GSdx-ogl: Drop AMD's SSO workarounds #2696
Conversation
| */ | ||
| if (v < 25.20f || (v < 25.21f && atoi(pEnd + 1) < 1407)) | ||
| amd_buggy_driver = true; | ||
| } |
This comment was marked as spam.
This comment was marked as spam.
Sorry, something went wrong.
This comment was marked as spam.
This comment was marked as spam.
Sorry, something went wrong.
| #endif | ||
| /*if (v < 15.3f) | ||
| amd_legacy_buggy_driver = true; | ||
| */ |
This comment was marked as spam.
This comment was marked as spam.
Sorry, something went wrong.
This comment was marked as spam.
This comment was marked as spam.
Sorry, something went wrong.
This comment was marked as spam.
This comment was marked as spam.
Sorry, something went wrong.
lightningterror
changed the title
| #ifdef _WIN32 | ||
| GLfloat v = _strtof_l(strrchr((char*)s, ' '), &pEnd, _create_locale(LC_NUMERIC, "C")); | ||
| #else | ||
| GLfloat v = strtof_l(strrchr((char*)s, ' '), &pEnd, newlocale(LC_NUMERIC, "C", NULL)); |
This comment was marked as spam.
This comment was marked as spam.
Sorry, something went wrong.
This comment was marked as spam.
This comment was marked as spam.
Sorry, something went wrong.
This comment was marked as spam.
This comment was marked as spam.
Sorry, something went wrong.
This comment was marked as spam.
This comment was marked as spam.
Sorry, something went wrong.
This comment was marked as spam.
This comment was marked as spam.
Sorry, something went wrong.
This comment was marked as spam.
This comment was marked as spam.
Sorry, something went wrong.
This comment was marked as spam.
This comment was marked as spam.
Sorry, something went wrong.
| // Unfortuntately after testing for at least these 2 cases we check, I found that had same | ||
| // versions across the buggy and fixed "branch". Threfore look for the kernel driver build. | ||
| // | ||
| // Screw Wx screwings (it's nice we have decimal seprator localized... but dlls do not) |
This comment was marked as spam.
This comment was marked as spam.
Sorry, something went wrong.
This comment was marked as spam.
This comment was marked as spam.
Sorry, something went wrong.
Also account for latest release bug fixes (see PCSX2#1552 for details on the used values) Legacy_driver variable was left there, because technically speaking that'd be supposed to be yet another case to remotely care
|
I don't see any harm in leaving the workaround there for a bit longer. If we remove it for 1.6 release and AMD breaks the driver again then we'll end up with a not working stable release. |
Forsaken float method for int, generic C, one (thanks @turtleli)
|
Here we are with another spin (RIP "first decimal digit" check) I'm a bit skeptical about sanitization of the arrays (e.g. what happens when GL_version is missing, or hasn't spaces?) so there's that. |
| // Coincidentally the same position is also where Intel and Nvidia put their only version id | ||
| bool check_driver_version(const unsigned char *s, std::string v) { | ||
| std::stringstream iss(v); | ||
| std::vector<int> fixed_version((std::istream_iterator<int>(iss)), std::istream_iterator<int>()); |
This comment was marked as spam.
This comment was marked as spam.
Sorry, something went wrong.
This comment was marked as spam.
This comment was marked as spam.
Sorry, something went wrong.
This comment was marked as spam.
This comment was marked as spam.
Sorry, something went wrong.
|
Needs rebase. |
|
Status? |
|
I'm sure somebody wittier than me will be able to handle it. |
Details on the otherwise random-looking values I used at #1552 (comment)
I have some reserve on whether it's me to be stupid, Wx, or both, considering the time I needed to find the right code (for as much ugly) that worked in every case.
And yes, @lightningterror,
amd_legacy_buggy_driveris just there waiting for your June <~15.11 hack to land ^^