Intel Core i5-L16G7 is the first "Lakefield" SKU Appearance, Possible Prelude to New Nomenclature?

[R0H1T]

Because of context switching.
So in the Blender example mentioned earlier: heavy rendering would be run on the "big" cores, while light background stuff (OS, networking etc) could be kept on the small ones. That means less switching and less performance lost.

That would necessitate a major overhaul of the kernel(?) & the OS scheduler, didn't we see how inefficient Windows is thanks to bouncing threaded workloads from core to core?
What you're suggesting is not possible on Windows now, though I also believe this is the most interesting chip from Intel in a long, really long time!

 

[bug]

That would necessitate a major overhaul of the kernel(?) & the OS scheduler, didn't we see how inefficient Windows is thanks to bouncing threaded workloads from core to core? What you're suggesting is not possible on Windows now, though I also believe this is the most interesting chip from Intel in a long, really long time!

I don't image that would be any different from trying to pin demanding tasks to "favorite cores".

Unfortunately big.LITTLE doesn't work like that. You can have either the big or the little core active, but not both at once. Of course, Intel doesn't have to copy what ARM did verbatim, but there's a TDP issue trying to run both cores at the same time.

 

[R0H1T]

Yes & big Little has been superseded by DynamIQ for quite some time now. I'd imagine that would be more of the template Lakefield could follow, although it will still need an OS scheduler update at the very least as well as application support. You can't restrict programs to certain cores on Windows by default, & I don't believe favorite cores work that way. Process Lasso does that but it's more of a heavy handed approach & slightly inefficient at that.

 

[bug]

Yes & big Little has been superseded by DynamIQ for quite some time now. I'd imagine that would be more of the template Lakefield could follow, although it will still need an OS scheduler update at the very least as well as application support. You can't restrict programs to certain cores on Windows by default, & I don't believe favorite cores work that way. Process Lasso does that but it's more of a heavy handed approach & slightly inefficient at that.

Don't worry about it, if we have "favored core" on the market, the academia is years away as far as ideas are concerned. I mean, you don't seriously think Intel would trouble themselves with a heterogeneous design and forgot all about scheduling.

 

[notb]

That would necessitate a major overhaul of the kernel(?) & the OS scheduler, didn't we see how inefficient Windows is thanks to bouncing threaded workloads from core to core?
What you're suggesting is not possible on Windows now, though I also believe this is the most interesting chip from Intel in a long, really long time!

I don't see why Microsoft would include such a change if Intel decided to provide this kind of tech in mainstream CPUs. There's benefit for all.
It's already working well in the Windows 10 ARM. So it's not like they can't or are against the idea.

Also, I'm not entirely sure what you mean by Windows being inefficient. You mean that some exotic CPU architectures aren't properly designed for the OS that has 80% PC market share? It's not Microsoft's fault, is it?

 

[R0H1T]

Don't worry about it, if we have "favored core" on the market, the academia is years away as far as ideas are concerned. I mean, you don't seriously think Intel would trouble themselves with a heterogeneous design and forgot all about scheduling.

The question isn't whether Intel's looking/working for this, it's whether MS can actually deliver it. At least to the extent we see it with linux (based OS) especially Android.

It's already working well in the Windows 10 ARM.

Actually it's not, that's the reason why Snapdragon 8xx based devices aren't all that better in terms of battery life. I'll see if I can find a recent article about it but I do remember this point being mentioned specifically in regards to Windows on ARM.

 

[dont whant to set it"']

For anyone who needs Skylake core's I've got a pair of dual core chips >1000miles away. Free , postage not included. Intel need a bit of more than a trowing around as a ragdoll for the lack of innovation since Conroe/Nehalem and come to think of it without other knowledge from my part, those arch names are genuine.

 

[GlacierNine]

Also, I'm not entirely sure what you mean by Windows being inefficient. You mean that some exotic CPU architectures aren't properly designed for the OS that has 80% PC market share? It's not Microsoft's fault, is it?

What kind of bass-sackwards logic are you running on here? There was a time windows couldn't even see more than a single CPU core, let alone schedule multiple cores effectively. It was never AMD or Intel's responsibility to *not* develop better hardware in those circumstances - it was Microsoft's responsibility to react to superior hardware existing so that users could utilise it.

If the industry had worked on this argument then, we'd still not have multitasking.

 

[dont whant to set it"']

@GlacierNine , and probably more than 16/32threads for high-end mainstream, if Intel/AMD/Ms and the like's would of cooperated a lot better ,we'd be a heck of a lot closer to assimilating ourselves. Khala. Do you not?
Le: