Intel Core Ultra 5 235TA: What Real Users Say

The Asus motherboard paired with the Core 9 285K actually sees a small performance regression in gaming after the patch — the unpatched 285K configuration is 3% slower than the newly-patched configuration.

1. Come up with new naming scheme to make things 'simpler'. 2. Come up with bloated sku variants that are hard to follow. The marketing never learns.

LOL an 'intel' fix that helps AMD more than it does intel.

With Intel it's always 'the next big thing'. Especially in this sub, Intel's comeback is always right around the corner, and when it fizzles out 'we just have to wait for the next one, THAT is the ACTUAL big one'.

It's based on TSMC 3nm (same as Apple M3). Should be a lot cooler and power efficient.

Intel bet it all on Arrow Lake only for it to be almost as bad as Rocket Lake, which was also a regression from the previous generation. Where exactly does Intel go from here?

It can't be worse than 'Ryzen HX 390 AI Max+ Pro'

On notebook we're in pretty good shape. We're expecting this quarter to be a pretty good quarter for Lunar Lake, so notebook is good, solid but we fumbled the high-performance desktop side.

If this cpu can deliver 13900K level of performance in games and isn't overpriced then it'll be really attractive.

Oh god, two more letters to add to the intel vocabulary.

I think this is more of a Intel 7 being such a mature and high volume node than a N3B/LNC design issue. We saw a similar story play out with Intel 10nm/7 as well. Intel 14nm was so mature that Intel wasn't beating 14nm skylake TVB turbo freq till raptor lake.

I wish they never abandoned the previous naming scheme. It was great for so many years...

Arrow Lake just has additional latency that causes frametime spikes. It's great in everything else (not gaming).

To be fair Alder Lake was pretty great after years of Skylake clones and Lunar Lake was also pretty impressive. Arrow Lake is good in the mobile segment.

235 is 3% faster in ST than 14500 at the same clock, but 285K is 9% faster in ST than 14900K despite being clocked 300 MHz lower. I'm confused.

ARL is decently competitive with vanilla Zen 5. The problem remains that Vanilla Zen5 isn't the only Zen5 products for sale, and that X3D exists. NVL vs Zen 6 is going to be mainly a generation focused on big increases in nT performance from both vendors.

How many suffixes does Intel have at this point? We have K, KS, F, KF, X, H, HX, P, U, V, E, KE, Y, C, B, and now A...

Arrow Lake Mobile does have better clocks than Meteor Lake. Core Ultra 5 225H had its P cores clocked at up to 4.9Ghz compared to 4.5Ghz of Ultra 5 125H and its E cores up to 4.3Ghz as opposed to 3.6Ghz.

Gaming is the single largest market for high perf desktop chips. And it's not like ARL is exceptional in the rest. The real problem with ARL is cost. It probably costs 2x or more for Intel to produce relative to RPL, but is at best an incremental upgrade for desktop.

I actually like arrow lake and recommend it for people who want good workstation performance on a budget. The 265k is stellar at that use case.

The problem is they keep switching the suffixes around and don't stick with one consistent one.

It was there since meteor lake from 2023. How much more time are you going to give? Also it's related to its tile-based design and removing the memory controller from the main CPU die.

I think they met expectations, moreso than previous generations. The biggest win is they don't seem to be oxidizing or self-immolating; your CPU not failing due to engineering and design defects is probably the biggest expectation. Aside from that, there have been noteworthy efficiency gains and raw speed is up.

At least they're admitting it instead of spinning it.

How many months of this same song and dance do we need to go through before people accept that Intel's statements cannot be trusted, and Arrow Lake does actually suck.

Decent chips, they just need a big cache to compete with x3D. Non x3D chips are essentially the same as arrowlake.

As long as it performs, it doesn't really matter whatever combination of numbers and letters they name it.

Was expecting a larger ST perf uplift to be honest.

Basically 2 whole node jumps, and we get what essentially amounts to a margin of error as an upgrade? Something is off here.

isn't these are just variant from main SKU for some form factor like all in one PC?

The wildest thing about all this is the hype surrounding Arrow Lake which is made on TSMCs node, while Intel only sees it as a stop gap to put something competitive on the market until their 18A node hits.

The lower end ones are on fire sale. That may be why.

This is wrong. T is power efficient SKU. No one knows what A is... Assuming Chinese market variant.

T series CPU's are only sold to Dell, HP, and Lenovo (and a few other OEM's like Asus in smaller numbers). Thus you only find them on ebay as parts pulls, or in the 1 liter corp desktop systems.

It's just a repackaged version of 'ARL-U', which is completely misleading naming as it's really just an Intel 3 port of Meteor Lake, not Arrow Lake.

I don't get it, if they are power efficient version. Shouldn't they config with more E-Cores and less P-Cores? Instead of 6+8, they could have gone with 4+16 or 4+12?

The one using Intel 3 is interesting, would be nice if some content creator made a small comparison against the TSMC one.

Silently launches because they know their stuff is shit.

Thankful that Intel isn't giving me a reason not to switch to AMD next upgrade tbh.

You could just get a standard 235 (or even better, a 245k because they're cheaper right now) and put a power limit on it in the BIOS to get the same effect.

You need to weigh it up against your specific use case. In a homelab NAS the 245t will probably be overkill. But if you're happy to throw the money at it, it will be energy efficient (as efficient as x86 can be) and will take many years before you outgrow it. In my use case (legacy app - heavily single-threaded), the 245 (non T) made a fantastic choice when weighing price vs single threaded performance.