Intel unlocks APO optimizations for new Arrow Lake CPUs, expands game support

zohaibahd

Posts: 934   +19
Staff
In context: Intel's Application Optimization is a utility that makes sure hardware is firing on all cylinders for supported apps and games. By intelligently optimizing thread scheduling along with application threading for a given workload, APO can squeeze out improved performance – up to a 31-percent boost in some gaming scenarios.

Intel has officially "verified" its top-tier Arrow Lake CPUs will play nice with APO software. The new Core Ultra 9 285K, Core Ultra 7 265K, and Core Ultra 7 265KF processors are the first Arrow Lake chips to get the APO stamp of approval. Games taking advantage of the optimization tech could see some decent performance gains.

Users rocking older Intel processors can also use APO, though the chip must have at least six P-cores, and the software will only run the more limited 'Advanced Mode' instead of the verified experience. The new Ultra 5 Desktop Processors (Series 2) can take advantage of the feature, albeit with limited support in Advanced Mode.

Besides the new chips, Intel also confirmed a dozen new game titles that employ the APO optimizations, including several big-name hits. Here's a list of the newly added games:

  • Company of Heroes 3
  • Counter-Strike 2
  • Cyberpunk 2077
  • Dota 2
  • Fortnite
  • Naraka: Bladepoint
  • Riftbreaker
  • Shadow of the Tomb Raider
  • Tiny Tina's Wonderlands
  • Total War: Pharaoh
  • Total War: Three Kingdoms
  • Total War: Warhammer 3

It's great to see Intel continually expanding game support for APO, as the software seems truly helpful in extracting some extra oomph from a rig's hardware. It's also nice that the company includes popular games for benchmarking, like Cyberpunk and Tomb Raider, in these APO updates. That should make it easier for enthusiasts and websites to test the real-world benefits without scrambling to find compatible games. The app also allows users to toggle between the modes and select which games to apply the optimizations to.

The APO software is not yet available through the Intel download center. Instead, users have to download the interface through the Microsoft Store.

Permalink to story:

 
This is not a good feeling that system performance is so dependent on opaque, black-box, per-application adjustments. And one has to wonder, if it is really that sensitive to unique configurations, is the Intel provided configuration really right for all scenarios involving that game? What if you're playing with higher or lower graphics settings, or have more additional activities going on at the same time?

I'd like to hear more about what these settings are doing, if there are similar user-accessible direct controls, and why the O/S can't get closer to optimal using its own real time adjustments to current workload.

 
With the cpu business so neck and neck( some may argue )Intel was hoping that developers will optimize for it's hardware and learned the hard way. AMD is learning the same thing with it's gpu business while Nvidia is loving the brute force of it all!

Unfortunately if you are not going to spend resources on optimization for your hardware the developers especially now will not do it for you unless there is an incentive like 88% market dominance.
 
Please note in those benchmarks Intel posted for Arrow Lake showing it beating 9950X they used APO in several games to pump the score. That is not apples-to-apples and if those titles are excluded it looks a lot worse for Arrow lake. Not saying it's not a good thing for the gamer, but we are trying to compare cpu's not a software and see what the architecture is doing.
 
Shouldn't this be something Windows pulls in automatically if it's free performance? Or rather why isn't it part of Windows own scheduler?
Would something similar benefit Linux or is Linux getting the most out of it already by default?

The more AMD and Intel start deviating from the classic all the same cores with the same clocks design the more it seems to highlight shortcomings in Windows. meanwhile Linux has been dealing with various mobile architectures that mix multiple different core designs for over a decade?

I wouldn't mind seeing a deep dive article comparing CPU bound games on one of these arrow lake chips and a 14th Gen and a dual CCD AMD chip.

And perhaps throw in the people's champion 5800X3D for comparison. An awful lot of work but that's Steve's superpower ;).
 
Last edited:
Maybe AMD should do something similar, so their dual CCD chips (3D especially) don't suffer too hard in games and need tinkering with Process Lasso. 99% of users won't do this anyway.

7800X3D beats 7950X3D in gaming just like 9800X3D will beat 9950X3D in gaming, and they do it at much lower power usage as well + lower price.

Single CCD has always been a nobrainer for gaming perf. No hassle, just best gaming performance as games don't use the wrong cores here or try to use second CCD which introduce latency and reduces performance.

7900X3D is meh due to only having 6 cores with 3D cache and AMD seems to repeat this mistake with 9900X3D sadly.

Which I find strange, considering 7900X3D sales numbers were terrible. Sounded good on paper, 12 cores 24 threads, however, only 6 cores with 3D cache which forced many games to utliize 2nd CCD reducing fps alot.

7900X3D performs closer to 7600X3D than 7800X3D as a result. Both have 6 cores with 3D cache so it makes sense somehow.

7900X3D and 9900X3D probably only exist so AMD can use faulty CCDs. Only needs 75% of the CCD working.
 
Last edited:
When you have "thread-director" to optimize the core usage, I don't understand why you need another bespoke optimization tool on top of it. They are just creating more software bloat. While such bespoke optimization is still free performance, I feel it is a very feeble way of squeezing more performance because it is another software layer and may break for whatever reasons.
 
When you have "thread-director" to optimize the core usage, I don't understand why you need another bespoke optimization tool on top of it. They are just creating more software bloat. While such bespoke optimization is still free performance, I feel it is a very feeble way of squeezing more performance because it is another software layer and may break for whatever reasons.

Probably because Thread Director is not working right, this is true for AMD as well.
 
So thread director is directing what? Is APO intel’s way of overriding the windows thread scheduler ? Is this a tacit admission that windows is unoptimized especially for hybrid core architectures? Are these intel chips way faster than “stock” windows (with no apo) would have you believe? Does intel not need its own version of x3d cache to compete?
 
Maybe AMD should do something similar, so their dual CCD chips (3D especially) don't suffer too hard in games and need tinkering with Process Lasso. 99% of users won't do this anyway.

7800X3D beats 7950X3D in gaming just like 9800X3D will beat 9950X3D in gaming, and they do it at much lower power usage as well + lower price.

Single CCD has always been a nobrainer for gaming perf. No hassle, just best gaming performance as games don't use the wrong cores here or try to use second CCD which introduce latency and reduces performance.

7900X3D is meh due to only having 6 cores with 3D cache and AMD seems to repeat this mistake with 9900X3D sadly.

Which I find strange, considering 7900X3D sales numbers were terrible. Sounded good on paper, 12 cores 24 threads, however, only 6 cores with 3D cache which forced many games to utliize 2nd CCD reducing fps alot.

7900X3D performs closer to 7600X3D than 7800X3D as a result. Both have 6 cores with 3D cache so it makes sense somehow.

7900X3D and 9900X3D probably only exist so AMD can use faulty CCDs. Only needs 75% of the CCD working.
Pretty sure they're putting X3D Cache on both CCDs for Ryzen 9000.
 
Back