New -ish to Garuda, and have played around with Arch for a bit, along with a bunch of Debian-based distros as well, and have been tinkering with Linux in general since the 90s. So far, really enjoying the stability and overall experience. So much so, that Garuda Dr460nized is at the top of my list as a Windows alternative that I'd very much like to shift into full time, and then virtualize W10, as needed. A few things to consider:
I'm not a huge gamer, but I do enjoy the occasional video game here and there. Used to play a lot of ESO, Mass Effect, etc. That said, I do some occasional video editing and photo editing, so to the degree I can take advantage of GPU acceleration here, and transcoding, the better.
I do a LOT of VM work, and while most of the VMs I've worked on have been on Type 2 Hypervisors (virtualbox/vmware), I plan on taking advantage of KVM/QEMU for Type 1 virtualization as much as I can (likely to run W10 for some proprietary software for work that only runs on windows, and potentially for any games that just aren't supported well on Linux yet). With this in mind, I do want to make sure I can pass-through a dedicated GPU in those conditions, and also make sure I am able to leverage the dedicated GPU on the native Garuda environment as well. Additionally, I do some work using docker containers as well, so figured it deserves a mention here.
I will be using some hardware from my existing desktop, primarily the 3 storage devices. On my current (quite old) build, I have:
Samsung SSD 850 Evo -
500GB: Primary C: drive. This is used almost exclusively for the Windows OS, and any software that cannot be overridden to install onto another HD, so all C:\Program Files, and C:\Program Files (x86). Currently about 27% storage capacity used.
1.5TBHDD: Secondary D: drive. This is used almost exclusively for all installed software, and in cases where I can make symbolic links to hard-coded C:\Program Files or C:\Program Files (x86) software installations - so that the software and OS think the software is installed on the C: drive, but is actually taking up storage on the D: drive. Currently about 13% storage capacity used.
4TBHDD: Tertiary E: drive. This is used almost exclusively for User files. I've already symbolically linked my C:\Users directory to E:\Users, and all the Documents, Downloads, Videos, Photos, etc are saved on this drive. Currently about 19% storage capacity used.
I will likely also cannibalize the optical drive into this new build. The rest of the hardware is still kind of up in the air, and that's where I can use this amazing community's help. I've been reading a lot on the forums on different solutions/ideas to multi-storage device configuration in a linux environment and how that fits with btrfs, etc, and I'm kind of at a loss of where to go from here, and how to, during install, configure the storage schematic. I want to take advantage of the SSD here as much as possible, and leverage its speed for the OS - and has been recommended - as well as for software that will be installed. That leaves the user files (/home) and potentially the /tmp and logging directories to be added to the HDD drives, so as not to use up too many of the Read-Write cycles on the SSD.
From what I think I was able to distill, in broad terms, is that I would use LVM to combine the two HDDs into one volume of 5.5TBs and maintain the SSD as its own volume. Does this make sense? Will this actually work? And if so, what mount points should I dedicate to the SSD volume and which mount points to the combined HDD volumes? If I understood correctly, I think the following should make sense/work:
I feel like I'm definitely missing something and would definitely appreciate some advice and guidance on this. I understand that if I don't specify a specific mount point for a directory or subdirectory, it will all default to wherever "/" is mounted - which would be the SSD. Also, where would the swap partition make the most sense? On the one hand, the speed of the SSD would likely be useful, but might use up too many of the write cycles, and so the swap partition may be better suited on the HDD lvm. Thoughts here would be great.
Then comes the question of which GPU to get. I'm somewhat brand-agnostic here, and would sooner buy a dedicated GPU that is most supported in the linux/Garuda environment, which would also allow me to use it within the context of a pass-through on a type 1 vm. From what I've read, the AMDs are really good, and have open source drivers/kernel modules, whereas the NVidia gpus, while some work, require a lot more tinkering, and don't have open source drivers/kernel modules and are generally a bigger PITA, but perhaps may be a "better" GPU - so, happy to take advice on this as well.
Now, onto the motherboard, cpu and RAM....I'd prefer to purchase a mobo which supports ECC DDR4 udim ram (some of the work I do on windows requires some sensitive computations, and having ECC RAM is very helpful here, but not entirely a deal breaker if there really isn't a compatible mobo that supports ECC). I don't even know where to begin in terms of making sure the kernel can utilize and leverage the ECC ram - while I know it's supported in general since it's most common use-case is in data centers/servers, I imagine it must also be possible on a desktop environment.
Additionally, considering my virtualization needs, is an intel or amd processor better here? From what I've read, KVM virtualization is far superior, and is most supported on intel cpus - but happy to learn more on this. I'm thinking at minimum 32gb of RAM, with the option of being able to expand that over time. Additionally, I'll be using sata for the storage needs (in case that needed to be specified). Nearly all mobos today come with all kinds of other integrated features such as LAN ports, and sometimes integrated GPUs. Nearly all the intel cpus have an integrated gpu, whereas the amds do not - which I guess is fine, since I plan on mostly utilizing the dedicated GPU, but also understand the benefits of the integrated gpu as a failsafe.
Oof...that was a lot and I'm already overwhelmed. I know enough to be dangerous, and can do a lot of googling to fix whatever I may have broken - also part of why I LOVE garuda's timeshift snapshots. Truly an amazing feature which has already saved my test system more times than I care to admit LOL. So, yeah.....I'd humbly appreciate this amazing community's guidance, advice, support and suggestions. Thanks so much in advance and I look forward to corresponding with y'all!
Thank you for the warm welcome! I've already learned so much about Garuda, and in my 3-months of using it exclusively on my oldish laptop - where it's been given a new lease on life - it's really helped me feel more comfortable, and excited, with the idea of switching over to it as my daily driver.
Not at ALL. I'm still going to do all my research and look into it further, but I was hoping to pique the interest of other arch/garuda enthusiasts who, perhaps have more experience than I, might be able to at least alert me to things I wouldn't know to think of (ie, I don't know what I don't know), or might be able to point me in the right direction. Frankly, I'll take what I can get in terms of advice, and I'm hoping () I'll touch on someone's sympathy who'd be willing to help me.
Also, what I think I'm saying is that I'm willing to pursue the hardware where it intersects with what I will primarily be using the build for, within the context of Garuda, specifically, and linux, generally. There are many more knowledgeable than I who would know how best to help me answer those questions, and I'd love to tap into whomever is willing to help.