Intel Core Ultra 9 290HX Plus vs NVIDIA RTX Spark (MediaTek AHJ11488B) Comparison

FAQ

Q: How do the core and thread counts compare between the two processors?

A: The Intel Core Ultra 9 290HX Plus has 24 cores and 24 threads, while the NVIDIA RTX Spark (MediaTek AHJ11488B) has 20 cores and 20 threads. Both processors use a 1:1 core-to-thread ratio, meaning neither supports simultaneous multithreading.

Q: What clock speeds does each processor offer?

A: The Intel processor has a base clock of 2.70 GHz and a boost clock of 5.50 GHz. The NVIDIA RTX Spark runs at a base clock of 1.70 GHz and a boost clock of 4.00 GHz. The Intel part holds a substantial clock advantage in both metrics.

Q: Which processor has higher memory bandwidth?

A: The NVIDIA RTX Spark delivers 273.1 GB/s of memory bandwidth through a quad-channel LPDDR5X interface. The Intel Core Ultra 9 290HX Plus provides 102.4 GB/s via dual-channel DDR5. The NVIDIA part offers more than double the bandwidth.

Q: What is the thermal design power for each chip?

A: The Intel Core Ultra 9 290HX Plus is rated at 55 W TDP. The NVIDIA RTX Spark is rated at 38 W TDP, making it the lower-power part by 17 W.

Q: Does either processor include integrated graphics?

A: Yes, both include integrated graphics. The Intel part uses Arc Xe-LPG Graphics with 64 execution units, while the NVIDIA RTX Spark integrates a GB20B "Blackwell" GPU.

Q: How do the processors compare in terms of production status?

A: The Intel Core Ultra 9 290HX Plus is marked as Active production and was released on 2026-03-16. The NVIDIA RTX Spark is marked as unreleased, with a release date of 2025-12-31.

Architecture Differences

The two processors represent fundamentally different architectural approaches despite both being built on a 3 nm process at TSMC. The Intel Core Ultra 9 290HX Plus belongs to the Core Ultra Series 2 family under the Arrow Lake-HX Refresh codename. It is a 24-core, 24-thread design with a 55 W TDP and uses the Intel BGA 2114 socket. The NVIDIA RTX Spark, codenamed N1X under the Spark (GB10) generation, is a 20-core, 20-thread chip with a 38 W TDP and no socket specification listed.

Cache hierarchies differ sharply. The Intel chip allocates 192 KB of L1 cache per core, 3 MB of L2 cache per core, and 36 MB of shared L3 cache. The NVIDIA part uses 128 KB of L1 per core, 2 MB of L2 per core, and only 8 MB of shared L3 cache. The Intel processor carries a 36 MB L3 pool that is 4.5 times larger than the NVIDIA chip's shared cache, a difference that matters for workloads with large working sets.

The integrated graphics also diverge. Intel uses Arc Xe-LPG Graphics with 64 EU, a general-purpose iGPU design. The NVIDIA RTX Spark embeds a GB20B "Blackwell" GPU, which aligns with the company's RTX branding and points toward graphics and compute acceleration beyond simple display output.

Memory architecture separates the two significantly. The Intel part supports dual-channel DDR5 with 102.4 GB/s bandwidth and ECC memory. The NVIDIA chip uses quad-channel LPDDR5X with 273.1 GB/s bandwidth and no ECC support. The NVIDIA part's memory subsystem delivers substantially higher throughput, which can benefit bandwidth-sensitive workloads.

PCIe connectivity is another major split. The Intel processor provides Gen 5 with 20 CPU lanes. The NVIDIA RTX Spark lists PCIe as N/A, indicating that the chip does not expose conventional PCIe lanes in the same manner. The Intel processor also has an unlocked multiplier, while the NVIDIA part is locked.

Transistor count and die size are documented only for the Intel part: 17,800 million transistors on a 243 mm² die. The NVIDIA chip has a 208 mm² die size, but transistor count is not recorded. The Intel die is larger by 35 mm² despite carrying fewer memory channels.

Where Each One Wins

The Intel Core Ultra 9 290HX Plus wins in scenarios that demand high core counts, high clocks, and deep cache. Its 24 cores and 5.50 GHz boost clock position it for multi-threaded rendering, compilation, and simulation workloads. The 36 MB L3 cache and 3 MB per-core L2 cache reduce memory pressure in data-heavy tasks. The unlocked multiplier allows tuning for sustained performance in mobile workstation-class systems.

The NVIDIA RTX Spark wins in scenarios that favor memory bandwidth and power efficiency. Its 273.1 GB/s quad-channel LPDDR5X interface is more than double the Intel chip's bandwidth, which benefits workloads that stream large datasets. The 38 W TDP makes it a lower-power option for compact mobile designs. The integrated GB20B "Blackwell" graphics suggests strong GPU-side compute capability, though benchmark data for the NVIDIA part is not recorded in the database.

The Intel part has the advantage in raw CPU throughput potential based on core count, clock speed, and cache size. The NVIDIA part counters with a faster memory subsystem and lower power envelope.

Specification Differences

| Specification | Intel Core Ultra 9 290HX Plus | NVIDIA RTX Spark (MediaTek AHJ11488B) |

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

| Cores | 24 | 20 |

| Threads | 24 | 20 |

| Base clock | 2.70 GHz | 1.70 GHz |

| Boost clock | 5.50 GHz | 4.00 GHz |

| TDP | 55 W | 38 W |

| Codename | Arrow Lake-HX Refresh | N1X |

| Generation | Ultra 9 (Arrow Lake-HX) | Spark (GB10) |

| Die size | 243 mm² | 208 mm² |

| L1 cache | 192 KB (per core) | 128 KB (per core) |

| L2 cache | 3 MB (per core) | 2 MB (per core) |

| L3 cache | 36 MB (shared) | 8 MB (shared) |

| Memory support | DDR5 | LPDDR5X |

| Memory bus | Dual-channel | Quad-channel |

| Memory bandwidth | 102.4 GB/s | 273.1 GB/s |

| ECC memory | Yes | No |

| PCIe | Gen 5, 20 Lanes (CPU only) | N/A |

| Integrated graphics | Arc Xe-LPG Graphics 64EU | GB20B "Blackwell" |

| Socket | Intel BGA 2114 | Not listed |

| Multiplier unlocked | Yes | No |

| Production status | Active | Unreleased |

Head-to-Head Benchmarks

The database contains no head-to-head benchmark entries between the Intel Core Ultra 9 290HX Plus and the NVIDIA RTX Spark. The Intel processor has a full benchmark suite recorded, while the NVIDIA RTX Spark has no benchmark scores, no average score, and no nearest rivals listed.

The Intel Core Ultra 9 290HX Plus posts an average benchmark score of 79,574 and sits at the 95th percentile among all CPUs. Its nearest rivals in the database are the Intel Core i9-14900KF at 79,371 average score (0.3% ahead), the AMD EPYC 7413 at 80,041 average score (0.6% behind), the Intel Core i9-14900K at 79,097 average score (0.6% ahead), and the Intel Xeon w5-2565X at 80,671 average score (1.4% behind). These deltas place the 290HX Plus in a tight cluster with high-end desktop and server parts.

In Cinebench tests, the Intel chip scores 39,684 in R23 multi-core and 2,356 in R23 single-core. The R20 results are 21,198 multi-core and 2,992 single-core. The R15 results are 5,981 multi-core and 340 single-core. These numbers indicate strong scaling from single-core to multi-core, with the 24-core design multiplying single-thread performance effectively.

PassMark results for the Intel part show 594,39 in multi-thread and 4,951 in single-thread. The data compression score is 658,724, data encryption is 50,008, and extended instructions score 51,290. Floating point math reaches 201,773, integer math 164,839, and physics 3,387. Random string sorting scores 80,327, and find prime numbers scores 519.

Because the NVIDIA RTX Spark has no recorded benchmarks, no direct comparison of measured performance is possible from the database. The available data allows only a specification-level comparison. The Intel part is the only one of the two with verified performance results, and those results place it at the 95th percentile of all CPUs in the database. The NVIDIA part sits at the 50th percentile by default, as its average benchmark score is 0.

The specification gap in cache and clocks is large. The Intel chip's L3 cache is 36 MB versus 8 MB, its boost clock is 5.50 GHz versus 4.00 GHz, and its core count is 24 versus 20. The NVIDIA part's memory bandwidth advantage, 273.1 GB/s versus 102.4 GB/s, is its most notable lead. The power envelope also favors the NVIDIA part at 38 W versus 55 W. Without benchmark results for the NVIDIA RTX Spark, the database cannot confirm how these architectural differences translate into application performance.