Intel Thread Director Virtualization Patches Boost Some Workloads By ~14%

[phoronix]

Phoronix: Intel Thread Director Virtualization Patches Boost Some Workloads By ~14%

Intel's hybrid core handling for modern Intel Core CPUs with a mix of P and E cores has largely been in good shape under Linux for a while. Intel Thread Director support has come along with various Linux kernel improvements to better handle task placement between the P and E cores. One area seeing new work now though is for virtual machines (VMs) running on Intel hybrid systems with a new Linux kernel patch series working on Thread Director Virtualization...

https://www.phoronix.com/news/Intel-Thread-Director-Virt

[stormcrow]

I find it even more interesting to note the layout than the results... Intel engineers are purposely looking at Windows guest VM performance running on Linux hosts regardless of the game-oriented benchmark on desktop processors. Most of the time this kind of setup is on the server side with Windows client/Server VMs running on either a VM director, or Linux host.

It makes me wonder how much Intel is concerned about the stability, privacy, and/or security implications of Windows going forward, even in an enterprise SKU environment, cuz I doubt this is a common setup for desktop client systems in the wild when compared to Intel's overall desktop hardware customer market. Granted, that's very tin foil hat (despite the known issues with Windows code quality, ads, and telemetry - most of which aren't present on enterprise SKUs), and it may just reflect one developer's personal interests that may not even be officially sanctioned, but still...

[ms178]

To my knowledge, the original (non-VM) Thread Director patchset hasn't landed in mainline yet and I haven't seen any benchmarks with/without it yet, but hope that it yields some improvements. There was also some related work around core topology by Thomas Gleixner. Considering that Alder Lake has been in the wild for a while, it is a bit surprising that all this work takes such a long time to materialize. While it is a hard task, their army of engineers were able to implement this feature for Windows on time after all. Maybe Linux simply wasn't a priority as no server core went with such a P/E-design.

[RealNC]

E-cores are a waste of silicon on desktops. Why are we even going down this road? That stuff is only relevant on battery-powered devices.

[and.elf]

E-cores are a waste of silicon on desktops. Why are we even going down this road? That stuff is only relevant on battery-powered devices.

They can say the cpus have more cores than the AMD equivalent..

[Joe2021]

E-cores are a waste of silicon on desktops. Why are we even going down this road? That stuff is only relevant on battery-powered devices.

Quite shortsighted POV. First, you are not distinguishing between the concept of big.LITTLe and the specific implementation of intel nowadays, and secondly even today there are many workloads where E-cores shine (on the desktop).

[Kjell]

there are many workloads where E-cores shine (on the desktop).

Shine how?

E-Cores are a joke.

It's a cheap attempt of trying to catch up with the superior power efficiency of AMD.

Keep coping

[espi]

E-cores are a waste of silicon on desktops. Why are we even going down this road? That stuff is only relevant on battery-powered devices.

E-cores are denser than P-cores (0.5x performance for 0.25x area, you can put 4 E-cores in the space of a single P-core, with double the mt performance). So for multithreading they are better than cramming more P-cores.

In fact I think there is very little point to having more than 8 P-cores at all. Any workload that scales to more than 8 cores probably scales to n-cores. Outside of gaming, Intel could get away with only 4 or 2 P-cores even.

The weird part is that E-cores aren't actually all that efficient, they barely beat the P-cores on some workloads, while on other they might even be less efficient.

[ms178]

Shine how?

E-Cores are a joke.

It's a cheap attempt of trying to catch up with the superior power efficiency of AMD.

Keep coping

There are people that care about more than gaming and while the current implementation might be lackluster there are applications and workloads where E-Cores do help, just look at the usual compilation and rendering benchmarks, you'll see the 14700K well above 7900X which are similar in price. Of course I'd like to see benchmarks with AMD's implementation of their compact cores but for some reason they still haven't released such a SKU yet.

I want a great AVX-512 implementation and 8P + 12C cores that just work everywhere as intended. Neither company wants to give me that at the moment.

[geerge]

P cores are the waste of silicon IMO, why do the top parts have 8 of them? It probably makes sense down the line to reduce P cores to 4/2/1/0 to increase the number of E cores.