 29840), while integer math sits at 86.1% (172379 versus 92603). Find prime numbers gives the 285K a 51.5% advantage (541 versus 357).

The PassMark multi-thread score favors the 285K by 75.6%, at 67260 versus 38298. Physics also goes to the desktop chip, though by a smaller 15.7% margin (3938 versus 3404). Single-thread PassMark results show a 14.2% lead for the 285K, scoring 5087 against 4453.

The 378H wins two single-core Cinebench tests, and these are not narrow margins. In Cinebench R23 single-core, the 378H scores 4595 against 2377, a 48.3% advantage over the 285K. In R15 single-core, it scores 462 versus 359, a 22.3% lead. However, Cinebench R20 single-core tells a different story: the 285K wins at 3388 versus 1929, a 75.6% gap. This inconsistency across Cinebench versions is curious. The R15 and R23 results suggest the 378H has a meaningful single-thread advantage, yet the R20 result contradicts that pattern entirely. The data does not explain why, but the discrepancy is worth questioning.

The average benchmark score reinforces the overall picture. The 285K averages 83807 across all recorded tests, placing it in the 96th percentile of all CPUs. The 378H averages 47468, placing it in the 89th percentile. The nearest rival data shows the 285K sitting within 1.5% of several server-class EPYC parts, including the AMD EPYC 7F72 at 85072 (a 1.5% deficit) and the AMD EPYC 4584PX at 83090 (a 0.9% edge). The 378H, by contrast, trades blows with desktop parts like the Intel Core i7-13700KF (47330, 0.3% behind) and the Intel Core i9-12900F (47176, 0.6% ahead).

Architecture Differences

The two processors come from different Intel families. The 285K belongs to Core Ultra Series 2, uses the Arrow Lake architecture with the Arrow Lake-S codename, and is built for the desktop socket Intel Socket 1851. The 378H belongs to Core Ultra Series 3, uses the Panther Lake codename (listed architecture is null in the database), and is a mobile part on Intel BGA 2540.

Both are manufactured on a 3 nm process, but they use different foundries. The 285K is produced by TSMC, while the 378H is produced by Intel. The 285K carries 17,800 million transistors on a 243 mm² die. The database records no transistor count or die size for the 378H.

Core counts differ substantially. The 285K has 24 cores and 24 threads, with no hyperthreading. The 378H has 16 cores and 16 threads. Base clocks are 3.70 GHz for the desktop part and 2.00 GHz for the mobile part. Boost clocks reach 5.70 GHz and 5.00 GHz respectively. The 285K has an unlocked multiplier; the 378H is locked.

Cache layouts show both similarities and differences. Both parts use 192 KB of L1 per core. The L2 cache is 3 MB per core on the 285K and 2.5 MB per core on the 378H. Shared L3 totals 36 MB on the 285K and 18 MB on the 378H. Neither part uses 3D V-Cache.

Memory support diverges. The 285K uses DDR5 with a dual-channel bus and 102.4 GB/s bandwidth, and it supports ECC memory. The 378H uses LPDDR5X with a dual-channel bus and higher bandwidth at 153.6 GB/s, but it does not support ECC. PCIe lanes also differ: the 285K provides Gen 5 with 20 lanes from the CPU, while the 378H provides Gen 5 with only 4 lanes.

Integrated graphics differ as well. The 285K uses Arc Xe-LPG Graphics with 64 execution units. The 378H uses Arc B390. The database does not record execution unit counts for the B390.

Power envelopes are far apart. The 285K has a 125 W TDP, while the 378H has a 25 W TDP. That fivefold difference in power target likely explains much of the multi-thread performance gap, though the data does not include direct power measurements.

Release dates also differ. The 285K launched on 2024-10-23 with a launch MSRP of $589. The 378H launched on 2026-04-03 with no recorded launch MSRP. Both are currently marked as Active in production.

FAQ

Q: Which processor has more cores and threads?

A: The Intel Core Ultra 9 285K has 24 cores and 24 threads. The Intel Core Ultra X9 378H has 16 cores and 16 threads.

Q: Which chip wins in Cinebench R23 multi-core?

A: The 285K wins with a score of 42522 against 32553, a 30.6% advantage.

Q: Is the mobile chip faster in any benchmark?

A: Yes. The 378H wins Cinebench R23 single-core (4595 versus 2377, a 48.3% lead) and Cinebench R15 single-core (462 versus 359, a 22.3% lead). The 285K wins Cinebench R20 single-core (3388 versus 1929, a 75.6% lead).

Q: What is the difference in average benchmark scores?

A: The 285K averages 83807 across all recorded benchmarks, placing in the 96th percentile. The 378H averages 47468, placing in the 89th percentile.

Q: Do both chips use the same process node?

A: Both are built on a 3 nm process, but the 285K is fabricated by TSMC while the 378H is fabricated by Intel.

Q: Which chip supports ECC memory?

A: The 285K supports ECC memory with DDR5. The 378H uses LPDDR5X and does not support ECC.

The Verdict

The benchmark data separates these two processors by workload type and platform intent. The 285K is a desktop part with 24 cores, a 125 W TDP, and a 96th percentile standing. It wins 15 of 17 head-to-head tests, often by large margins. Its nearest rivals are server-class EPYC processors, which indicates the performance tier it occupies.

The 378H is a mobile part with 16 cores, a 25 W TDP, and an 89th percentile standing. It wins two single-core Cinebench tests, including a 48.3% margin in R23 single-core. Its nearest rivals include desktop parts like the Intel Core i7-13700KF and Intel Core i9-12900F, which places it in a strong mid-range desktop performance class despite being a mobile chip.

The single-core Cinebench results present a puzzle. The 378H leads R15 and R23 single-core by substantial margins, yet loses R20 single-core by 75.6%. The database does not explain this inconsistency, and it complicates any simple narrative about which chip has the better single-thread architecture. The PassMark single-thread score gives the 285K a 14.2% edge, which aligns with the R20 result but not with R15 or R23.

The 378H achieves its multi-thread scores with 16 cores and a 25 W TDP. The 285K uses 24 cores and a 125 W TDP. The performance per watt is not directly measured in the database, but the power envelope difference is so large that the mobile chip's efficiency profile is clearly different.

For users whose workloads match the recorded benchmarks, the 285K is the stronger processor in almost every category. The 378H is the better choice only in specific single-thread Cinebench scenarios, and even then the R20 result contradicts that pattern. The data does not support a general recommendation for the 378H outside of its intended mobile platform context.

Specification Differences

| Field | Intel Core Ultra 9 285K | Intel Core Ultra X9 378H |

| --- | --- | --- |

| Series | Core Ultra Series 2 | Core Ultra Series 3 |

| Cores | 24 | 16 |

| Threads | 24 | 16 |

| Base Clock | 3.70 GHz | 2.00 GHz |

| Boost Clock | 5.70 GHz | 5.00 GHz |

| TDP | 125 W | 25 W |

| Socket | Intel Socket 1851 | Intel BGA 2540 |

| Codename | Arrow Lake-S | Panther Lake |

| Generation | Ultra 9 (Arrow Lake) | Ultra X9 (Panther Lake-H) |

| Foundry | TSMC | Intel |

| Transistors | 17,800 million | Not recorded |

| Die Size | 243 mm² | Not recorded |

| L2 Cache | 3 MB (per core) | 2.5 MB (per core) |

| L3 Cache | 36 MB (shared) | 18 MB (shared) |

| Memory Support | DDR5 | LPDDR5X |

| Memory Bandwidth | 102.4 GB/s | 153.6 GB/s |

| ECC Memory | Yes | No |

| PCIe | Gen 5, 20 Lanes (CPU only) | Gen 5, 4 Lanes (CPU only) |

| Integrated Graphics | Arc Xe-LPG Graphics 64EU | Arc B390 |

| Market Segment | Desktop | Mobile |

| Release Date | 2024-10-23 | 2026-04-03 |

| Launch MSRP | $589 | Not recorded |

| Multiplier Unlocked | Yes | No |

| Part Number | SRQD5 | Unknown |

Shared specifications include the 3 nm process node, 192 KB L1 cache per core, dual-channel memory bus, and Gen 5 PCIe support. The 285K uses 24 threads, the same as its core count, and the 378H also matches threads to cores at 16. Neither part uses 3D V-Cache.

Where Each One Wins

The 285K wins in multi-core rendering. Cinebench R15, R20, and R23 multi-core all go to the desktop chip, with margins ranging from 30.6% to 97.9%. The PassMark multi-thread test also favors the 285K at 75.6%. Any workload that scales across cores will favor the 285K based on these results.

The 285K also wins in data processing tasks. Data compression shows a 104.4% lead, and random string sorting shows a 112.6% lead. Encryption, extended instructions, floating point math, and integer math all favor the 285K by margins between 86.1% and 98.9%. The physics test is closer at 15.7%, but still goes to the 285K.

The 378H wins in two specific single-core Cinebench tests. The R23 single-core result is particularly strong, with the 378H scoring 4595 against 2377. The R15 single-core result also favors the 378H at 462 versus 359. These wins suggest the mobile chip has a competitive single-thread architecture, though the R20 single-core result contradicts this by favoring the 285K at 75.6%.

The 378H also holds an advantage in memory bandwidth on paper, with 153.6 GB/s against 102.4 GB/s for the 285K. The database does not include a benchmark that isolates memory bandwidth, so the practical impact of this difference is not directly measured.

The 378H uses a 25 W TDP, which indicates a much lower power draw than the 285K's 125 W TDP. The database does not record power consumption measurements, but the TDP figures alone distinguish the two parts by platform intent. The 378H is built for mobile use on a BGA socket with 4 PCIe lanes, while the 285K is a desktop part with 20 PCIe lanes and an unlocked multiplier.

The nearest rival data adds context. The 285K competes with AMD EPYC server processors, sitting within 1.5% of the EPYC 7F72 and 1% of the EPYC 9135. The 378H competes with desktop processors like the Core i7-13700KF and Core i9-12900F, sitting within 0.6% of the latter. This places the 378H's performance in a desktop-class range despite its mobile form factor.