Cogs, originally released as a humble bundle game http://www.humblebundle.com/ also hosted on Desura http://www.desura.com/games/cogs is not playable on GMA 4500 due to severe fps issue. The bug https://bugs.freedesktop.org/show_bug.cgi?id=40401 is fixed for gallium divers by the commit http://cgit.freedesktop.org/mesa/mesa/commit/?id=ce44bae366ade59fb2dbdfbfe5a1ab8d24518a57 but the issue is still present for Intel graphics cards. Tested on: Ubuntu 11.10 linux 3.2.0-030200-generic X.Org X Server 1.11.2.902 xf86-video-intel 2.17.0 Mesa 8.0-rc2 Same problem for older Mesa versions 7.11.2 and 7.10.3 tested on both 32 and 64 Bit Linux system
still present in the git and stable Mesa
still present in Mesa 8.0.3
Please fix this bug, Only DRI drivers are effected by this bug, even old Intel GMA 950 with gallium driver i915g runs this game with acceptable speed. Cheers
Hello Developers, Please, send me a personal message so that I can give the download link of the game for testing purpose. Best Regards, Imam
Just some info, Windows version of the game uses Direct3D and runs fine with WINE. when monitored with intel_gpu_tops, render busy only showing 20% with Linux version. Cheers
Should be fixed by the series ending with: commit 174d44a9c4d39a030fe3528acf07f9ac9aa617a1 Author: Kenneth Graunke <kenneth@whitecape.org> Date: Wed Aug 22 00:43:14 2012 -0700 mesa: Use a new, more specific hook for shader uniform changes.
Still not fixed at least for Intel GMA 4500 device id 2a42. Checked against: Linux 3.5.0 (64 bit) libdrm-2.4.39 xf86-video-intel 2.20.5 Mesa master 8.1-devel (Mesa DRI Mobile Intel® GM45 Express Chipset) Thank you for reviewing
(In reply to comment #7) > Still not fixed at least for Intel GMA 4500 device id 2a42. > > > > Checked against: > > Linux 3.5.0 (64 bit) > libdrm-2.4.39 > xf86-video-intel 2.20.5 > Mesa master 8.1-devel (Mesa DRI Mobile Intel® GM45 Express Chipset) > > Thank you for reviewing and Linux 3.5.0 (32 bit)
Also slow for Intel HD graphics (device id 0106)
What a spectacularly broken application. They're handing us 900kb of vertex data and using 4-6 verts out of it. We would need to detect the app-being-stupid case and rebase out their indices. It would never be as fast as a good implementation on their side, but should help at least.
Tempted to close as NOTOURBUG.
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.