Intel Core Ultra X7 358H New Benchmarks are out on PassMark
Intel Core Ultra X7 358H is one of the upcoming Panther Lake mobile microprocessors, the results for which have surfaced on the PassMark platform. Scores from this more credible platform show that the new mid-range chip is significantly slower than Intel's currently available Arrow Lake-H CPUs.
Here are core performance details of the Core Ultra X7 358H processor:
Single-threaded score: 4,282 and multi-threaded score: 29,426. The PassMark entry further confirms the known specification of 16 cores (4 Performance, 8 Efficiency, 4 Low Power Efficient cores) and an 18 MB L3 cache.
The performance gap compared to current Arrow Lake-H variants is significant:
- Single-threaded performance costs a bit over both the Core Ultra 7 255H (4,347) and the Core Ultra 7 265H (4,433).
- In multi-threaded tests, the X7 358H lags behind the 255H by 4%.
- The multi-threaded gap widens to about 15% when compared to the Core Ultra 7 265H (34,678).
Examining the Integrated Arc B390 GPU Benchmarks
The PassMark figures also have entry performance figures for an integrated GPU attached to the chip, the Arc B390, based on the Xe3 architecture. The iGPU is rated to get 9,339 points in the G3D Mark Test.
Performance of this nature would place it:
- Close to a desktop GeForce GTX 1650 SUPER.
- Approximately 23% slower than a GeForce RTX 3050 Laptop GPU (12,144).
This result is quite a departure from earlier Geekbench leaks, which claimed that the Arc B390 could even compete with an RTX 3050 Ti laptop GPU. PassMark is considered a more trustworthy platform for making such comparisons.
Early Results and Implications Ahead
It is critical to keep in mind that these are preliminary benchmarks and represents an early engineering sample. The final performance may differ. Even with that, they do not seem particularly impressive, and a poor score would not impress if Panther Lake cannot excel over its Arrow Lake predecessors.
The dates for launch of the Panther Lake lineup into mobile platforms are set in January 2026. If the final performance does not improve significantly, it puts more pressure on the subsequent Nova Lake architecture expected from Intel in the late part of the year 2026.
