[mindcrash]

Framework 13 with the Ryzen AI 5 340 and 2.2K display costs around $900. https://frame.work/laptop13

Add RAM and SSD from a third party, e.g. Amazon, which saves you some money.

Also note that since it's a Framework laptop you can always upgrade the mainboard and display later when you want to.

The cheapest full setup will be around ~1300 (including RAM, storage and all necessary accessories like the expansion cards for external IO).

Also note that if you can set aside ~32 Gb of RAM for virtual machine usage you can run Linux with potentially a GPU accelerated VM containing Windows on this thing which allows you to run native Windows (not simulated, like in WINE) and thus native Windows apps on top of Linux on pretty much bare metal speeds.

[mattbrewsbytes]

Can you elaborate more on your last sentence and running VMs? How does 32GB set aside for VMs and GPU accelerated let you run native Windows? Is there special/newer VM software that is used? Is it still virtualbox?

Outside of docker containers for servers and work I haven't delved into VMs on PCs in a while but this sounds compelling to run linux and windows at once from the same hardware instead of a dual boot.

[mindcrash]

Certainly.

KVM (standard Linux kernel feature) will get you virtualization. On top of that there's a hardware virtualization feature which is called IOMMU (https://en.wikipedia.org/wiki/Input%E2%80%93output_memory_ma...) allowing you to "detach" physical hardware from your host PC and assign it to a VM. This allows you to run a operating system (in this case Windows) inside a virtual machine, but it can use the physical GPU for graphics acceleration because you attached the GPU to it.

When you attach the GPU to it, the VM will also behave like a "real" PC meaning it will discover your graphics card as one of its installed devices and will ask you to install AMD or Nvidia drivers (depending on what hardware you assigned to it) on the virtualized Windows system.

Although it's a bit easier and less tricky on a laptop with 2 GPUs (a integrated GPU embedded with the CPU and a discrete GPU for example) it is also very much possible on a system with a single integrated or discrete GPU.

The only thing you will need to consider on a single GPU configuration is that all GPU acceleration on the host will be gone as long as the GPU is deattached from the host, and it will solely depend on the frame buffer for graphics output (so in this particular case you should not run apps requiring 3D acceleration and the GPU accelerated VM on the host at the same time).

When you have two GPUs you can use one for applications on the host and the other for GPU acceleration in the virtual machine, which means you can run 3D accelerated apps on the host, and a GPU accelerated virtual machine at the same time.

Do note however that you can't run any Windows games with kernel level anti cheat this way, that cat and mouse game has been played and lost. A lot of other use cases will work fine, though.