Intel's Lion Cove P-Core and Gaming Workloads

Not “no effort to make sure it performs well in the first place”, that isn’t fair. Lots of effort probably went into it performing well, just this case isn’t handled yet and to be fair this only impacts some people currently and there is a chance to update it.

This just reads like “if they haven’t handled this specific case they’ve made no effort at all across the board” which seems extreme.

Which is funny because the pushback the PS3 got from developers was too much. Maybe it was "too heterogeneous"

But I guess the future was already set.

> Software working poorly in such setup means no effort was made to actually make it perform well in first place.

How do you optimize for a system architecture that doesn't exist yet?

(e.g. CoD could probably be fixed quite easily, but you can't do that before the hardware where that problem manifests even exists - I'd say it's much more likely that the problem is in the Windows scheduler though)

> Games requiring desktop cases looking like a rainbow aquarium with top everything will become a niche

PC gaming has been declared dead ever since the late 90s ;)

There's a giant pile of software - decades worth of it, literally - which was already written and released, much of it now unmaintained and/or closed source, where the effort you cite is not a possibility.

By all means anything released in 2025 doesn't have that excuse - but I can't fault the authors of 15+ years old programs for not having a crystal ball and accounting for something which didn't exist at the time. Intel releasing something which behaves so poorly in that scenario is.... not really 100% fine in my eye. Maybe a big warning sticker on the box (performs poorly with pre-2024 software) would be justified. Thankfully workarounds exist.

P.S. At least I would have expected them to work more closely with OS vendors and ensure their schedulers mitigate the problem, but nope, doesn't look like they did.

People have been making the same argument for more than a decade now, meanwhile PC gaming has only kept growing. If that was the case, why hasn't it happened yet?

Is it?

No one has come close to solving the problem of optimizing software for multiple heterogeneous CPU's with differing micro-architectures when the scheduler is 'randomly' placing threads. There are various methods in linux/windows/etc which allow runtime selection of optimized paths, but they are all oriented around runtime selections (ex pick a foo variation based on X and keep it) made once, rather than on cpu/context switch.

This means that one either uses a generic target and takes the ~1.5-2.5x perf hit, or one optimizes for a single core type and ignores the perf on the other cores. This is probably at least partially why the gaming IPC numbers are so poor. The games are optimized for a much older device.

So a 1.5-2.5x perf difference is these days could be 5-10 years of performance uplifts being left on the table. Which is why AMD seems to be showing everyone a better path by simply using process optimizations and execution time optimization choices with the same zen cores. This gives them both area, and power optimized cores without having to do a lot of heavyweight scheduler/OS and application optimization. Especially for OS's like Linux which don't have strong foreground/background task signalling from a UI feeding the scheduler.

In other words heterogeneous SMP cores using differing micro-architectures will likely be replaced with more homogeneous looking cores that have better fine grained power and process optimizations. Ex, cores with even wider power/performance ranges/dynamism will win out over the nearly impossible task of adding yet another set of variables to schedulers which are already NP-hard, and adding even finer grained optimized path selection logic which will further damage branch prediction + cache locality the two problems already limiting CPU performance.

You can still buy and build (I do) powerful gaming-capable systems that do not emit rainbow-puke or have transparent panels and other gimmicky stuff like that.

>desktops are actually laptops nowadays.

With concepts such as "chromebook" (semi-merging with android phones), apple convergence between iPhone and iComputer chipset and code, Samsung DeX (run samsung phone as a desktop), in my opinion the endpoint is:

- Your phone is your desktop is your laptop. It may dock with a keyboard/monitor. This will happen when software kit is integrated between phone and computer (Microsoft has tried this in the past - too early), and the formfactor is sorted. Also a key gating item is phone CPUS matching laptop performance - Also some form of local AI will be present everywhere - Also, sadly, the way things our headed, such a system will be hardcore locked down if it inherits more from the phone ecosystem than the OS ecosystem

Old games won't get updates. That is why there are multiple separate tools that try to force the situation. e.g. Process Lasso

Tell that to AMDs bulldozer. There is something to be said for considering theoretical performance, but one can't ignore how hardware works in practice, now.

Is this a problem with Intel or with the OS scheduler? Haven't they had this kind of CPUs for a few years now, with this being touted as a reason to move from Win10 to 11?

Scheduling is the OS’s decision not the CPU’s.

Although Intel’s configurations are unhelpful to say the least: it’s hard to make good scheduling decisions when Intel sells a 2P, 8E, 2LE as a 12 cores to the user.

The recommended actions are built around the assumption optimal gaming performance is the only market segment to please. Of course, efficiency cores were never about pleasing that segment anyways - gamers want fewer huge cores, not lots of big cores or tons of efficient cores. That Windows gaming is slightly imperfect due to scheduling difficulties will not stop AMD, Intel, ARM, etc from continuing to use different sized cores on non-uniform interconnects if it means larger changes in perf/watt and perf/cost for that price.

In practice this is more of a problem for gaming news site headlines than users anyways. For all but the biggest enthusiast the uplift from going to a new system is usually so much they'd not notice it could be a couple percentage points better or that one game is still a bit hitchy. It's us ultra-nerds that care about the 1% lows between different top end CPUs, and we tend to be quickly convinced the problem is Microsoft's to solve for us when the scheduling problem hits ARM, AMD, and Intel based Windows devices equally.

So, for one in other software the new processors do better. The 285K beats the i9-14900KS by a bit in my app benchmark collection (which is less extensive, but still). And second yes, according to https://www.computerbase.de/artikel/prozessoren/intel-core-u... for example they are less extreme in their energy usage and more efficient in general, albeit not more efficient than the AMD processors.

But it is a valid question.

> I don't fully follow this, so what has been gained with the new models?

There were power efficiency gains, as well as nice productivity improvements for some workloads: https://www.youtube.com/watch?v=vjPXOurg0nU

For gaming, those CPUs were a sidegrade at best. To be honest, it wouldn't have been a big issue, especially for folks upgrading from way older hardware, if only their pricing wasn't so out of line with the value that they provide (look at their GPUs, at least there the MSRP makes the hardware good value).

> I seem to remember you'd need dedicated industrial cooling for the 14700k.

Those CPUs run hot, but it got exaggerated a lot online. It’s not hard to handle their heat with a good air cooler (even some of the $50 units like the Peerless Assassin) or a run of the mill closed loop water cooler.

There are a lot of old notions in the gaming community that you need to keep CPUs under arbitrary temperature thresholds or that any throttling is bad. Modern CPUs and GPUs run themselves deep into the performance curves and slight throttling is barely noticeable.

I see how that can be misleading. It's globally not a percentage based score.

The benchmark is creating a global order based on which one is faster, including indirect comparisons. But that only takes care of the position, not the rating. That one is based on the percentage between direct neighbor iff there is a direct comparison in the benchmark data, otherwise it's a small .1 increment. So many small percentage increases that don't necessarily match the direct comparion between parts that are not neighbors. Sometimes that works great, sometimes not.

Here the example looks a bit pathological, that difference is further off than I expected when introducing the calculation recently. For the 12700K and 13700K, the direct comparison sees the 12700K at 85% of the 13700k:

https://www.pc-kombo.com/us/benchmark/games/cpu/compare?ids%...

For apps it's 79%:

https://www.pc-kombo.com/us/benchmark/apps/cpu/compare?ids[]...

So yeah, sorry, that part is misleading right now. I'll check the calculation.

But the ingested individual benchmark numbers and the ranking, so the position, is very solid I'd say. With the caveat that ranking position can change with more data.

I found a bug in the score calculation that inflated the score, this case is closer now.

Hey, I do try to make the site as transparent as possible - but admit that the site does not make it obvious. For such a doubt, go into the comparison of the two (https://www.pc-kombo.com/us/benchmark/games/cpu/compare?ids%...) where all benchmarks used that the two processors share are listed. The benchmark bars are clickable and go to the source.

It does get really complicated to address something like that when all comparisons are indirect. Thankfully, that's not the case here.

The 13900K and 14900K in games really have been that close, see https://www.computerbase.de/artikel/prozessoren/intel-core-i... for an example, where the two have a 2% FPS difference.

It is an interesting but particularly non-actionable analysis: only a handful of engineers at Intel are in a position to design an improved Lion Cove, while the main takeaway for the game programmers who care about game workload performance is that nothing performs too badly and therefore general performance improvement techniques like accessing less memory are a good fit for this processor.

How so?

Not that I disbelieve, I just wasn't especially picking that up from the article.

exactly. I'm very happy to notice there are no 'x86 bad arm good' comments here as of now. a welcome change.

also - or maybe, first and foremost - it's just a very good article.

Interesting. You realize this by identifying the offending assembly instructions and then see that one operands comes from memory?

That's more of a graduate level book. If you would like a book by the same author, that is lighter in content but much more approachable, I recommend this version: https://www.cs.sfu.ca/~ashriram/Courses/CS295/assets/books/H...

And don't forget to look at the online appendices on the MK website. They are also all very good.

I'm pretty sure memory is still the main bottleneck in the vast majority of operations, and why the X3D is such a monster with so much L3 cache. There's almost nothing that doesn't benefit from having twice the memory bandwidth.

Apple's M-series of chips make use of more than two memory channels. I don't think there is evidence that it is actually beneficial for anything but marketing though.