15
votes
Three weeks of Steam Deck game compatibility data
I've been checking in each Friday since the release of the Steam Deck to see the number of games that have been added to the Deck's different compatibility categories. I felt like it was a bit past time to keep bumping the release thread, so I went with a new topic.
Here's where we're at currently:
| 2022-02-25 | 2022-03-04 | 2022-03-11 | 2022-03-18 | Week 1 Change | Week 2 Change | Week 3 Change | |
|---|---|---|---|---|---|---|---|
| Deck Verified | 433 | 535 | 721 | 798 | +102 | +186 | +77 |
| Deck Playable | 398 | 471 | 580 | 678 | +73 | +109 | +98 |
| Deck Unsupported | 389 | 711 | 775 | 837 | +313 | +64 | +62 |
| Steam Total Games | N/A | 67,165 | 67,399 | 67,627 | N/A | +234 | +228 |
Concerns have been raised by Linus of LTT that "verified" is worthless after about five hours of gameplay. I'm paraphrasing of course, but I do believe that QC is going to be a very long tail issue.
What happens after five hours?
He experienced frequent crashing in one of the Horizon titles at least. The concern is that Valve can't afford the resources to properly playtest and do QC on all of the verified titles, so further in to games, instability and bugs are going unnoticed, even on verified titles.
It makes sense. There are issues in programs of any stripe that don't bite you until the program has been running for a while. Memory leaks, issues with the data cache, subtle data corruption, etc.
Yeah. Even the studios and publishers' play testing and QC struggles with long runtime bugs. Linus wondered if publishers would raise a fit with their games being branded as "verified" by Valve when the experience may not be as advertised on other platforms.
I would like to see publishers step up and support the steam deck but... So I am curious to see what the "verified" category looks like a year down the road.
The fact that long runtime issues are so common among so many games strikes me as a good thing. They don't have these problems running native - it's the same game code. That means a flaw (or more likely, a class of flaws) in the emulation layer. That means it can probably be fixed with a careful cataloging and patching of the cracks there rather than trying to sort out god knows how many games individually. If the publishers chip in a little help, Proton could come a very long way in that year.
You get past where the QA team did
I see. But isn't 5 hours, like, enough? Do console plataform test for more than that?
Depends what game it is and how they test. It seems like there are a lot of games that you can't beat in five hours, and sometimes higher levels use different code than lower ones?
Doing a playthrough to the end of the game seems important, just to make sure it's winnable. I wonder how they handle that, if at all?
Maybe test for 5 hours, but not sequentially? Like get saves for different spots and test 1 hour each. IDK. I know nothing about this. Just thinking out loud.
Yeah, that and I assume cheat codes are added to games for QA's use. But part of this is knowing the game really well, which the original QA testers probably did, but maybe not Steam's testers?