Beiträge von vit9696

  • CFG Lock

    I would say we will need to rename VerifyMsrE2 to something like MsrE2Ctl and make it accept arguments like -check (and do what VerifyMsrE2 currently does) and -unlock/-lock (and perform your tool logic). Without the arguments it can possibly have some interactive mode.


    A good start will be a PR of this change to OpenCorePkg. Then we could review and bring the changes to the project codestyle and such. For the arguments you can use GetArguments function from Library/OcMiscLib.h.


    Thanks a lot!

  • RadeonBoost.kext - Benchmark Scores wie am echten Mac / unter Windows

    Falcon SMC firmware loading is available on X4000~X6000, but it might actually be unused in some(?) drivers. It is very easy to check by adding an OC patch _TF_PhwSIslands_UploadFirmware → 0F 0B for AMDRadeonX****HWLibs.kext. If the firmware is used, the system will not boot. Otherwise it will crash very early. Probably worth checking to be sure, which cards it affects.

  • RadeonBoost.kext - Benchmark Scores wie am echten Mac / unter Windows

    Yeah, I think that makes very good sense to me, sorry for a slow reply. Not sure if it is a bug, could just be some unimplemented feature, but very plausible.


    Other than that, I noticed that RadeonBoost gives HBM2 memory bandwidth boost, right? Could you tell us which exact parameter causes the boost? We might be able to track down the reasons behind it. Currently in our view we there are only two relevant parameters:

    — PP_WorkLoadPolicyMask

    — ATY,EFIVersion / Force_Load_FalconSMUFW


    But it will be nice to confirm that we miss nothing. Thanks.

  • RadeonBoost.kext - Benchmark Scores wie am echten Mac / unter Windows

    Zitat von Toskache

    vit9696 Very interesting, but how do you explain the better Photoshop performance (pugetbench)? https://forums.macrumors.com/t…st=28377684#post-28377684

    Zitat von CMMChris

    Yup, I had the same thought. In specific areas however, there definitely are performance improvements. Some games (Dirt4, Borderlands 3) run notably smoother with my RadeonBoost kext at least on the Radeon VII and also Photoshop seems to benefit as demonstrated on Macrumors.


    Edit: Neue Version 1.1 im Startpost. AGPM Injector ist nun direkt in RadeonBoost integriert. matpro64 Check if your Vega64 fans run normal now.

    Well, like I said, the two parameters work, others do not as they just seem to be blindly copy-pasted. For Polaris only the policy applies, as it has no SMU. We can retest with this benchmark, but are you sure you have any difference with just the two parameters specified and RadeonBoost?

  • Thunderbolt HOT-PLUG bei Onboard Thunderbolt Chip!

    DSM2 sorry for jumping in, but could you please do clarify why this kind of secrecy is involved in the matter? Is it just that you need some time/will to do a proper write-up for the problem? Or perhaps it is quite a bit complicated and more thought is needed to make it useable by everyone? I believe that we could work it out together =)


    Just, I mean, if neither of these, the situation is awkward at least. If you look around, the development community is not hiding their research from others for obvious reasons. Avarice of this kind sounds like a disgrace for everyone involved in macOS community development, including Download-Fritz and myself. Obviously I did not want to offend you in case I misunderstood something.

  • Funktionierender NVRAM nun mit Clover


    The point is that UEFI does not allow you to use global variables in runtime services. Yet AMI does it @_@.
    So the point of fixing variable functions is to fix a bug in one of the AMI modules.
    While it is terrible that bugs exist in their other modules too, adding preliminary hacks is a bad practice.


    Looking forward for the function list.

  • Funktionierender NVRAM nun mit Clover

    Spoiler anzeigen


    Hallo,


    Could you make a list of the functions that crash and the ones that work fine?
    It may be acceptable to proxy more functions in AptioFix if necessary.
    So far I had to use ResetSystem, and it worked fine for me, but others may indeed crash.


    By the way, do you know whether Apple changed AppleEFIRuntime interface across different versions of macOS? I currently call functions directly via gPEEFIRuntimeServices, but it is safer to invoke AppleEFIRuntime. The only reason stopping me from reversing the header is that they may change it at any time.


    P.S. Sorry, don't know German, hoping for Google Translate.