v0.8.10

Caution

Please see https://github.com/ChefKissInc/NootedRed/discussions/422 about NRed and AppleALC

Enhancements

• Further compatibility work for macOS Mojave.

Bug Fixes

• Fix conflict of new AppleGFXHDA code with AppleALC's.

Note

Use the RELEASE build. RESEARCH_RELEASE is DEBUG with optimisations.
The latter two will be way slower and inefficient than the former build. Use them only when debugging an issue with NootedRed.

Full Changelog: v0.8.9...v0.8.10

v0.8.9

Enhancements

• Added the boot argument -NRedNoAccel for running FB-only.

  Primarily for debug purposes.

• Miscellaneous improvements to the PenguinWizardry SDK.

• Miscellaneous improvements to the GPUDriversAMD SDK.

• X6000FB and AppleGFXHDA compatibility with macOS Mojave.

  However, the Mojave-compatible X6000FB kext and instructions will only be released on v0.9.0!

Bug Fixes

• Added missing call to calcAndSetVrrTimestampInfo in GFX9DCNDisplay implementation.

  This missing piece of code is needed on macOS Big Sur and newer, to tell the accelerator about the display timings.

• Fixed the max display count for Renoir to 4 also in the GFX9DCNDisplay implementation.

  I missed it in v0.8.4.

Note

Use the RELEASE build. RESEARCH_RELEASE is DEBUG with optimisations.
The latter two will be way slower and inefficient than the former build. Use them only when debugging an issue with NootedRed.

Full Changelog: v0.8.8...v0.8.9

v0.8.8

Enhancements

• Moved X6000FB VBIOS fix code out of iVega, into Hotfixes.

Bug Fixes

• Gated getTriageHardwareData behind macOS Big Sur and macOS Monterey.

  It doesn't exist on macOS Catalina. Also replaced usage of scnprintf with snprintf as it may not be available there and cause an Invalid Parameter error when loading the kext.

Note

Use the RELEASE build. RESEARCH_RELEASE is DEBUG with optimisations.
The latter two will be way slower and inefficient than the former build. Use them only when debugging an issue with NootedRed.

Full Changelog: v0.8.7...v0.8.8

v0.8.7

Bug Fixes

• Neutralised setupBootWatermarks in X6000FB.

  This function applies some hardware quirk through the DCN watermarks on specific RDNA 1 dGPUs. On Raven the register offsets are even completely wrong and changing random DCN registers on specific ASIC revisions. So, I just neutralised the function.

• Added custom implementation for getTriageHardwareData (exists in macOS Monterey and below).

  This function prints a few DCN register values in a GPU diagnosis report, however, the offsets aren't quite matching for Raven and Renoir, along with the max display pipe count being 4.

NOTE: The "Research Release" and "Debug" builds will be way slower and inefficient than the "Release" build. Do not use them if not debugging an issue with the software.

Full Changelog: v0.8.6...v0.8.7

v0.8.6

Bug Fixes

• Worked around spurious kernel panic in recovery likely caused by Lilu bug.

NOTE: The "Research Release" and "Debug" builds will be way slower and inefficient than the "Release" build. Do not use them if not debugging an issue with the software.

Full Changelog: v0.8.5...v0.8.6

v0.8.5

Enhancements

• Cleaned up X6000FB branding code, removing the subsystem cruft and the weird "Renoir Graphics D4" and similar models.
• Specialised fallback model to "Radeon RX Raven Graphics", "Radeon RX Picasso Graphics", and "Radeon RX Renoir Graphics".

NOTE: The "Research Release" and "Debug" builds will be way slower and inefficient than the "Release" build. Do not use them if not debugging an issue with the software.

Full Changelog: v0.8.4-2...v0.8.5

v0.8.4

Enhancements

• New code style for improved readability.

• Some small code cleanup and optimisations.

  This includes improving and fixing up the dm_logger_write reimplementation that's enabled with -NRedDebug.

• Replaced useless backtrace on SMU/SMC command errors with plain logs including the command ID.

• Improved the memory type handling inside the X6000FB code.

• Replaced legacy model mechanism with its driver-native counterpart.

Bug Fixes

• Fixed potential out-of-bounds string comparisons throughout the code.

• Fixed the waitForFunc timeout calculation in nanoseconds being implicitly truncated.

• Fixed the saved framebuffer size calculation of the X5000DCNDisplay implementation being implicitly truncated.

• Fixed copy paste error on the ARGB1555 surface format case of the X5000DCNDisplay implementation.

• Fixed the display pipe count on Renoir-based ASICs.

  Renoir and Raven both have only 4 display pipes. Not sure where I got the previous assumption of 6 for Renoir, maybe accidentally read from DCN 2.0 (Navi) instead of 2.1 (Renoir).

NOTE: The "Research Release" and "Debug" builds will be way slower and inefficient than the "Release" build. Do not use them if not debugging an issue with the software.

Full Changelog: v0.8.3...v0.8.4-2

v0.8.3

Enhancements

• Removed redundant GPUDCCDisplayable setting.
• Removed unnecessary PP_ToolsLogSpaceSize setting. This setting is only applicable for SMU 11.
• Removed redundant PP_WorkLoadPolicyMask setting. The value specified was 0, which is already the default.
• Removed IOPCITunnelCompatible property.
  iGPUs can't run through Thunderbolt.
• Moved SMU 12 firmware wait to separate function.
• Some TTL event handler cleanup.
• Miscellaneous firmware constant cleanup.

Bug Fixes

• Rechecked driver settings and adapted them correctly to AMD Adrenaline's.
• Added missing DalForceSingleDispPipeSplit to Picasso settings.
• Removed Raven2 RLC A0 firmware.
• Activate A0 RLC firmware only on non-Picasso or Picasso AM4.
• Disable MMHub PG as per the driver settings.
• Added missing GPUTaskSingleChannel accelerator setting.
  Possibly added in Sonoma or something, non-existent in Big Sur, default false, but set to true in the accelerator kexts (Navi and Vega).
• Corrected "wait for" logic.
  Corrected timeout to PP default of 2s from Linux's 100s and used IODelay instead of IOSleep.

NOTE: The "Research Release" and "Debug" builds will be way slower and inefficient than the "Release" build. Do not use them if not debugging an issue with the software.

Full Changelog: v0.8.2...v0.8.3

v0.8.2

Enhancements

• Code clean-up.
• Add back the -NRedOff, -NRedBeta boot arguments.

Bug Fixes

• Fix cut off strings in stringifyCRTHWDepth.

NOTE: The "Research Release" and "Debug" builds will be way slower and inefficient than the "Release" build. Do not use them if not debugging an issue with the software.

Full Changelog: v0.8.1...v0.8.2

v0.8.1

Enhancements

• Code clean-up.
• Added debug prints for pixel mode, format, depth for display issue debugging.

Bug Fixes

• Fixed AUX/OLED backlight code always returning failure.
• Fixed AUX/OLED brightness transition time (15 seconds -> 50 milliseconds).
• Fixed the addrlib setting depthPipeXorDisable being set to true on all Raven/Non-Renoir even Raven2 instead of Raven1, which was incorrect.

NOTE: The "Research Release" and "Debug" builds will be way slower and inefficient than the "Release" build. Do not use them if not debugging an issue with the software.

Full Changelog: v0.8.0...v0.8.1