Intel Core Ultra 275HX Outshines Core i9-14900HX by 33% in Early Passmark Appearance

[GGforever]

A recent Cinebench R23 result portrayed the upcoming Intel Core Ultra 9 275HX lagging behind its predecessor by a few points in single-core performance, despite pulling well ahead in multicore performance. Now, the high-end Arrow Lake-HX chip has made its debut on Passmark, and the result appears quite enticing, to say the least. In single-core, the Core Ultra 9 275HX leads the i9 14900HX by around 10% - a fair generational uplift. In overall performance, however, the Core Ultra 9 275HX shines bright, pulling off a 33% lead over its predecessor. Of course, the actual improvements are likely to be lower, considering that the Passmark database contains over 1800 entries for Core i9 14900HX-powered systems with varying thermal capabilities, while only a single one so far for the 275HX.

For a refresher, the Core Ultra 9 275HX debuted at CES 2025, and packs 8 Lion Cove P-cores along with 16 Skymont E-cores. Intel has left Hyper-Threading in the rearview mirror with its Arrow Lake lineup, although the Passmark entry seems to suggest Arrow Lake-HX will do just fine without it. Unsurprisingly, for laptops, the performance of the system will boil down to its thermal capabilities, which basically means that there will be a plethora of systems where the 275HX will be unable to fully spread its wings. Besides that, as with all pre-release performance benchmark leaks, be sure to accept this information with a grain of salt. The Ryzen 7945HX3D is also left behind, albeit by a far smaller margin of just around 7% in overall performance. With the Ryzen 9 9955HX3D just around the corner, however, Intel's high-end laptop reign might be short-lived after all.



View at TechPowerUp Main Site | Source

 

[RandallFlagg]

Sounds right. Intel's push has gone whole hog into the large laptop market. Unfortunately, it doesn't translate very well into the desktop market.

 

[ScaLibBDP]

This is due to higher Clock Speed, that is 2.7 GHz vs. 2.2 GHz. If a normalization is applied the performance improvement will be ~8.2%.

Also, it is Not clear where the 33% came from?

 

[Rjc31]

This is due to higher Clock Speed, that is 2.7 GHz vs. 2.2 GHz. If a normalization is applied the performance improvement will be ~8.2%.

Also, it is Not clear where the 33% came from?

61010 is 133% of 45615... Or 33% more

 

[RandallFlagg]

This is due to higher Clock Speed, that is 2.7 GHz vs. 2.2 GHz. If a normalization is applied the performance improvement will be ~8.2%.

Also, it is Not clear where the 33% came from?

Your first statement is only using base clock speeds, which are meaningless during benchmarking.

To add, when you look at turbo speeds, the exact opposite of what you assume is happening here.

The reason it doesn't do better vs the 14900HX in single thread is likely due to having ~7% slower max turbo speed. That affects single thread, where the CPUs can maintain that turbo on one core.

So if you normalized in the correct direction, if both were running at the same clock, the 275HX would be ~17% faster in single thread. That is just mental masturbation though, the chips run at the frequencies they run at.

In multi-thread, they can't maintain max turbo and it's anyone's guess where they land without a benchmark. However, based on this benchmark, the 275HX is able to maintain its turbo much better than the older 14900HX. Plus it has an IPC advantage.

This shouldn't be very surprising though. 14900HX is an Intel 7 part and derivative of a design launched 4 years ago. 275HX is TSMC N3B and a new design.

 

[Punkenjoy]

What a pity time for hardware when its news worthy to have a new gen beating in some scenario the old gen.

 

[Tek-Check]

he Ryzen 7945HX3D is also left behind, albeit by a far smaller margin of just around 7% in overall performance. With the Ryzen 9 9955HX3D just around the corner, however, Intel's high-end laptop reign might be short-lived after all.

Way more balanced narrative and bringing more context to readers. Thank you for the effort and well done.

 

[InVasMani]

So works out to like a 100MHz net gain average clock speed bump between base the base and turbo clock speeds at least for what appears to be the P cores. I'm not sure how it looks with E cores, but hopefully a similar approach and maybe even slightly more aggressively at reducing a touch of turbo clock speeds in favor of base clock and a higher average between the two. At least this score uplift in this test appears more pronounced. Intel really needs to turn things around though overall more aggressively. It looks like they made a small bump to the L1 and L2 cache to.

 

[watzupken]

I would think it will be wise to wait for official testing results to confirm the performance of the Core Ultra 2xx mobile chips. Knowing that this is no different from the desktop variant, I do expect it to underwhelm when it comes to gaming performance. So for people who are getting a new laptop for work that benefits from the new chip, then it will make sense to get it. For gamers, I see no point in getting this over say an older Raptor Lake or AMD based laptop if they are cheaper. Battery life may be better with the Core Ultra 2xx, but on gaming laptops, battery life is generally not a big problem.

 

[Vayra86]

Intel: 'we made our CPU less bursty and lowered the peak frequencies so now it performs better'

Me: so wtf have you been doing previously?
This feels like progress Intel already had and then threw in the shitter

 

[phanbuey]

What a pity time for hardware when its news worthy to have a new gen beating in some scenario the old gen.

Especially considering the last generation is actually just a rebrand of the generation before that.

 

[ScaLibBDP]

61010 is 133% of 45615... Or 33% more

It also could be calculated as 45615 is ~74.8% of 61010... Or 25.2% less...

Both results are Valid and final calculated result depends on what number is selected as a base for 100%, that is 45615 or 61010.

Not a big deal to continue talks...