Tuesday, November 14th 2023

Intel Confirms APO Feature Not Coming to 13th Gen and 12th Gen Core Processors

Intel Application Performance Optimization (APO) is a unique feature that sets 14th Gen Core "Raptor Lake Refresh" processors apart from 13th Gen "Raptor Lake," despite the two being practically the same microarchitecture. APO is software-based, application-specific processor optimization that is found to offer an up to 16% performance boost in "Metro Exodus," and an up to 13% boost in "Rainbow Six: Siege." These are the only two games that Intel has released APO optimizations for, and for now, APO is only supported on the Core i9-14900K and i9-14900KF. Hardware Unboxed learned that while the company might bring APO to more 14th Gen Core processor models in the near future; it won't make it to 13th Gen Core "Raptor Lake" and older 12th Gen Core "Alder Lake."

APO is an extension of Intel Dynamic Tuning Technology framework that provides a greater degree of handholding for the OS scheduler, to give a specific application the best possible allocation of hardware system resources. For Intel to release APO profiles to new games, it needs extensive testing and validation specific to processor models and the applications themselves, which is probably why the company is limiting APO to only its current 14th Gen processors, and only specific processor models within the lineup. You can catch the Hardware Unboxed presentation with their testing of APO on the two supported titles, and Intel's statement, in the source link below.
Source: Hardware Unboxed (YouTube)
Add your own comment

75 Comments on Intel Confirms APO Feature Not Coming to 13th Gen and 12th Gen Core Processors

#26
AusWolf
thestryker6Nope Thread Director by itself analyzes the threads, but it cannot place them that's the job of the scheduler. APO is seemingly bypassing the scheduler at least to some degree.

Preferred cores just identify to the OS the highest performing cores this does not guarantee Windows won't ping pong threads across the CPU which it certainly does (I'm not referring to the high power threads here). APO is clearly keeping this behavior in check better than the scheduler does by itself.
So both Thread Director and APO are supplements to the scheduler - just as I thought.
Posted on Reply
#27
ZoneDymo
That's fine, I'll go AMD for the next build
Posted on Reply
#28
Eternit
Typical for Intel. Their P E cores cause performance issue in games (and other apps) so they created software solution that greatly improves the performance, but requires some refactoring in the games code. They require their latest high end CPU to work, so game devs ignore it, so they paid for implementation in two titles. It shows the potential, but it will be ignored because of its costs and small customer base benefiting from it.
Posted on Reply
#29
AusWolf
EternitTypical for Intel. Their P E cores cause performance issue in games (and other apps) so they created software solution that greatly improves the performance, but requires some refactoring in the games code. They require their latest high end CPU to work, so game devs ignore it, so they paid for implementation in two titles. It shows the potential, but it will be ignored because of its costs and small customer base benefiting from it.
It's a stopgap generation anyway, only being released to boost publicity. The next "Lake" will be fundamentally different if they stay true to their words.
Posted on Reply
#30
Eternit
AusWolfIt's a stopgap generation anyway, only being released to boost publicity. The next "Lake" will be fundamentally different if they stay true to their words.
Fundamentally different but still with P and E cores and games have problems with proper utilisation of them. Also I do not trust Intel with their claims they will have high performance CPU in 20A next year.
Posted on Reply
#31
thestryker6
Eternitbut requires some refactoring in the games code
It doesn't touch game code at all and game developers have nothing to do with it.
Posted on Reply
#32
Eternit
thestryker6It doesn't touch game code at all and game developers have nothing to do with it.
So why it works only with two games?
Posted on Reply
#33
thestryker6
EternitSo why it works only with two games?
The easiest way to explain it without spending a ton of time is think of it similarly to a video card game profile. In effect Intel has generated a model based upon how those specific titles work on those specific CPUs and APO implements it.
Posted on Reply
#34
lemonadesoda
I thought Apple had the monopoly on software enhancing/software crippling new vs. old.

I've always suspected that the new architecture with its asymmetric design makes thread scheduling horribly complex and in most situations non-optimal. You can keep most of the threads happy most of the time, but you can't keep all of the threads happy all of the time!

This new Intel "Thread Director profiling" utility is a proof of that.
Posted on Reply
#35
InVasMani
Intel may have seen my post comment on TPU awhile back talking about E cores and the shared cache and how it's larger than that of a individual P core. I pointed out that similar to disabling a CCX for latency reasons with stacked cache similarly could be applied to E cores with the shared cache access to reduce latency and heat output likewise that also has a great side benefit of reducing power and allowing for longer turbo duration and/or higher turbo boosting.

Anyway that said 12th gen had smaller E core cache size per cluster, but also didn't have individual multiplier for E cores within a cluster either so wouldn't work as well dynamically in terms of reducing power and heat output nor would the share cache latency uplift be nearly the same in terms of effectiveness. In the case of 13th it wouldn't be too different. The only minor difference for 14th gen relative to 13th gen is base/boost would be a little different and thus behave bit more effectively, but the cache is the same. In the case of 14700K relative to 13700K it naturally would be more effective though due to the additional set of E cores.

The bottom line is tasks don't always benefit from more multi-threading and don't all have heavy CPU utilization. Certain tasks are more latency driven or just have a certain threshold of interment latency driven points where cleverly utilizing the E cores better makes sense and would lower latency and temps to improve turbo performance and overall performance. It seems like that is what APO is doing. Are they doing things beyond that scope of that I have no idea, but it stands to reason their leveraging what they can in whatever ways they can to raise performance, reduce heat, and drop power consumption.
Posted on Reply
#36
FrostWolf
The fact that APO isn’t supported on the 14700k just makes this another non-starter for me.

It’s not even “only supported by the 14th gen” it’s “only supported by two CPUs”. I see no value in it.
Posted on Reply
#37
TheoneandonlyMrK
Harsh, it's clear there are no architectural reasons just more artificial segregation, I hope they are roasted in the forums for this shit move.
Posted on Reply
#38
A Computer Guy
lemonadesodaI've always suspected that the new architecture with its asymmetric design makes thread scheduling horribly complex and in most situations non-optimal. You can keep most of the threads happy most of the time, but you can't keep all of the threads happy all of the time!

This new Intel "Thread Director profiling" utility is a proof of that.
I'm not sure if you can do this yet but it would be nice to be able to program thread A to P-core thread B to E-core (etc...) more specifically to optimize at the application level much in the same way you can create high and low priority threads.
Posted on Reply
#39
phanbuey
FrostWolfThe fact that APO isn’t supported on the 14700k just makes this another non-starter for me.

It’s not even “only supported by the 14th gen” it’s “only supported by two CPUs”. I see no value in it.
Not only is there no value in it. It's actually damaging the brand. The whole shtick that Pat was trying to foster was "The nerds are back in town" i.e. - the 1st year business school 'tactics' they're notorious for were done with and they were going to focus on the tech.

What they very clearly demonstrated to the market is that those shenanigans are alive and well -- and they did so at almost no benefit to sales or their product stack. It's basically all downside.



Reviews are turned off at the store for me... probably because it was 1 star. lol. What a PR blunder.
Posted on Reply
#40
Prima.Vera
Look like nowadays everybody is taking nGreedia callous model of providing "features" only to the latest generation.
Vote with your brain and wallet people.
Posted on Reply
#41
Lord Romulus
Nothing that a political pen can't solve, forcing the company to provide support. 12th/13th gen are still right there behind and there is no justification for this, just greed to force sales of new products.
Posted on Reply
#42
AusWolf
lemonadesodaI thought Apple had the monopoly on software enhancing/software crippling new vs. old.

I've always suspected that the new architecture with its asymmetric design makes thread scheduling horribly complex and in most situations non-optimal. You can keep most of the threads happy most of the time, but you can't keep all of the threads happy all of the time!

This new Intel "Thread Director profiling" utility is a proof of that.
TheoneandonlyMrKHarsh, it's clear there are no architectural reasons just more artificial segregation, I hope they are roasted in the forums for this shit move.
Prima.VeraLook like nowadays everybody is taking nGreedia callous model of providing "features" only to the latest generation.
Vote with your brain and wallet people.
Agreed, agreed, and agreed. As long as Intel is "all about the e-core p-core bullshit" (Limp Bizkit, anyone?) going hand in hand with software optimisations that are restricted to their highest-paying buyers, I'm gonna stay with AMD. Same with Nvidia and their closed DLSS/RT ecosystem (whatever the heck they mean by this word) and its newest version being locked to the latest gen. Thanks, but no, thanks.
Posted on Reply
#43
BoggledBeagle
Why only 2 games so far?

One could expect Intel will get on some schedule, as adding one new game in the supported list a day, but the number 2 has not changed since introduction. Are they stuck for some reason? Is something broken? Have they problem finding games which can be improved? What is going on?
Posted on Reply
#44
AusWolf
BoggledBeagleWhy only 2 games so far?

One could expect Intel will get on some schedule, as adding one new game in the supported list a day, but the number 2 has not changed since introduction. Are they stuck for some reason? Is something broken? Have they problem finding games which can be improved? What is going on?
Probably one of these reasons. I don't think Intel cares much, though. They just need the publicity to sell 14th gen.
Posted on Reply
#45
A Computer Guy
BoggledBeagleWhy only 2 games so far?

One could expect Intel will get on some schedule, as adding one new game in the supported list a day, but the number 2 has not changed since introduction. Are they stuck for some reason? Is something broken? Have they problem finding games which can be improved? What is going on?
They need to do something like partner with Steam and get access to data of who has what processor and games and let Steam deliver the optimization and incentivise devs to do the optimization in priority of who is using what and playing what. At least that is how I would do it unless Intel is going to piggyback on the ARC team to blindly optimize games for ARC and CPU's using APO at the same time. Spreading out the work among the vested interests would be the better cost effective way to do it and get them all working in the same direction optimizing for Intel chips. A sort of different way to undercut the competition by ecosystem building and squeezing out their time they would otherwise spend with AMD technologies like FSR.
Posted on Reply
#46
BoggledBeagle
What if only games which work in some odd or very specific way can be improved? If it was some easy thread scheduling tweak, there would be most likely much more games supported.
Posted on Reply
#47
A Computer Guy
BoggledBeagleWhat if only games which work in some odd or very specific way can be improved? If it was some easy thread scheduling tweak, there would be most likely much more games supported.
That's kind of what I am getting at with my threads idea. If windows updates their threads api to tag threads for P or E cores scheduling (a suggestion to the thread scheduler) then devs would just do that. Simple!
Posted on Reply
#48
lexluthermiester
GoldenXTypical Intel.
As much as I'd like to be the optimist, I have to agree with this.

@ INTEL
This is shady, dishonest, borderline unethical crap. :shadedshu:
Morons..
Posted on Reply
#49
dyonoctis
ChaitanyaWhile watching Gamersnexusreview they showed screenshot of MS store where the APO was rated 1.5 stars but when I checked the MS App store it seems like reviews have been scrubbed.
phanbueyNot only is there no value in it. It's actually damaging the brand. The whole shtick that Pat was trying to foster was "The nerds are back in town" i.e. - the 1st year business school 'tactics' they're notorious for were done with and they were going to focus on the tech.

What they very clearly demonstrated to the market is that those shenanigans are alive and well -- and they did so at almost no benefit to sales or their product stack. It's basically all downside.



Reviews are turned off at the store for me... probably because it was 1 star. lol. What a PR blunder.
I think that it's just the store being weird.
Posted on Reply
#50
InVasMani
It seems like a lot of people are kind of hopping on the Intel hate train w/o really any of us being briefed fully on how APO operates and why it can or isn't being supported for 12th gen and 13th gen. I suspect their gate keeping pretty hard highhandedly against 13th gen while 12th wouldn't provide the same tangible uplift we see with 14th gen for APO that could probably be nearly identical for 13th gen as well.

I saw that based upon how I suspect APO operates based on post I had made on TPU around E cores shared cache and intelligently utilizing it to reduce latency, but also has a side benefit of dropping power consumption and temperatures as well that translates to power savings and/or higher performance in tandem with boosting algorithms. It could provide higher peak performance for a individual cooling restraints or longer boost duration for higher sustained performance. In either case it's a net positive.

Basically suspect Intel has mimicked disabling of CCX cores on AMD side to provide uplift within the E core cluster. It's a bit like treating each core within a cluster of E cores as a CCX if it operates as I suspect it does and then cleverly dynamically enabling or disabling them based upon the workload. It's not going to to provide performance uplift for the same reasons, but in lighter workloads that are more latency and thermal sensitive it can show positive gains while also increasing efficiency.
Posted on Reply
Add your own comment
Dec 22nd, 2024 16:02 EST change timezone

New Forum Posts

Popular Reviews

Controversial News Posts