Intel Working on Fixing "Arrow Lake" Gaming Performance with Upcoming Patches

[Hyderz]

im guessing they will improve performance by about 5% and move onto the next lake...
i wouldnt mind a core ultra cpu...
it will be a decent upgrade to my current cpu..

 

[LittleBro]

They should fix it in time, not like with RL and ML, nearly two years after the release.
They should fix everything in Arrow Lake that could possibly cause chip degradation.

Man sick of crashes sues Intel for allegedly hiding CPU defects

Intel’s faulty 13th- and 14th-gen CPUs trigger lawsuit out for blood.

arstechnica.com

 

[DevZone]

So what do we guess the fix will be?

More power usage to perform better and likely some adjusting the voltage to to CPU. ?

 

[trparky]

So what do we guess the fix will be?

More power usage to perform better and likely some adjusting the voltage to to CPU. ?

Who knows? However, as I said before, the damage to this generation has already been done; there's no coming back from it.

 

[DevZone]

Who knows? However, as I said before, the damage to this generation has already been done; there's no coming back from it.

I agree and with the problems thay had before, it´s not sure, if they eveer come back from it as the same company atleast

ARM tech is probably going to take over the whole thing both in PC, smartphone and tablets and so on.

 

[Darmok N Jalad]

So what do we guess the fix will be?

More power usage to perform better and likely some adjusting the voltage to to CPU. ?

More scheduling fixes, I'm sure. Now that there's no HT, who knows what Windows is doing with threads during games. I guess one could find out by disabling the E cores and see how things go. It'll be back to running a 9700K.

 

[Crackong]

So what do we guess the fix will be?

More power usage to perform better and likely some adjusting the voltage to to CPU. ?

My guess is,
A big bandage on the thread director and windows scheduler to try to 'Evade' the E-cores while in a gaming load ( White listed manually )

 

[theglaze]

This sounds like a bunch of nonsense to me.
Let's stop pretending anything caught Intel by surprise, they have tested and validated qualification samples for months prior to release, they knew exactly the performance characteristics of the hardware on different OS'.

An updated BIOS shouldn't matter for application performance

PC Gamer did post this comment:

...will Intel's incoming Windows and BIOS fixes make a huge difference?

That's certainly possible, especially regarding the firmware on motherboards, as the [Asus ROG Maximus Z890 Hero] I used for reviewing the Ultra 9 285K and Ultra 5 245K was hugely better in our Factorio test than either of the MSI Z890 boards I have.

Who knows what MSI screwed up with their launch boards, BIOS, drivers, w/e. And just because one manufacturer made mistakes and ruined performance, doesn't mean there's hidden leaps in performance for others to find. But I do think the platform launched half-baked and Intel will tune it for more gaming performance, even if the gains will be small.

 

[chrcoluk]

I see there is a few people that remain convinced HT is a good thing for games. Despite the reports that when its turned off things improve.
To me there is an issue with scheduling (which I am not confident will be fixed by Intel), and the memory latency issue.

The windows scheduler which me and some others ramble about, is far from optimised. Some people still think a 2nd logical core is faster than a real e-core, this simply isnt the case. But the windows scheduler is somewhat stupid, I expect it still schedules on these chips like it expects the HT core to be there, I have observed it on my RL chip when I disable HT core scheduling. Instead of putting the extra thread on a different core, it just puts the extra thread on the same core instead of the 2nd HT core when HT scheduling is disabled, its dumb.

This problem however is mitigated by some manual tuning in the power scheduler settings, the heterogeneous policy 4 (windows 10 default) as an example behaves better than policy 0 (windows 11 default). I also think policy 1 is better than both 0 and 4. The rest of fixing it can be done with something like system informer or process lasso.

A big problem that Windows scheduler has, and I wonder if this will be part of the Intel fix, is if p cores are set as priority for foreground, the scheduler does not like to push extra threads on to e cores, this is pretty dumb, again it seems designed to assume HT is the way.

Intel are to blame for these issues, the same way I said AMD's issues they hold the blame for, to be fair to the Intel rep, he was pushed to blame Microsoft, but he said its all on Intel.

Productivity apps usually use e-cores and p-cores side by side ok, although certain scheduler settings will break this as well.

Also whilst some (a minor amount) of games will use up to 14 threads on consoles, its still much more common to be 7 or less.

 

[Chrispy_]

I'd forgotten that Rob Hallock moved to Intel.
AMD's loss.

 

[Prima.Vera]

They are just polishing this big turd....
Probably the worst CPUs since Pentium 4. Not probably, for sure.

 

[pavle]

They are just polishing this big turd....
Probably the worst CPUs since Pentium 4. Not probably, for sure.

I believe you're on to something there. If we look at L2 cache speed/throughput rate it's drastically lower (half of what it was) than intel's previous 13th and 14th gen or AMD's CPU's.
Nothing can patch cache speeds I believe. But let them cook...

 

[tfp]

AMD was able to address the high cross chip latencies that were seen with Zen5 9900X and 9950X with a BIOS update it could be possible for Intel to correct some latency issues. If intel is lucky they will improve the games that are vastly lower performance moving up the overall averages but I don't expect any real improvement on applications and games that already perform well.

 

[Chrispy_]

Is Arrow lake really that much of a turd? The 285K is up with Ryzen 9000 in the application benchmarks and whilst gaming is behind the X3D chips, it's not terrible - they're Ultra7 is better than the 7700X and 9700X.

Look, I'm not fond of Intel - they've always been the incumbent giant who stamp out competition with shady and illegal practices - but there's no denying that Arrow lake is objectively a decent processor, even if it's neither the best application CPU nor the best gaming CPU - it certainly does a reasonable job of both. On top of the objectively decent performance, the reduction in power consumption compared to 13th/14th gen is good news for everyone. Yes, they're still hotter than, and less efficient than Ryzens of any flavour, but they're more power-efficient than their predecessors which is progress we should be happy about.

 

[efikkan]

Bulldozer vibes with this comment
TBF I am glad I skipped that gen, went from Phenom to Core i7 920 and spent a few years with Intel before going back to AMD when Ryzen landed and I could get a 6c/12t CPU for less than $200 which was unheard of before Ryzen and kept to Intel HEDT lineup

While getting higher core counts to the mainstream was inevitable one way or another, did we actually win anything with with pushing high-power CPUs to the mainstream platforms? What we have now with up to 16 "p-cores", heavy throttling and burst power draw, with very limited IO and "high-end" mainstream motherboards that are severely hamstrung and we are still paying "HEDT prices", so what have power users and enthusiasts really gained from this?
It would have been much better to cap off mainstream platforms at 100W, and keep the proper HEDT segment in all its glory, with a good selection of affordable and flexible motherboards, lots of IO, 4 memory channels and "standard cooler" compatibility. With the popularity of NVMe drives, more PCIe lanes are more important than ever.

Threadripper and Xeon W platforms do exist, but in very limited availability and very high entry (Threadripper), needing special coolers, usually no store stock, and have been lagging behind their mainstream counterparts in getting to the market. (They are rock solid though, for those concerned about long-term stability.)

And back to what you said about Ryzen bringing more cores to the mainstream; Intel had originally scheduled the cancelled Cannon Lake (10nm shrink of Kaby Lake) with 8 cores to launch around the same time as Zen 1, quickly to be followed by Ice Lake (again 8 cores) the following year. While we all know about their 10nm disaster, it is important for historical accuracy to point out that it was node limitations preventing them from adding more cores earlier.

An updated BIOS shouldn't matter for application performance, not unless it somehow incorrectly detects or configures hardware

PC Gamer did post this comment:

The second half of the sentence in my post that you cut explains it; if the BIOS incorrectly detects/configures hardware, it can certainly affect performance. But this would be a bugfix of something reference implementations did correctly during qualification testing, so such fixes are not an optimization what will change how the hardware behaves.

 

[sephiroth117]

8 core cost a lot...
And if gaming 1440p or 4K no need for fastest CPU

I was speaking just for that price range, because i7 and i9 are really either aligned or more expensive.

For 4K some games may benefit from a strong CPU: Flight simulator 2024 and Monster Hunter Wilds (RE Engine is cpu intensive), civ vii or stellaris just for reducing loading time for each turns but that's far less important.

Other than that yes at 4K you are in general GPU limited.

 

[soulphie]

the pre release material showed that intel expected 5% lower performance in games then the 14900k and it reaches round about that performance, there is nothing major to fix in performance this is as advertised it just sucks.