I’m not 100 % sure how it exactly works, but I think Microsoft recompiles/translates the games and you then download the changed binary instead of playing off your disc (which is also why texture streaming should be a lot faster).

This is most likely a process that’s automated for the most part though. And I highly doubt it’s recompiled from source, that’s why I called it “translated”.

Games using the id tech engine were often affected by visible texture pop in, and apparently the PS3 version was affected more than the 360 version, but the latter still was noticeably affected. Rage uses id tech 5, but I remember playing BRINK (id tech 4) on PS3 which had no mandatory install (it ran from the disc without installing anything to the HDD upfront), but used the HDD extensively for caching texture data. After I upgraded from the standard 5400 rpm HDD to a 7200 rpm HDD I remember texture pop-in was noticeably reduced.

Xbox 360 emulation on Xbox One or Series isn’t really accurately emulating the hardware, instead it translates the original code to something the One and Series understand.

So it’s just a worse Analogue Pocket for almost the same price…?

The rpm seems to be Fedora (and maybe RHEL) only, it has broken dependencies on openSUSE Tumbleweed.

Is this app anything else than just an Electron wrapper for the web app?

git commit --amend

git push --force

I don’t have the money, can I kiss you twice instead?

Today’s CPUs usually expose some USB connections directly. Ryzen 7000 desktop CPUs expose 4 USB 3.2 Gen 2 directly on their on-package I/O die for example. So if you connect your USB drives directly to the ports your mainboard connects directly to the CPU, the chipset (“southbridge”) and any third-party USB controllers are out of the equation.

This is just information, I’m not advising to use USB for fixed storage.

I was in a similar situation at the end of last year losing all WhatsApp chats dating back to ~2016.

I got a new iPhone 15 Pro (and didn’t keep it, but that’s another story), wanting to upgrade from my 13 Pro.

Now normally, I just transfer all data from the old phone to the new phone which always worked fine. But WhatsApp has its own backup mechanism, which normally works in addition to the regular iCloud backup. But a year ago or two they introduced end-to-end encrypted backups, which I enabled. Apparently this blacklists WhatsApp chats from being backed up to iCloud as part of the standard iPhone backup (even when WhatsApp is explicitly enabled in the backup settings) and also blacklists them from being transferred via direct phone transfer for some reason only some developer at Meta knows.

So, that meant the new chats weren’t on the new iPhone 15, but I didn’t know that yet. I then launched WhatsApp on the iPhone 15, going through verifying my phone number again, which is the normal procedure. It then asked me to generate a new encryption phrase, and this should have been the first sign something went wrong. Basic WhatsApp configuration apparently transferred just fine, which is why WhatsApp never asked to restore the backup it made. This was my first failure, as I didn’t suspect anything and simply set a new encryption phrase. Sure enough, WhatsApp wiped the existing (WhatsApp specific) backup off of iCloud and started a new (empty) backup. Apparently WhatsApp is designed to only support one backup per iCloud account/phone number.

The “old” iPhone conveniently asked if I wanted to factory reset/wipe it after the transfer was successful. This was my second failure, as I simply confirmed the prompt, thinking the transfer was completed (which it was, except for the WhatsApp chats) and also thinking I had the iCloud backup of the phone (which I had, but that excluded the chats as well).

My third failure then was not having any local backup of the phone. I thought I had a backup that was at most a few weeks old on my Mac, but apparently I deleted it and I also moved to a new NAS recently and didn’t transfer the Time Machine backup of the Mac, so it wasn’t in that backup either (I wiped the old NAS).

So it was a chain of UX nightmares/stupid design on Meta’s part and my own stupidity on multiple occasions.

I somewhat cared that my chats were gone as they were a decent database for actually useful stuff and WhatsApp is searchable quite fast, so I frequently used it to find older stuff like links, photos or documents. There were some more personal chats that I’d have like to keep as well. I actually got some chat history back by asking one guy I share a lot of groups with to export their shared group chats and send them to me. I also got a more personal chat back from a person I’m close with.

Most chats are still missing and gone for good, but I’m mostly over that as I shouldn’t live in the past anyway. I also tried contacting Apple to restore iCloud to an older state, but as the WhatsApp backup isn’t on the actual user-facing “iCloud Drive” there’s no way to recover this data.

What bothered me more back then and sometimes still bothers me is how my own mistakes contributed to losing all my chat history. Just not fucking up one single part of this would’ve resulted in me having a working backup of all chats/a working live version on the old phone. It was completely unnecessary to instantly wipe the old phone for example. I absolutely hate fucking something up that I’m not able to fix/do anything about.

One thing I can tell you is to focus on what lessons you take away from this. What I took from this is to of course be more careful, but also to not trust proprietary/cloud-based backups. In the back of my head I always wanted to backup my iCloud Photo Library locally to my NAS, but I never did it. I searched for an app that automatically backs up original versions of all photos/videos to my Synology NAS and I now have a regular, automated, append-only backup of all my iCloud photos. I backup personal data from my NAS to an external SSD weekly, and have a separate cloud backup of the most important data running every night.

Then there’s also time. Time lets you get past most things. Sure, you’ll probably think back once in a while and think “oh damn”, but then you’ll move on the next second and it’ll be fine. Trust me.

“ASRock” and “ASRock Rack” are two different series of motherboards.

Here’s the QVL of one of their AM5 mainboards: https://www.asrockrack.com/general/productdetail.asp?Model=B650D4U-2L2T/BCM#Memory - it doesn’t limit these modules to specific CPUs. All CPUs with ECC compatibility also support these modules on this mainboard. Some of these Rack boards are over a year old, and they always had some ECC modules on their QVL. This - again - isn’t EPYC 4004 specific, they couldn’t have validated it with EPYC 4004 CPUs a year ago. In fact, their CPU support list doesn’t even list EPYC 4004 CPUs as of today, as they haven’t released a BIOS update adding (official) compatibility in for these CPUs (it will probably be released shortly though).

ASRock Rack AM4 mainboards also officially support ECC memory. So if you wanted verified ECC support on a comparatively cheap AMD platform you could’ve always gone for one of these boards with a regular Ryzen CPU (not an APU). The boards are a bit on the expensive side but if you want official support (for whatever reason you’d need that in a homelab environment) you can get it.

The server/workstation focused ASRock Rack AM5 mainboards list plenty of ECC modules in their QVL. The “gaming-focused” ASUS B650E-E I’m using even has two ECC modules listed in its QVL.

So you could’ve already gotten verified ECC support, the fact that the same CPUs now exist with a different (EPYC) branding doesn’t change that. Finding these mainboards isn’t particularly tricky either.

Could’ve just gotten a Ryzen then. These Epycs are essentially relabeled Ryzen CPUs.

Many boards support ECC even when not mentioned. Most ASUS and ASRock boards do for example.

PCIe is back- and forwards-compatible, so even an RTX 4090 would in theory work in a PCIe 2.0 x4 slot.

Now, PCIe 2.0 x4 is obviously really slow compared to what most current cards support (PCIe 4.0 x16), but I doubt transcoding is bandwidth sensitive on the PCIe link.

So pick pretty much any card, be wary though that some consumer-focused cards artificially limit the amount of concurrent transcoding sessions they support. Seems like Nvidia is limiting consumer cards to now 8 sessions, which is probably plenty.

I tried the odd distro here and there over the years on my old laptop/“secondary” device or whatever you want to call it, and while you get some experience with the different desktop environments and whatnot, I can’t fully replicate what I’m doing with my main computer on my older laptop. Gaming is a big thing for me for example, and my laptop has an old GeForce MX250 which doesn’t cut it for today’s games.

What I ended up doing to really give desktop Linux a try was installing it onto a separate SSD on my main machine. You don’t need a separate SSD of course, you could just repartition your existing one and install a Linux distro side by side with your existing Windows installation.

And then I just used my computer (or attempted to) for whatever I would’ve done on Windows at that time. So if I wanted to play a game, I tried installing it (via Steam or Bottles for example) and checked whether it worked. Same with voice chat, screen sharing, development stuff, photo editing, media consumption etc. The few times I booted up Windows again was to update the firmware of a game controller and to transfer save games I hadn’t backed up elsewhere (my Windows partition is BitLocker encrypted and while you can certainly mount it under Linux, I didn’t feel like it, rebooting was much quicker).

Sure, using an old device can work for checking out some things, but at least for me I wasn’t sure at all until I tried doing stuff on Linux that I also did on Windows. YMMV.

Or better yet try it on your main notebook/desktop. Try to get the same things done you did before with Windows and if it works for you, stick with it.

Like others said an Intel CPU with iGPU, alternatively the cheapest Intel Arc GPU (A380?) supports the latest spec of Intel QSV as well.

That’s what started it all for me tbh :D

I also recommend a stable/LTS distro like Debian or AlmaLinux (or other RHEL-based distros). Or just keep using Ubuntu Server LTS.

The OS packages being hopelessly outdated doesn’t really matter when you’re running most services inside containers.

Couldn’t agree more. Wireguard isn’t really hard once you understand the basics.

As others said, “Websocket Support” enables support for them and is required for some applications. “Cache Assets” caches (likely static) assets in the proxy so they don’t have to be loaded from the backend service - I’d leave this disabled unless the backend service is hosted on another network entirely, and even then only enable it if you know the implications. “Block Common Exploits” is a very primitive filter against SQL injection (and similar) attacks. It also blocks some user agents. I wouldn’t enable it as it won’t do much to block a dedicated attacker and some filters may falsely trigger in edge cases, causing errors.