News Intel announces Arrow Lake fix coming within a month — Robert Hallock confirms poor gaming performance is due to optimization issues

[Admin]

Intel's Robert Hallock says a fix for Arrow Lake will arrive within a month, and that Arrow Lake's poor gaming performance is a result of tuning and optimization issues — not memory latency or the company's decision to switch to a tile-based architecture.

Intel announces Arrow Lake fix coming within a month — Robert Hallock confirms poor gaming performance is due to optimization issues : Read more

 

[why_wolf]

So either he is lying or their internal QA is completely incompetent.

 

[flowingbass]

Intel can wince, cry, beg. All i see are excuses.

It's good to see their past dirty, evil, manipulative business tactics come back and bite them in their complacent balls.

Desperately clinging onto their withering and rotting laurels, counting very heavily on "too big to fail" mantra.

Their contempt for the entire world with quad cores but with hyperthreading! Got em! was like the dark ages.
Just because the competition was stuck in contract with lackluster semiconductor manufacturing process.

Karma finally put its foot down on intel.

 

[blargh4]

So after the last couple generations self-destructed, they're still unable to properly qualify and test a chip before shoving it on the market? Very encouraging.

 

[bluvg]

Zen 5 launch, mixed bag of results: "stick with Zen 4 for now, but it'll get better"
Intel ARL launch, mixed bag of results: "ARL is a disaster, Intel must die"

It may be fashionable (and warranted in some cases) to pig-pile on Intel right now, but maybe cut them the same slack as others for a product fresh out of the oven? A strong Intel is good for the industry.

 

[diminishedfifth]

If all these "third party reviewers" are coming to the same conclusion, which does not align with Intel's conclusion.
Maybe it's an Intel issue with how they test their products, cause we all know third party reviewers. All have the exact same setups and do everything in coordination with each other to coordinate their attacks against Intel 😭

 

[bit_user]

Wow, why can no one seem to pull off a decent launch, these days?

I guess AMD redeemed itself with the 9800X3D, at least.

 

[bit_user]

Zen 5 launch, mixed bag of results: "stick with Zen 4 for now, but it'll get better"
Intel ARL launch, mixed bag of results: "ARL is a disaster, Intel must die"

That's a false equivalence. Zen 5 was faster than Zen 4 in nearly every game. Arrow Lake was slower than Raptor Lake in nearly every game.

It may be fashionable (and warranted in some cases) to pig-pile on Intel right now,

You're just trying to dismiss legitimate concerns as bias. You're not fooling anyone who actually followed these launches. Maybe you didn't?

A strong Intel is good for the industry.

There are plenty of other competitors, coming up now. If Intel leaves the scene (and I'm not saying they will), AMD won't have a chance to get lazy.

 

[hotaru251]

Honestly ignoring the performance of their new cpu's the Intel "quality" they have earned over 20+ yrs is basically dead at this point.

With a company the size of Intel (and its massive compared to amd) you shouldnt be launching cpu's with these issues as should been caught when testing them.

A strong Intel is good for the industry.

yes, but that doesn't get them free ride.
They have made blunder after blunder and thats a them issue. Nobody should be putting em on a pedestal for that. Its real life not a children's game where everyone gets a trophy win or lose.
People WANT Intel to be like they were in the past (as again competition keeps prices low & progress moving).

They have to earn it though not have it handed to em freely.

 

[endocine]

Rob Hallock went from bashing intel, to working there, and now apologizing for them. Very trustworthy individual, and should believe every word when he says its an optimization issue and will get fixed.

 

[thestryker]

It was very obvious from some of the reviews that there were some outliers in gaming performance. I highly doubt any of these updates/fixes will fundamentally change anything overall conclusion wise, but I'd expect results like when all 3 ARL CPUs are slower than Zen 3 (non-X3D) to go away.

When Steve from HUB did their testing on the 245K he found even more dramatic outliers where just that SKU was slower than the rest by a fair margin. I think it was Jay (JayzTwoCents) who saw some random gaming performance upticks from E-core overclocking which really shouldn't happen unless it's a super core heavy title.

Pretty sure Intel knew about the performance issues since Steve said as much in a recent HUB podcast where he talked about the communication difference between Intel and AMD and how Intel told him his results weren't notably different from theirs so I'm not sure what Hallock is referring to (though I haven't watched the video so I do not know exactly what was said and am basing on what this article indicates).

Edit:
I watched the interview and he says it was when they were getting reports from reviewers and verifying is when they found the issues. So it wasn't as simple as just Intel seeing one thing and reviewers another like the article here says.

If I had to guess Intel probably does like AMD and uses canned benchmarks so things can be automated more easily. This is something several reviewers have gotten away from for a lot of CPU testing as many canned benchmarks are very GPU heavy even with games that are CPU heavy. CP2077 would potentially be an example in that Intel's results showed 21% behind 7950X3D while TPU and TechSpot/HUB show 33-40% difference. I'm assuming these circumstances would be the only ones where Intel's data didn't match reviewer.

We'll see if what he says comes to pass with regards to changing processes and speaking publicly about the specific performance issues. I'll believe it when I see it.
/Edit.

I'm tired of the bog standard deflect which is why I liked Tom Peterson talking about Arc and just being honest about it needing to take time to fix what could be. Still no excuse for a messy launch, but at this point a company being honest about what's going on is an anomaly.

 

[DingusDog]

Zen 5 launch, mixed bag of results: "stick with Zen 4 for now, but it'll get better"
Intel ARL launch, mixed bag of results: "ARL is a disaster, Intel must die"

It may be fashionable (and warranted in some cases) to pig-pile on Intel right now, but maybe cut them the same slack as others for a product fresh out of the oven? A strong Intel is good for the industry.

Can't lump the 9800X3D in with the Zen 5 launch.

 

[Varsaggo]

So either he is lying or their internal QA is completely incompetent.

Because?

 

[watzupken]

“According to Hallock, Arrow Lake's performance from third-party reviewers did not align with what Intel saw in its internal testing. Hallock noted a massive disconnect between third-party review performance and Intel's internal testing. ”

What is perplexing is how did Intel test these internally to arrive at their results which they claim is very different from external review results. They said the issue is due to BIOS and OS, so why release a BIOS to retail when they have something more refine and better? Same for OS, what version of OS are they running that yields a significant delta in performance? And if they have a newer driver or whatever at the OS level, why not release it? It’s not like they just figured out something they missed out previously. So the statement he made don’t make sense to me.

 

[-Fran-]

Hm... I strongly suspect this man is lying on behalf of Intel's managerial line (execs).

I highly doubt they "didn't see this in test". I strongly suspect they knew, but still decided to launch and say "we'll fix it after", like with ARC.

As they say: "fool me once, shame on you; fool me twice, shame on me".

The level of incompetence shown by Intel in this launch was quite honestly baffling, even edging out AMD's own incompetence by a mile. Just baffling.

And to all the weird nerds: stop jumping in front of your darling Companies when rightful criticism is headed their direction, please. It's sad.

Regards.

 

[TheHerald]

That's a false equivalence. Zen 5 was faster than Zen 4 in nearly every game. Arrow Lake was slower than Raptor Lake in nearly every game.

And it's going to remain that way. Whatever patch they are cooking is probably for these few mega problematic games (like cyberpunk) that have performance drop like a rock. I do not expect this patch will actually make arrow faster than raptor in gaming. It's literally impossible due to the IMC being on a different planet.

 

[tom2tec]

classism leads to corruption which breeds incompetence

 

[meski42]

Is Intel trying to be the Boeing of CPUs?

 

[magbarn]

Is Intel trying to be the Boeing of CPUs?

They're already there.

 

[KraakBal]

Too much politics put into cpus vs engineering

 

[bit_user]

Too much politics put into cpus vs engineering

The marketing departments at these companies are always going to try and spin whatever they have in the best possible light. It's our challenge to cut through the BS and focus on the actual tech.

 

[bluvg]

That's a false equivalence. Zen 5 was faster than Zen 4 in nearly every game. Arrow Lake was slower than Raptor Lake in nearly every game.


You're just trying to dismiss legitimate concerns as bias. You're not fooling anyone who actually followed these launches. Maybe you didn't?


There are plenty of other competitors, coming up now. If Intel leaves the scene (and I'm not saying they will), AMD won't have a chance to get lazy.

No, it was exaggerating to make the point that many product launches have historically often been initially disappointing but improve over time, yet the reaction from some is wildly different. I'm dismissing nothing ("warranted in some cases"). I've followed these products from rumor through launch a little too closely at times.

If by plenty of other competitors coming up you mean nVidia, Qualcomm, and MediaTek, they've already been around for quite some time, and for many reasons, they're not 1-1 competitors. Aside from a brilliantly managed and executed merger, Intel "leaving the scene" would not be a good outcome for the industry or consumers, particularly the US.

 

[bit_user]

If by plenty of other competitors coming up you mean nVidia, Qualcomm, and MediaTek,

Not only the ARM crew (which includes Apple, for what slice of the market they can play for), but also Zhaoxin x86, Loongarch, and the RISC-V folks coming up behind them.

I'm not saying gaming PCs will stop being predominantly x86 any time soon, but the performance desktop segment is now a minority share of the overall computing market. Intel and AMD are worried about much more than just each other.

I think the main reason AMD is looking at ARM is that they're well aware that even if they continue increasing their share of the x86 cloud computing market, their overall datacenter revenues could plateau if their penetration of the x86 segment slows to a pace less than or equal to the decline of x86 in the cloud.

 

[cyrusfox]

I really believe this lackluster launch is a direct result of the cost cutting and transformation Intel is attempting. You don't cut 15% without seeing a drop off, key people leave and the remaining workforce has gaps. You don't know what you don't know. Intel is forced to relearn lessons they never properly documented due to their silo'd need to know exclusionary culture.

 

[bit_user]

I really believe this lackluster launch is a direct result of the cost cutting and transformation Intel is attempting.

That seems like a stretch. It happened too recently. Whoever was managing the release would've had a schedule with the remaining work planned and would know who's on the hook for what. Employees key to that delivery wouldn't have been terminated. Also, keep in mind that the 15% workforce reduction didn't occur on the day it was announced. It unfolded in the following month or more, I think. It's not like just turning up for work and finding 1 of every 6 colleagues randomly absent.

Any who quit of their own accord would've given notice (usually 2 weeks). Their manager could've arranged time for knowledge transfer to their colleagues who were remaining.

Also, don't forget that Lunar Lake was basically the same architecture and launched a couple months earlier. That should've provided some initial experience tuning big/little scheduling, etc. Furthermore, the launch date for Arrow Lake seemed to slip a few weeks, which should've given them even more time.

Basically, if this does really explain the botched launch, then Intel already would've had to be a clown show.