This is how xdk-3911 runtime does it, but it seems to result in half-width rendering at the moment (possibly due to something else assuming that multisampling and supersampling are the same).
#CXBX RELOADED XBOX ONE CONTROLLER CODE#
The code contains a code path checking for X_D3DMULTISAMPLE_SAMPLING_MULTI sampling type and divides scaling by 2 if that is the case. Scaling factors should be recalculated when binding a render target, and should be derived from multisampling type only when binding the back buffer.Scaling should be performed on Clear calls too.Introduces those changes to MSAA scaling: MSAA improvements basing on xdk-3911 behaviour With this change, NASCAR Heat 2002 progressed from being completely broken to reaching in-game for the first time. The existing implementation was shared with xdk-4721 with a wrong assumption that both functions are the same, but they were not. This variation puts Stage on the stack and pTexture in eax. Group S Challenge can now reach the menusĪdd several D3D8LTCG function variations for xdk-3911 This PR adds a check for this false positive for a 0x01 rdtsc pattern (triggered also by the offset, as evidenced). Turns out, it's due to a very funny rdtsc false positive which detected the instruction in the middle of a call instruction: E8 0F 31 01 00 Group S Challenge was going ingame for years, but crashed on startup. Finally, resolve symbol scan unable to register kernel thunk addresses properly due to MapThunkTable was called first.Īdd an rdtsc false positive check for Group S Challenge.It should be checking against xbe than system type. Fix hidden xbeType bug for kernel thunk and entry point look up.Now it will be easy to relocate if necessary.
However, ThreadId is filled after the thread has started, and that creates a window for a race if ThreadId is used by the newly spawned thread.
In xbox::PsCreateSystemThreadEx, ThreadHandle is filled immediately after the host thread creation. PsCreateSystemThreadEx: Fill dwThreadId immediately after creating the thread Sound, graphics and performance, there’s something to everyone, but let’s take a look at our numbers before going deep into the changes:Īllow users to disable aspect ratio correction Īs an improvement to the feature implemented last month, this new option allows users to ignore the aspect ratio correction for whatever reason (want your graphics stretched to cover the screen or has some widescreen hack applied? We got you covered) The amount of games that improved this month alone is enough to send chills down my spine (seriously, I hope you have some time to spare reading the results I got…), and a lot of improvements to various parts of the code in general that fixed many long-standing issues. Welcome everyone, to our spooky scary October report! What is so scary about it? The size of this thing to begin with. NOTE: This was written and ready to go by November 2nd, I apologise for the delay posting, life got in the way, anyway, here goes: