Talk:Performance impact of Mesa glthread

From Linux Games Wiki
Jump to navigation Jump to search

Do we need to put stuff here which gets no impact? I think it's better to remove those entries. Shmerl (talk) 01:38, 11 July 2017 (UTC)

IMHO It's worth putting those games in here - a game that gets no impact will inform it's been tested. Unless you fear that it will prevent someone from testing it further. I wouldn't worry with the page getting larger, especially as the table is sortable :) Faalagorn/ 00:48, 12 July 2017 (UTC)

Improvements to the table

I have to yet do the tests to contribute myself (switched to RX 580 few days ago), but I've started to made some minor improvements to the table I thought are beneficial overall – so far I've swapped places of executable name with tester name, as I think it makes more sense to have the executable name near the game name (and the tester name near the date of test, too – and have game info separated from tester info). I may do some more minor fixes for clarity to satiate my OCD (which shouldn't affect anything), including unification of card names (which should help with table sorting), but I have a few questions to ask:

  • Shouldn't the games that were whitelisted be appropriately marked, similar to how games in Games broken on Mesa are crossed when fixed? Maybe just with something different than crossing, since that would suggest the test results being invalid. The tables at Games broken on Mesa are far from being perfect.
  • Should the operating system be listed at all? Results template suggest stating only the mesa version yet all the tests include it. If yes, then maybe it's best to split it (see the last point)?
  • Shouldn't GPU memory be stated in GPU names as well? E.g. Radeon RX 580 4 GB vs Radeon RX 580 8 GB – after all the extra memory can have influence on some titles plus some cards can have other differences between the amount of memory that come along it.
  • And finally, wouldn't it be better to have separate columns for GPU and CPU? Assuming CPU can be influential here (which make sense, after all it could be heavy bottleneck in some setups) That way it would be possible to group the same CPU alone without grouping the cards by sorting the table.

Faalagorn/ 20:44, 11 December 2017 (UTC)