MSI Afterburner

[George]

RTSS 7.3.6 ბილდი გამოვიდა ერთი 10 დღის უკან შემდეგი ჩეინჯლოგით:

Now the installer is skipping the process of downloading and deploying VC++ redistributables if the previous downloaded copy of redistributables is detected inside .\Redist folder. Installation on top of the previously installed version should be much faster now
· Now installation and uninstallation attempts are blocked until reboot when the previous installer’s or uninstaller’s reboot prompt is ignored
· Improved protection against attempts to run incompletely installed application when installer’s reboot prompt is ignored by user and some pending file operations are scheduled to be performed on the next reboot. Previously it was possible to bypass the protection and still start application in such incompletely installed state with scheduled startup task after user logoff/logon. Also, the protection is now verifying the product integrity and doesn’t allow starting it if some component is missing due to corrupted installation
· Some of previously introduced NVIDIA Reflex injection related options are no longer power oriented and can be changed via “Injection properties” settings group in “General” tab:
o New “Inject NVIDIA Reflex latency markers” option can be used for troubleshooting, you may use it to disable markers injection either globally or for some specific applications
o New “Inject NVIDIA Reflex sleep call” option can be used to control NvAPI_D3D_Sleep injection behavior, you may tweak it to improve injected Reflex efficiency in applications performing simulation and input sampling on rendering thread

ესეც გადმოსაწერი ლინკი: https://download-eu2.guru3d.com/rtss/[Guru3D]-RTSSSetup736Build28004.rar

[George]

RTSS 7.3.6 ახალი ბილდი გამოვიდა შემდეგი ჩეინჯლოგით:

 Now the installer and uninistaller are performing active application instance checking on earlier stages before displaying any GUI
· Now the installer is performing locked files testing and displays the list of third-party processes, which may lock Vulkan layer and hook libraries
· Now the installer is skipping the process of DirectX web installation when required d3dx10/11 runtimes are detected
· Now OverlayEditor plugin is forcibly uninstalling IO driver on OS shutdown to prevent the driver from being hibernated and reloaded due to fastboot till the next clean system boot event
· Now OverlayEditor plugin allows you to override CPU bus clock with “bclk” environment variable. You may use it to manually correct CPU clock readings on the systems where Intel OC mailbox and precise bus clock monitoring are not available

- Now RivaTuner Statistics Server defaults to hooking original OS version of OpenGL32.dll if both original and third party proxy versions of OpenGL32.dll runtime are loaded in context of the game process. This change is aimed to improve compatibility with ReShade 6.0.x in some OpenGL games like “Wolfenstein : The New Order”. You can still prioritize OpenGL32.dll proxy hooking via enabling “Enable Custom Direct3D” option at application profile level, which previously controlled Direct3D runtime proxy hooking behavior only, but now it also controls OpenGL runtime proxy hooking strategy

სულ მალე ფინალური ბილდიც გამოვა ამ დღეებში უკვე გურუ3დ-ზე, ესეც მაგ ბილდის გადმოსაწერი ლინკი:

 

https://download-eu2.guru3d.com/rtss/[Guru3D.com]-RTSSSetup736Build28010.rar

[George]

RTSS 7.3.6 ფინალური ბილდი გამოვიდა შემდეგი ჩეინჯლოგით:

 Forced Detours hooking mode for ID3D11DeviceContext::ClearRenderTargetView hook to prevent crashing on the final stage of track load in “Le Mans Ultimate”
· Extra synchronization added to Detours uninstallation implementation in the previous version is now active inside ID3D12CommanQueue::ExecuteCommandLists hook on application shutdown only to avoid performance issues with “Elden Ring”
· Now RivaTuner Statistics Server defaults to hooking original OS version of OpenGL32.dll if both original and third party proxy versions of OpenGL32.dll runtime are loaded in context of the game process. This change is aimed to improve compatibility with ReShade 6.0.x in some OpenGL games like “Wolfenstein : The New Order”. You can still prioritize OpenGL32.dll proxy hooking via enabling “Enable Custom Direct3D” option at application profile level, which previously controlled Direct3D runtime proxy hooking behavior only, but now it also controls OpenGL runtime proxy hooking strategy
· Now the installer and uninistaller are performing active application instance checking on earlier stages before displaying any GUI
· Now the installer is performing locked files testing and displays the list of third-party processes, which may lock Vulkan layer and hook libraries
· Now the installer is skipping the process of downloading and deploying VC++ redistributables if the previous downloaded copy of redistributables is detected inside .\Redist folder. Installation on top of the previously installed version should be much faster now
· Now the installer is skipping the process of DirectX web installation when required d3dx10/11 runtimes are detected
· Now installation and uninstallation attempts are blocked until reboot when the previous installer’s or uninstaller’s reboot prompt is ignored
· Improved protection against attempts to run incompletely installed application when installer’s reboot prompt is ignored by user and some pending file operations are scheduled to be performed on the next reboot. Previously it was possible to bypass the protection and still start application in such incompletely installed state with scheduled startup task after user logoff/logon. Also, the protection is now verifying the product integrity and doesn’t allow starting it if some component is missing due to corrupted installation
· Now OverlayEditor plugin is forcibly uninstalling IO driver on OS shutdown to prevent the driver from being hibernated and reloaded due to fast startup till the next clean system boot event
· Now OverlayEditor plugin allows you to override CPU bus clock with “bclk” environment variable. You may use it to manually correct CPU clock readings on the systems where Intel OC mailbox and precise bus clock monitoring are not available
· Some of previously introduced NVIDIA Reflex injection related options are no longer power oriented and can be changed via “Injection properties” settings group in “General” tab:
o New “Inject NVIDIA Reflex latency markers” option can be used for troubleshooting, you may use it to disable markers injection either globally or for some specific applications
o New “Inject NVIDIA Reflex sleep call” option can be used to control NvAPI_D3D_Sleep injection behavior, you may tweak it to improve injected Reflex efficiency in applications performing simulation and input sampling on rendering thread
· Switched to alternate digital signature

www.guru3d.com/download/rtss-rivatuner-statistics-server-download https://www.guru3d.com/download/rtss-rivatuner-statistics-server-download/

[George]

 

[George]

 

მალე გამოვა ახალი RTSS 7.3.7 ბეტა ბილდი დიდი ჩეინჯლოგით, ნაწილი მაგ ვიდეოშია აღწერილი რეებიცაა დამატებული ახალში

[George]

 RTSS 7.3.7 ბეტა ბილდი გამოვიდა შემდეგ ჩეინჯლოგით:

· Improved hypertext parser:
o Now bar range graphs are rendered when the range width is less than one pixel
· Added late initialization of Vulkan device function pointers aimed to improve Vulkan hooking reliability in the applications dynamically loading vulkan-1.dll on the systems with multiple installed third party Vulkan layers
· Improved OverlayEditor plugin:
o Improved NVIDIA Reflex metrics access functionality. Previously NVIDIA Reflex metrics were represented by hardcoded internal "NVIDIA Reflex XXX timespan" and "NVIDIA Reflex XXX latency" groups of data sources. Such internal data sources are deprecated since this version, now you can use more robust functions for accessing NVIDIA Reflex data:
§ New reflexlatency(markerFrom, markerTo) function can be used to calculate a latency between two arbitrary NVIDIA Reflex latency markers related to the current frame. You can specify one of the following marker identifiers: 0 – input_sample, 1 – simulation_start, 2 – simulation_end, 3 – rendersubmit_start, 4 – rendersubmit_end, 5 – present_start, 6 – present_end, 7 – driver_start, 8 – driver_end, 9 – osrenderqueue_start, 10 – osrenderqueue_end, 11 – gpurender_start, 12 – gpurender_end, 13 – virtual gpurender_start (gpurender_end minus GPU active time). You may also set markerFrom to -1 to specify the very first marker and set markerTo to -1 to specify the very last marker if necessary. This function allows you to calculate all NVIDIA Reflex latencies previously represented by hardcoded internal "NVIDIA Reflex XXX latency" data sources and implement additional custom latencies (e.g. a latency between simulation_start and present_start markers)
§ New reflextimespan(markerFrom, markerTo, base) function can be used to calculate and encode a timespan between two arbitrary NVIDIA Reflex latency markers related to the current frame. This function allows you to calculate all NVIDIA Reflex timespans previously represented by hardcoded internal "NVIDIA Reflex XXX timespan" data sources and implement additional custom timespans (e.g. a timespan between simulation_start and present_start markers). Also the previous "NVIDIA Reflex XXX timespan" data sources calculated timespan using fixed base 100 (meaning that the timeband graphs visualizing the timespan had just 100 gradations), now you may specify custom base to implement more detailed timebands
§ New reflextime(id) function can be used to access absolute times reported by NVIDIA reflex for the current frame. You may set time identifier to 0 to get GPU active or 1 to get GPU frametime
o Built-in reflex.ovl overlay layout has been modified to demonstrate new reflextime/reflextimespan/reflexlatency functions usage instead of deprecated "NVIDIA Reflex XXX timespan" and "NVIDIA Reflex XXX latency" data sources. Please apply similar modifications to your custom overlays, if you used deprecated data sources there
o Updated helper PresentMonDataProvider.exe application:
§ Added PresentMon V2 service support. Please take a note that due to major changes between PresentMon V1 and V2 service API, service API V1 is no longer supported
§ Added PresentMon V2 console application support. Unlike the service, console mode is backward compatible with the previous versions of PresentMon, so you may switch to the previous versions of console PresentMon if necessary via the configuration file. The most recent versions of console PresentMon V1 and V2 applications are included into distributive
§ Added configuration file setting allowing power users to specify additional command line parameters for console mode (e.g. -v1_metrics for forcing condole PresentMon V2 to report V1 metrics or -track_input for enabling experimental input tracking functionality in PresentMon V1)
o Updated PresentMon data provider:
§ Added PresentMon V2 metrics support. The most of PresentMon V1 metrics are also reconstructed in realtime from V2 metrics for your convenience and for backward compatibility with previously existing PresentMon based overlay layouts. Please take a note that PresentMode metric encoding changed between V1 and V2 versions of PresentMon service API, so now OverlayEditor use new encoding too. All PresentMon based overlay layouts have been updated to reflect new PresentMode encoding. Please apply the same changes to your overlays if you decode PresentMode there
§ PresentMon V1 service API specific hardware telemetry is no longer available in PresentMon data provider
o Added presentmon_frame_scale_V2.ovl overlay layout:
§ The layout demonstrates usage of PresentMon V2 specific data sources and keyboard based overlay transformations. You may press <F> to switch between present-to-present (PresentMon V1) and start-to-start (PresentMon V2) frametime sampling modes in realtime
§ The layout demonstrates PresentMon V2 ClickToPhotonLatency data source usage to render input events graph and calculate averaged latency for the last 512 frames
· Bundled DesktopOverlayHost has been upgraded to v1.4.0:
o Added Direct3D10, Direct3D12 and Vulkan rendering backends. New backends do not affect the primary application’s functionality, however now you may use DesktopOverlayHost as a built-in overlay rendering test for all supported 3D APIs in x86 applications
o Added optional 64-bit version of DesktopOverlayHost. You may use it as a built-in overlay rendering test for all supported 3D APIs in x64 applications
o Added power user oriented config file switch allowing using idle based rendering loop instead of default timer based rendering loop
o Updated DesktopOverlayHostLoader helper application can launch both 32-bit and 64-bit versions of DesktopOverlayHost
· Updated profiles list

ესეც გადმოსაწერი ლინკი: https://download-eu2.guru3d.com/rtss/[Guru3d]-RTSSSetup737Beta1Build28020.rar

[George]

 

დაიდება მალე 7.3.7 ბეტა 2 ბილდიც ახალი ცვლილებებით, ნაწილი ვიდეოშია აღწერილი, ხო რაც შეეხება DWM და explorer ლატენტურობას და ლაგებს, ლაგების ნაწილი მცირდება nvidia კონტროლ პანელში adaptive-ან მაგის გამშვებებს თუ მოუნიშნავ maximum performance-ზე, მარტო გლობალზე დაყენება მაგისი არ შველის, ცალკეც უნდა მოუნიშნო, ვნახე და მაინც adaptive-ზე ეყენა ორივე, ამცირებს ეგ ხშირათ ლაგებს ბრაუზინგშიც და ყველგან სადაც გამოიყენება ეგ ორი სერვისი

[George]

RTSS 7.3.7 ბეტა 2 გამოვიდა შემდეგი ჩეინჯლოგით:

o As I promised I included my custom version of PresentMon-2.0.0-x64.exe in this build to address regress with data streaming/buffering efficiency. Intel also fixed it in future versions of console PresentMon, but no new version is available yet to I included my own custom build as a temporary solution. This should address issue with increased CPU load in PresentMonDataProvider.exe
o Now you may press <Ctrl>+<A> inside the editor window to select all overlay layers or press <Ctrl>+<Shift>+<A> to deselect all selected layers
o Now new layout creation command deletes all data sources and text tables, instead of deleting just overlay layers before
o Save prompt is now displayed when you try to load new overlay layout without saving the previously modified overlay
o Added new “Merge” command to “Layouts” menu. Now you can easily combine multiple overlay layouts or copy data sources, layers or text tables between the layouts. All layers of source layout are automatically selected after merging to allow you to move them to new position with a single mouse gesture
o Now PresentMon data provider calculates timestamps of different stages of rendering pipeline using PresentMon V2 metrics, when at least one PresentMon data source is added to layout, and allows accessing PresentMon timestamps in a way similar to accessing NVIDIA Reflex latency markers:
§ New presentmonlatency(markerFrom, markerTo) function can be used to calculate a latency between two arbitrary PresentMon latency markers related to the current frame. You can specify one of the following marker identifiers: 0 – input_sample, 1 – cpu_busy_start, 2 – cpu_busy_end, 3 – present_start, 4 – present_end, 5 – gpurender_start, 6 – gpurender_end, 7 – virtual gpurender_start (gpurender_end minus GPU active time), 8 - display. You may also set markerFrom to -1 to specify the very first marker and set markerTo to -1 to specify the very last marker if necessary
§ New presentmontimespan(markerFrom, markerTo, base) function can be used to calculate and encode a timespan between two arbitrary PresentMon latency markers related to the current frame. It behaves similar to reflextimespan function documented above
o Added new presentmon_latency_analyzer.ovl overlay layout. This layout replicates NVIDIA Reflex Latency analyzer using PresentMon’s latency markers, so you may use it as alternate solution in cases when NVIDIA Reflex Latency analyzer is not available (e.g. on Intel/AMD GPUs or in Direct3D9 or OpenGL application, where Reflex is not supported)
o Added buf(name, position) function support for accessing desired data source’s ring buffer contents. You may use this function to implement different forms of data filtering. For example PresentMon V2 DisplayLatency data source is designed to report zero for dropped frames, so you may apply (x > 0) ? x : buf(x,-1) formula to it to filter such data and report the previously sampled values instead
o Updated profiles list

ესეც გადმოსაწერი ლინკი: https://download-eu2.guru3d.com/rtss/[Guru3D.com]-RTSSSetup737Beta2Build28021.rar

[George]

MSI Afterburner 4.6.6 Beta 4 გამოვიდა სადაც RTSS 7.3.7 beta 3 დაემატა შემდეგი ჩეინჯლოგით:

MSI Afterburner 4.6.6 Beta 4 Build 16449:

- Added alternate V/F point locking mode. Now you may consequently press <L> key inside V/F curve editor to switch between voltage locking and max frequency locking modes
- Optimized GPU context switching implementation in HAL. This provides slightly faster GPU selection in GUI and slightly reduced CPU polling time for the systems with multiple GPUs
- Now MSI Afterburner uses extended 32KB RTSS text slot instead of previously used 4KB slot to display OSD, so legacy MSI Afterburner OSD can fit 8x more text items without hitting the limit
- Added Bulgarian localization
- Fixed ACL bypass vulnerability and NULL ptr dereference in RTCore driver (reported by Andres Roldan)
- Bundled RTSS version has been upgraded to 7.3.7 beta 3

RTSS 7.3.7 beta 3 Build 28022:

- Added Bulgarian localization
- Fixed ACL bypass vulnerability in OverlayEditor's RTCoreMini driver (reported by Andres Roldan)

ესეც გადმოსაწერი ლინკი: https://download-eu2.guru3d.com/afterburner/[Guru3D]-MSIAfterburnerSetup466Beta4Build16449.rar

[George]

 RTSS 7.3.7 beta 4 ბილდი გამოვიდა შემდეგ მცირე ცვლილებებით:

RTSS core in beta 4 is exactly the same as previous beta, there were no changes and it was simply recompiled to increment build number. The changes are localted inside imporved OverlayEditor plugin and summarized in this post.

Also it contains a few additional minor changes, which I didn't document there:

- Built-in profile for CS2 documented in this post: https://forums.guru3d.com/threads/msi-ab-rtss-development-news-thread.412822/page-229#post-6241977
- Now "with Radeon Graphics" text pattern is excluded by OverlayEditor from %CPU%/%CPUShort% macro names as requested in this post.
- Ryzen 9000 CPU hardware IDs added to OverlayEditor's hardware database. It is completely blind change, which was not tested on actual hardware, so I'm not sure if it will unlock full CPU monitoring functionality on new CPU family.

ესეც გადმოსაწერი ლინკი: https://download-eu2.guru3d.com/rtss/[Guru3D]-RTSSSetup737Beta4Build28023.rar

[George]

მოკლეთ unwinder-მა დახურა გურუ3დ-ზე ბეტა ვერსიების საპორტი, თუმცა ჰილბერტას თქმის არ დაუსრულებია მაგ აპზე მუშაობა, კიდე გააგრძელებსო ახალი ბილდების გამოშვებას, მაგრამ გარჩევების გარეშეო უკვე, ეგ ტიპი ისე ფაქტიურათ რუსია უკრაინული ფესვებით, ნიკოლაიჩუკია გვარათ და მგონი ტაგანროგიდანაა ანუ ეს თემა დაიხურა https://forums.guru3d.com/threads/msi-ab-rtss-development-news-thread.412822/page-233 ეს კიდე ჰილბერტას კომენტარი მაგ აპდთან დაკავშირებით სხვა თემაში : https://forums.guru3d.com/threads/rivatuner-stops-developing-debunked-no.454479/ ნელი ტემპებით წავა მემგონი მაგის დახვეწვა და ახალ-ახალი ბეტების გამოშვება აწი, თუმცა ბოლო RTSS ბილდი სექტემბერში გამოვიდა, MSI afterburner მანამდეც თქვა რომ დიდ ყურადღებას აღარ დავუთმობო აწი, უფრო RTSS გავაგრძელებო მუშაობას, როგორც ჩანს პირდაპირ გურუ3დ-ის გადმოწერების განყოფილებაში დაიდება შიგადაშიგ ახალი RTSS ბილდები, ანუ უფრო რედკათ ვიდრე აქამდე იდებოდა მაგ თემაში ცალკე

[George]

MSI Afterburner 4.6.6 Beta 5 Build 16555 გამოვიდა 3 თებერვალს უზარმაზარი ჩეინჯლოგით:

MSI Afterburner 4.6.6 Beta 5 Build 16554

• Added NVIDIA GeForce RTX 50x0 series graphics cards support:
o Added quad fan control support for reference design NVIDIA GeForce RTX 50x0 series graphics cards
o Added software interface for future G-Assist integration
o MSI Windows 11 Dark skin is selected as default application skin now. Please take a note that quad fan control GUI is currently available in Dark and Light editions of MSI Windows 11 skins only. If you prefer to use some older legacy skins on NVIDIA GeForce RTX 50x0 with 3 or 4 independently controllable fans, you may still switch between all fans in async fan control mode by pressing <Shift>+<F> repeatedly
o Added new MSI Special skin by Drerex design. New skin natively supports quad fan control feature on reference design NVIDIA GeForce RTX 50x0 series graphics cards
• Added voltage control support for low-cost revisions of NVIDIA GeForce RTX 40x0 GPUs
• Refactored CPU info database format provides more simplified and unified future database updates for MSI Afterburner monitoring HAL, CPU.dll sample plugin and RivaTuner Statistics Server’s own monitoring HAL
• Updated HwInfo.dll plugin. Now the plugin can automatically spawn new versions on HwInfo, which now store path to installed application in slightly different installer’s registry key
• Added experimental support for AMD Ryzen 9xxx CPUs
• Added experimental support for Intel Arrow Lake CPUs
• RivaTuner Statistics Server has been upgraded to v7.3.7 Beta 5

RivaTuner Statistics Server 7.3.7 Beta 5 Build 28209

• Improved hypertext parser:
o Improved bar embedding <B> hypertext tag syntax. Now the third tag parameter can specify rounding radius in pixels to allow you to render solid color filled bar with rounded corners. This feature was added to simplify implementation of overlay layouts similar to minimalist overlay designed by Devyn Johnston. You no longer need to implement similar rounded background with embedded images, now you may render such primitives directly from hypertext
• Added support for 64-bit hook library renaming. Now you may rename RTSSHooks64.dll and instruct RivaTuner Statistics Server to use renamed hook library instead of original one via configuration file. Power users may use this feature to diagnose and debug compatibility with various third party software products, which try to detect RivaTuner Statistics Server presence by hardcoding hook library name (e.g. bugged module name detection inside AMD 24.6.1+ OpenGL driver causing DaVinci Resolve to crash)
• Improved “Inject NVIDIA Reflex latency markers” option compatibility with 3D applications, performing runtime Direct3D11 device recreation (e.g. RetroArch when Direct3D11 rendering backend is selected)
• Improved render submission stage start detection for NVIDIA Reflex latency markers injected into Direct3D11 applications. Now render submission stage is detected inside ID3D11DeviceContext::ClearRenderTargetView hook as it was initially intended instead of doing it in ID3D11DeviceContext::RSSetState hook before
• Added alternate Direct3D12 command queue detection mode to provide compatibility with future NVIDIA software technology, which will hook Direct3D12 presentation calls similar to RivaTuner Statistics Server. New alternate mode is disabled by default, but it can be enabled by changing QueueDetection field to 2 inside [RendererDirect3D12] section of application profile
• Improved OverlayEditor.dll plugin:
o Updated helper PresentMonDataProvider.exe application:
- Both PresentMon console and service client based data importing modes updated to the latest PresentMon v2.3.0
- Slightly refactored frames parsing implementation in console mode to simplify integration of future console versions of PresentMon
- NaN values for dropped frames are now replaced with zeroes in service client mode to provide unified data format for both service client and console modes
- Dropped frames are now excluded from FramerateDisplayed metric calculation
o Added horizontal.ovl overlay layout:
- The layout demonstrates usage of improved <B> tag for rendering background with rounded colors
- The layout also allows you to switch between RivaTuner Statistics Server’s own framerate or framerate sampled with PresentMon via pressing <F> key. This allows you to use this layout either in traditional application hooking mode or in combination with Desktop Overlay Host to display framerate in applications like Destiny 2, which globally restrict hooking
o Updated helper LHMDataProvider.exe application:
- LibreHardwareMonitorLib.dll updated to the latest release v0.9.4
o Added experimental support for AMD Ryzen 9xxx CPUs
o Added experimental support for Intel Arrow Lake CPUs
o Added quad fan monitoring support for reference design NVIDIA GeForce RTX 50x0 series GPUs
• Updated bundled DesktopOverlayHost.exe application:
o Added keyboard shortcut for rendering backend restarting. Now you may press <Ctrl>+<Shift>+<R> when DesktopOverlayHost window is in focus to perform full restart of currently selected rendering backed, indicated by short background color flash. Rendering backend restarting feature can be used as universal vendor agnostic overlay stability test, allowing you to check how overlay handles full renderer restart from 3D application side

This version adds compatibility with NVIDIA Smooth Motion FG in Direct3D11 games. Also, Direct3D12 Smooth Motion compatibility path mentioned in the previous post is now enabled automatically when RTSS detects that NVIDIA Smooth Motion mode is in use. So no manual RTSS profile tweaks for Smooth Motion are necessary. Full changes list comparing to Beta 5 Build 28209 includes the following:

• Added compatibility with NVIDIA Smooth Motion frame generation technology:
o Now RivaTuner Statistics Server ignores presentation calls for invisible windows. This change is necessary to provide compatibility with NVIDIA Smooth Motion frame generation in Direct3D11 games
o Added alternate Direct3D12 command queue detection mode to provide compatibility with NVIDIA Smooth Motion frame generation technology, which hooks Direct3D12 presentation calls similar to RivaTuner Statistics Server. Alternate detection mode is automatically enabled at 3D application level when RivaTuner Statistics Server detects NVIDIA Smooth Motion hook module (nvpresent64.dll) in context of hooked application, but for testing purposes it can also be enabled globally by changing QueueDetection field to 2 inside [RendererDirect3D12] section of application profile
 

No changes in MSI Afterburner itself, the same version was recompiled to increment build number from 16554 to 16555. But the distributive now includes recently released RTSS 7.3.7 Beta 6 Build 28211 with improved NVIDIA Smooth Motion support. Also, the distributive includes new skin called MSI H2O by @Drerex. The same updated 4.6.6 Beta 5 Build 16555 will be published on MSI.com this week.

ნუ რამდენიმე ბეტა გამოვიდა 3 თებერვლამდე და საბოლოოთ მაგხელა ვებერთელა ჩეინჯლოგი აქვს ყველაფერს ერთად, MSI აფტერბარნერიც რამდენჯერმე განახლდა და ორი RTSS ბეტაც გამოვიდა თან, აგერაა კიდე დამატებითი ინფოები ნვიდიას 5000 სერიის კარტებზე ცალკე: https://forums.guru3d.com/threads/msi-afterburner-4-6-6-beta-5-for-nvidia-geforce-rtx-5000-series-cards.455155/

ესეც გადმოსაწერი ლიმკი: https://www.guru3d.com/download/msi-afterburner-beta-download/

[George]

ცალკე RTX 5000 სერიის მოდელებისთვის პატარა განახლება გამოვიდა კიდე:

Small trick for RTX 5000 series graphics cards, allowing you to extend memory overclocking range to +3000MHz:

- Download modified MSI AB hardware database file with extended memory clock adjustment range. This modified database file is intended for MSI Afterburner 4.6.6 Beta 5 Build 16555 only, it will not work with other versions of MSI AB:

https://download-eu2.guru3d.com/aft...burner466Beta5Build16555_ExtendedMemoryOC.rar 

- Open MSI AB installation folder ("C:\Program Files (x86)\MSI Afterburner" by default) and backup original MSI AB hardware database (MSIAfterburner.dat).
- Overwrite original database file with modified MSIAfterburner.dat extracted from archive you downloaded.

 

[George]

რაღაც მთლათ გაუბერა აფტერბარნერმა, აღარ მიხსნი სეთინგებს აფტერბარნერი და RTSS, რომ ვაკლიკავ არა აქვს რეაქცია და ფანჯარა არ იხსნება, შეიძლება ისევ ვინდოუსი უბერავს რაღაცას, ან კიდე VS redist ატრაკებს სადღაც, გადაყტენებამ არ უშველა, პროფილს უშვებს და აჩვენებს ყველაფერს, მაგრამ სეთინგებში არ მიშვებს

[George]

დავფიქსე ეგ ბაგიც, დიდი ალბათობით ან რაღაცა ფაილი აკლდა ან ტელემეტრია მქონდა VS დაბლოკილი სანლექსის ტვიკებში ცალკე რომ დადო და მაგი უშლიდა ხელს, რომ გავხსენი ტელემეტრია ჩაირთო ისევე ყველაფერი

[George]

მოკლეთ აი ასეთი ბაგი აქვს ბოლო ვინდოუს 11 აპდეიტებში 26100 პრევიუ აპდეიტი იქნება 3323 თუ 26120 დევ ბილდის განახლება:

https://forums.guru3d.com/threads/rtss-and-sfc.455588/

https://forums.guru3d.com/threads/bug-rivatuner-on-screen-windows-11-keyboard.455595/

ხო ერორს კი აგდებს მაგრამ ყველაფერი აგრძელებს მუშაობას მაგის შემდეგ, არა აქვს დაკრაშვები არსად, იფიქსება ან ბოლო აპდეიტის გათიშვით, ანუ 26120 ან 26100 უფრო ძველ ბილდებზე არ ატრაკებს, ან კიდე RTSS-ის ამ გალოჩკის მოხსნით, რაც თიშავს 64 ბიტიანი თამაშების და აპების საპორტს :

როდის დაფიქსავს მიკროსოფტი მაგ ბაგს არავინ არ იცის, ხო მარტო RTSS არ ახურებს, ქონია ეგეთივე ტიპის გაბერვები osk.exe და sfc/ scannow გაშვების შემდეგ აგრეთვე ნორტონის,ავირას და ბიტდეფენდერის დრაივერებზე, ეგენი დაფიქსავენ თუ არა მაგას თავისით არ ვიცი, შეიძლება მიკროსოფტის დასაფიქსია ეგ ბაგი, სხვა ვერ გაასწორებს დიდათ, კასპერსკიზე ეგ ერორი ნამდვილათ არა აქვს, მანდ სხვანარი ტიპის დრაივერებს იყენებს ალბათ კასპერსონი და ეგ ერორი არ ფიქსირდება, აქვს სტიმის დრაივერზეც ეგ ერორი აი აგერ დაპოსტეს კიდეც: https://steamcommunity.com/app/1787090/discussions/0/595140733594266095/ ანუ მარტო RTSS არ აგდებს მაგ ერორებს, კიდე ბევრგან ფიქსირდებაო თუ გიყენია ბოლო ვინდოუს 11 ინსაიდერული აპდეიტი, Canary რა ხდება არ ვიცი, მაგრამ ჩემთან დევ ბილდზე ატრაკებდა და 2323 ბილდზეც ახურებს, 2360 დევ ბილდზე აღარ დამიტესტია და მეზარება ეხლა დაყენება, თუ გაუბერა აპდეიტებმა მერე მოვსინჯავ მაგაზეც ალბათ, მაგრამ ეგ უკვე შემდეგ კვირას იქნება როცა გამოვა შემდეგი განახლება 26100-ზე, ერთადერთი არ ვიცი ერორს მარტო oks.exe და sfc.exe გაშვების შემდეგ აგდებს მაგ დრაივერებზე, თუ სხვა აპებიც ახურებენ, მე ეგ ორი დავაფიქსირე მარტო და მგონი სხვებმაც, რაღაცა რედკი ბაგია მგონი აჩერედნოი კრივოი ინდუსი დევების, ადრე რუსები ყავდათ ძლიერი ბილ გეიტსის დროს, მერე შეეშინდათ შპიონაჟის და დაითხოვეს რუსი სპეციალისტები მთლიანათ მგონი მიკროსოფტიდან და მასიურათ აიყვანეს ინდუსები, მაგის შემდეგ გაძვრა სული უკვე მიკროსოფტს, მეთაურობაც არ იყო მაღალი დონის თან, მილიარდებს კი შოულობენ, მაგრამ დღემდე გრძელდება ეგ ბარდაკი და ნაწილი მაგ პრობლემების რუს და ინდუს სპეციალისტებს ეხება როგორც არ უნდა ატრიალო, ნუ როგორც დირექტორი ბილ გეიტსი უფრო გამოცდილი იყო და არ უშვებდა მაგდენ შეცდომებს მაინც, ეხლა კი მაგარი ბარდაკია

[George]

აპდეიტი დადო ანვინდერმა მაგ ბაგთან დაკავშირებით: https://forums.guru3d.com/threads/rtss-and-sfc.455588/page-2 ბევრ დრაივერებზე ქონია ეგ ბაგი ბოლო აპდეიტებში, დარწმუნებულია თვითონ რომ მიკროსოფტი დაფიქსავსო თავისით მაგ ბაგს რომ ნახავს რომ ბევრს აქვს პრობლემები და არავინ არ აპირებს მაგის დაფიქსვას ცალკე, შეიძლება არც ეგეთი ადვილია მაგის გასწორება, სერიოზული ცვლილებები უნდა მოახდინო, ან არც კი შეიძლება თეორიულათაც კი, მაშინ ეგ ფუნქციები აღარ იმუშავებს ალბათ, ბოლო პოსტში უწერია კარგათ მაგ ბაგის შესახებ და დაახლოებით რაც ხდება, ანუ მაგ ყველაფრით მიკროსოფტი ვითომ ცდილობსო შეზღუდვას 3party DLL თავის სისტემურ აპებზე და დრაივერებზე, მაგიტომაა ალბათ რომ ერორს აგდებს მაგრამ არ იკრაშება აპი და განაგრძობს მუშაობას, მაგრამ ნერვებზე მოქმედებს ეგ შეტყობინება ყოველი გაშვების შემდეგ, მთელი ტიპის ანტივირუსებზე აქვს, სხვა ტიპის hooks, იგივე ბანდიკამი და სხვა აპები რასაც ხმარობენ + Rtss, უბრალოთ მარტო 64 ბიტიანზე ახურებს, 32 მუშაობს მაგ ერორის გარეშე, კასპერსონაზე რომ არა აქვს კარგია, თორე ტვინს შეჭამდა ეგ ბაგი, მე Rtss-ზეც დროებით გამოვრთე 64 ბიტიანის საპორტი, რა იცის კიდე რამდენი ხანი დარჩება ეგ დაუფიქსავი და მაგ ბაგის გამო აპდეიტების გამორთვაც არ ვაპირებ, თუმცა ჯერ მარტო პრევიუ და დევ ბილდების ბოლო განახლებაზე აქვს 26120-26100 ბილდებზე ეგ ერორი, ისე კი უკვე გაიყოო ორ ნაწილათ 26120 დევ ბილდი, ერთზე დევ ბილდია, მეორეზე ბეტა უკვე, დროებით ეგრე დარჩებაო, მე ეხლა ბეტა მიყენია, მაგრამ დევ ბილდზე რომ გადავიყვანო ამ უტილიტით რაღაცა სხვა აპდეიტს იწერს თურმე ხელმეორეთ ცოტა განსხვავებული ფუნქცუიონალით თუმცა ვოტერმარკი არც მაგაზე არ ქონია https://github.com/abbodi1406/offlineinsiderenroll/releases სტიმის დრაივერზეც ქონია ეგ ბაგი ისე, ესეც ზუსტათ იგივე ტიპის ბაგია https://forums.guru3d.com/threads/bug-rivatuner-on-screen-windows-11-keyboard.455595/ სხვაობა ისაა რომ პირველზე როცა უშვებ sfc/ scannow პოვერშელში ან command prompt-ში მაშინ აგდებს მაგ ერორს bad image-ის,მეორეზე კი როცა ვირტუალურ კლავიატურას უშვებ მაშინ უკვე, osk.exe ანუ, მაგის სანაცვლოთ სხვა რაღაცის გამოყენებაც შეიძლება, მაგაზე თურმე აღარ ახურებს რახან მიკროსოფტის არაა, არის მანდ ლევი აპი უფრო კარგი ფუნქციონალით, თუმცა მე ჯერ 64 ბიტიანის საპორტი გამოვრთე RTSS, თუ დიდხანს მოანდომეს მაგის დაფიქსვა შეიძლება მეც ჩავრთო ისევე 64 ბიტიანი საპორტი და ეგ მეორე აპი დავაყენო უკვე რედიტზე დადო ერთმა ლინკი მაგრამ აღარ დამიტესტია, იდეაში იმდენ რაღაცაზე ახურებს რომ არა მგონია მიკროსოფტმა ეგ ეგრე დატოვოს, თუმცა რა იცის რეებს ჩალიჩობენ ეგენიც, ტრამპმა ყველაფერი აირ-დარია და ეგენიც რაღაცეებს გოთვერნობენ ეტყობა ბოლო პერიოდია

[George]

ისე პოვერ სეთინგების გათიშვა დღემდე აქტუალური ყოფილა აფტერბარნერში სტატერინგის დასაფიქსათ თანამედროვე სისტემებზეც ამცირებსო კარგათ სტატერს, მეც ეგრე მაქვს მონიშნული რახან ეგ სტატერინგს არა მარტო თამაშებში, ვიდეოებში და ბრაუზერშიც იწვევს:

MSi Afterburner Potential 1% Lows & Stutter Issues - FIXED https://www.techpowerup.com/forums/threads/msi-afterburner-potential-1-lows-stutter-issues-fixed.330207/

[George]

წინა კვირას ახალი MSI Afterburner v4.6.6 Beta 6 Build 16591 გამოვიდა შემდეგი ჩეინჯლოგით:

- Extended +3000MHz memory clock overclocking range tweak for NVIDIA RTX 5000 series graphics cards, which was available as separately downloadable hardware database for the previous beta, is now integrated into distributive.
- Added some future NVIDIA RTX 5000 series GPUs to hardware database.
- Added unofficial AMD RDNA4 GPU family support
- Relaxing preemptive sleep delay is now excluded from cumulative polling time displayed by performance profiler on Overdrive8 capable graphics cards to avoid confusing beginners.
- Fixed missing CPU Power sensor on Intel Arrow Lake CPU family
- RivaTuner Statistics Server has been upgraded to v7.3.7 beta 7

RTSS v7.3.7 Beta 7 Build 28220:

- Reduced cumulative committed memory usage due to switching from statically to dynamically allocated hypertext parser context.
- Added built-in overlay profile for "Doom : The Dark Ages". The profile included into distributive applies exactly the same compatibility setting for this game as we previously published in forums: it forces RTSS overlay to be rendered at DXGI interoperability swapchain level to bypass native Vulkan overlay renderer's incompatibility with Vulkan AMD FSR frame generation implementation. The profile is not necessary if you don't use AMD FSR frame generation, so you may safely remove .\ProfileTemplates\DOOMTheDarkAges.exe.cfg on NVIDIA hardware if you prefer to see overlay rendered at native Vulkan API level.
- Improved displayed 3D API information for Vulkan/OpenGL applications, rendering RTSS overlay at DXGI interoperability swapchain level. Now 3D API for such applications is displayed as OpenGL/DXGI or Vulkan/DXGI instead of displaying interoperability DXGI 3D API only (Direct3D11 or Direct3D12).
- Updated bundled PresentMonDataProvider.exe application. Now it supports both console and service versions of the latest PresentMon v2.3.1. Please take a note that Intel implemented some fundamental changes in PresentMon API DLL loading, so the previous versions of the service are no longer supported by PresentMonDataProvider.
- Some fixes in compact CPU name macro formatting implementation for 16-Core CPUs.
- Slightly improved algorithm for MPERF based base clock and bus clock calculation implementations. Now the algorithm ensures that both MPERF performance counter and timestamp counter are sampled inside the same scheduler's time slice and filters the values if sampling was interrupted by scheduler.

ესეც გადმოსაწერი ლინკი: https://download-eu2.guru3d.com/afterburner/[Guru3D]-MSIAfterburnerSetup466Beta6Build16591.rar

ახალი ბეტა სულ მალე გამოვაო კიდე ამ ინფოში უწერია გურუ3დ-ის ფორუმზე:

Good news for NVIDIA RTX 4000/5000 series GPU owners:

As you know, NVIDIA implemented some architectural changes in framepacing with DLSS4 frame generation introduction. When DLSS4 frame generation is in use, framepacing is no longer performed at rendering pipeline input side, now it is done at rendering pipeline output side either by dedicated flip metering unit in display engine (on RTX 5000 series cards) or by display driver (on RTX 4000 series cards). In both cases programmable flip delay is applied to presented frame immediately before it is getting displayed. Due to such framepacing approach "fat" or "jigsaw" pattern is normal and expected to be visible on frametime monitored at rendering pipeline input, because frames are now entering the pipeline as soon as they are ready and framepacing is done at later stage.
NVIDIA actively promoted PresentMon's msBetweenDisplayChange metric as the only way allowing you to monitor what's going on at rendering pipeline output side and see the effect of flip metering with active DLSS4 frame generation. However, as I explained before, NVIDIA used their internal privately patched build of PresentMon, so it was not possible to monitor flip metering with any public versions of PresentMon or public PresentMon based tools like OverlayEditor in RTSS or CapFrameX. Things finally changed (better late than never) and NVIDIA finally worked with Intel and implemented flip metering support in public PresentMon source code. Unfortunately the latest version of PresentMon (v2.3.1) was released just a couple weeks ago, and NVIDIA's flip metering support related changes were committed after that. So DLSS4 flip metering support it is not yet available in any downloadable PresentMon installers (neither console nor service) and you'll probably need to wait a few more weeks or months until Intel launch new version. But it is possible to build PresentMon manually with DLSS4 flip metering related changes, commited by NVIDIA. And that's exactly what I plan to do: to let revewiers to play with it without waiting for future PresentMon release, I'll compile custom build of console PresentMon v2.3.1 with DLSS4 flip metering support and include it into RTSS. So one more RTSS beta with it is expected within a couple of days.

Stay tuned!

 

[George]

To demonstrate new build of PresentMon with DLSS4 flip metering support the next beta of RTSS will include new overlay layout for OverlayEditor plugin - presentmon_framegen.ovl. It will allow you to see and compare frametimes on both rendering pipeline input and rendering pipeline output sides and help you to understand how flip metering works on RTX 4000/5000 series cards. The following screenshot demonstrates presentmon_framegen.ovl layout usage in Doom : The Dark Ages with DLSS4 frame generation on RTX 4090. Top overlapped frametime and GPU busy graphs reflect traditional present-to-present frametimes on rendering pipeline input side. Left part of the graph was captured when the game was in menu, so frame generation was inactive during that period of time and you can see usual frametime and GPU busy graphs there. Once I left the game menu, DLSS4 frame generation was activated and both present-to-present frametime and GPU busy graphs became "fat". That's exactly what you're expected to see with active DLSS4 frame generation, such fat/jigsaw pattern is a result of repeating sequence of long(real)-short(ai)-long-short frames entering rendering pipeline. The same apply to GPU busy graph, where real frames take long time to be rendered and AI frames take short time and such sequence of long-short-long-short frames is repeating making the graph "fat". Numeric values indicating frametime/GPU busy for top graphs (reflecting rendering pipeline input side) have been slightly modified in presentmon_framegen.ovl overlay layout too. The values no longer display instant frametime/GPU busy, because this way depending on sampling time you'll see either real or generated frame's values there, which confuses a lot of inexperienced users. Now instead of showing instant values, those indicators in this overlay layout are showing the maximum frametime/GPU busy for last 4 sampled frames. Such approach is effectively filtering short values for generated frames (for both FG and MFG modes) and always show you real frame's statistics only. Limited by CPU/GPU indicator is also using maximums for the last 4 sampled frames due to the same reason.
Bottom graph is display-to-display frametime, PresntMon's msBetweenDisplayChange. This graph is showing you the frames on rendering pipeline output side, that's where NVIDIA flip metering engine apply framepacing to generated frames. As you can see, presented and displayed frametimes match when frame generation is inactive, but once FG is enabled the magic begins and you start seeing effect of flip metering on the bottom graph. Also, if you try to experiment with this layout in windowed applications, please keep in mind that in this case you'll see the results of DWM activity on displayed frametime, so it is normal and expected to see spikes/dips due to dropped frames, due to DWM composition etc.
Percentile based metrics (1.0%/0.1% lows) are also calculated and displayed on both pipeline input and output (i.e. for present-to-present and display-to-display frametimes), so you can compare them too.

ეგეთი ინფო დაიდო კიდე გურუ3დ-ის ფორუმზე შემდეგ ბეტაზე

[George]

RTSS v7.3.7 Beta 8 ბილდი გამოვიდა ეხლახან შემდეგი ჩეინჯლოგით:

- "Display maximum frametime" option in RTSS properties is no longer affected by minimum overlay refresh period. Previously maximum frametime spikes inside RefreshPeriodMin period (10ms by default) could be skipped in this mode.
- Custom version of console PresentMon 2.3.1 is now bundled with PresentMonDataProvider. PresentMon-2.3.1-x64-DLSS4.exe is an official version with additional changes from NVIDIA side, which finally add flip metering support for msBetweenDisplayChange metric. More details about it can be found in this post. Please take a note that PresentMonDataProvider can consume data from both service and console versions of PresentMon and the service has higher data streaming priority. So to ensure that you stream desired data from PresentMon-2.3.1-x64-DLSS4.exe please ensure that the service is not installed.
- Added burst mode to PresentMonDataProvider's data consumption loop for both service and console versions of PresentMon. Burst mode is activated when PresentMonDataProvider detects that PresentMon's reporting lag increases due to way it is batching frames in applications with extremely high framerate. In this mode PresentMonDataProvider starts consuming data batches from PresentMon without any relaxing sleep periods between batches to compensate increasing reporting lag.
- Added new presentmon_framegen.ovl overlay layout for OverlayEditor plugin. New layout allows you to track frametimes at both rendering pipeline input (present-to-present intervals) and rendering pipeline output (display-to-display interval) sides. In conjunction with custom PresentMon-2.3.1-x64-DLSS4.exe it helps you to monitor framepacing performed by flip metering unit in DLSS4 frame generation mode. More info about this overlay layout can be found in this post.

ესეც გადმოსაწერი ლინკი: https://download-eu2.guru3d.com/rtss/[Guru3D]-RTSSSetup737Beta8Build28221.rar თუ არ გადმოწერა გურუ3დ-ის ფორუმიდან დააკლიკეთ და გადმოწერს მაშინ 

[George]

დაახლოებით 1 კვირის უკან MSI Afterburner v4.6.6 Beta 7 ბილდი გამოვიდა შემდეგი ჩეინჯლოგით:

- Added MP2988 and MP29816A voltage controllers support. New PWM controllers support is reserved for future extreme overclocking oriented graphics card models with triple/quad channel voltage control. Please take a note that such voltage controllers are also used on the most of currently existing NVIDIA RTX 5080 and RTX 5090 series cards, but unfortunately I2C address range of PWM controllers on reference cards is blacklisted by NVIDIA display driver, making extended voltage control on currently existing NVIDIA RTX 50x0 series cards impossible.
- Name of voltage rail mapped to AUX or AUX2 voltage slider (e.g. VDDCI, MSVDD, PEXVDD or 1V8) is now displayed in context help of AUX or AUX2 GUI controls and inside MSI Afterburner system info window ("i" button).
- Added internal database format improvements, allowing us to remap AUX/AUX2 GUI controls to different physical voltage rail if multiple optional controllable voltage rails are available
- Added new GPU hiding functionality. Added "Hide GPU" button to GPU selection panel of default Win 11 Dark and Win 11 Light skins. Alternately, you may use <Ctrl>+<H> keyboard shortcut to hide currently selected GPU if you prefer to use legacy or third party GUI skins with no such button. You may use it to completely remove some GPUs from MSI Afterburner's GUI, including both hardware control and hardware monitoring sections. The functionality is intended for hiding integrated iGPUs. More info in this post.
- Empty compact names of some iGPUs in GPU selection panel are now replaced with abstract "AMD GPU" / "NVIDIA GPU" / 'Intel GPU" names.
- Fixed GUID, PCI location and registry key info displayed in MSI Afterburner system info window ("i" button) for secondary Intel GPUs.
- I2C cache of secondary GPUs is now saved inside proper hardware profile file. Previously it was always written to and read from the primary GPU's profile.
- Asynchronous fan control buttons are no longer displayed in GUI on GPUs with no fan control support (e.g. iGPUs).
- RivaTuner Statistics Server has been upgraded to v7.3.7 beta 9

https://download-eu2.guru3d.com/afterburner/[Guru3D]-MSIAfterburnerSetup466Beta7Build16756.rar

ცალკე დაიდო RivaTuner Statistics Server v7.3.7 Beta 9 ჩეინჯლოგიც გურუ3დ-ის ფორუმზე:

- Now RivaTuner Statistics Server forcibly enables "Use Microsoft Detours API hooking" compatibility option for processes, which load NVIDIA Smooth Motion runtime (NvPresent64.dll). This change is required to improve compatibility with hooks installed by Smooth Motion runtime and fix overlay flickering in Direct3D11 applications when Smooth Motion frame generation is enabled.
- Added new options to main RivaTuner Statistics Server window, allowing you to center overlay horizontally / vertically.
- Added new <FNT> hypertext format tag and new GUI option "Lock user settings" associated with this tag. New tag & option allow displaying OverlayEditor's overlay layout with native overlay creator's font and zooming ratio, which ensures that nothing is distorted due to different font selection on user side. More info in this post.
- Added new "Dark background" menu option to OverlayEditor plugin. You may uncheck it to switch editor's workspace to light mode, if you need to edit your own overlays containing dark background elements. More info in this post.
- Sadly EA did nothing to fix or at least document recent EA App launcher's bug, which broke RivaTuner Statistics Server's compatibility with some games like Battlefield I or Battlefield V. So I decided to provide temporary bandaid workaround to address the problem, which EA introduced on their side. Now RTSS supports two alternate shader compilation paths and can transparently switch to alternate compiler when EA App unintentionally (?) block loading of legacy compiler's runtime.

ესეც გადმოსაწერი ლინკი ცალკე RTSS-ის: https://download-eu2.guru3d.com/afterburner/[Guru3D]-RTSSSetup737Beta9Build28312.rar თუმცა msi ბოლო პაკეტში მოყვება ეგეც 

[George]

დღეს MSI Afterburner 4.6.6 ფინალური ბილდი გამოვიდა შემდეგი ჩეინჯლოგით:

Version 4.6.6

Ported to VC++ 2022 compiler. Please take a note that due to this change MSI Afterburner will no longer be able to start under Windows XP. Please stay on the previous versions of the product if you need this OS support
Please take a note that size of mandatory VC++ 2022 runtime redistributables roughly doubled comparing to the previously used VC++ 2008 redistributables, and we’d like to avoid providing overblown application distributive, drastically increased in size due to bundling newer and much heavier VC++ redistributables with it. To deal with this issue we provide our own original tiny web installer for VC++ redistributables, which allowed decreasing the size of final application distributive drastically even comparing to the previous VC++ 2008 based version. Please take a note that install time can be increased slightly due to downloading VC++ 2022 runtimes redistributables on the fly during installation. If you install MSI Afterburner offline, you can always deploy required VC++ 2022 distributives later with web installer by launching .\Redist\VCRedistDeploy.bat
Now the installer and uninstaller are performing active application instance checking on earlier stages before displaying any GUI
Added NVIDIA GeForce RTX 50x0 series graphics cards support
Added quad fan control support for reference design NVIDIA GeForce RTX 50x0 series graphics cards
Added software interface for future G-Assist integration
MSI Windows 11 Dark skin is selected as default application skin now. Please take a note that quad fan control GUI is currently available in Dark and Light editions of MSI Windows 11 skins only. If you prefer to use some older legacy skins on NVIDIA GeForce RTX 50x0 with 3 or 4 independently controllable fans, you may still switch between all fans in async fan control mode by pressing <Shift>+<F> repeatedly
Added new MSI Special and MSI H2O skins by Drerex design. New skins natively supports quad fan control feature on reference design NVIDIA GeForce RTX 50x0 series graphics cards
Extended +3000MHz memory clock overclocking range tweak for NVIDIA RTX 5000 series graphics cards
Added AMD RDNA4 GPU family and AMD RADEON RX 9000 series cards support. Please take a note that MSI do not sell any AMD RDNA4 based graphics cards, so RDNA4 support is completely unofficial
Added MP2988 and MP29816A voltage controllers support. New PWM controllers support is reserved for future extreme overclocking oriented graphics card models with triple/quad channel voltage control. Please take a note that such voltage controllers are also used on the most of currently existing NVIDIA RTX 5080 and RTX 5090 series cards, but unfortunately I2C address range of PWM controllers on reference cards is blacklisted by NVIDIA display driver, making extended voltage control on currently existing NVIDIA RTX 50x0 series cards impossible
Added voltage control support for AMD RADEON RX 7800XT series graphics cards
Added voltage control support for new low-cost revisions of NVIDIA GeForce RTX 40x0 GPUs
Added new GPU hiding functionality. Added "Hide GPU" button to GPU selection panel of default Dark and Light editions of MSI Windows 11 skins. Alternately, you may use <Ctrl> + <H> keyboard shortcut to hide currently selected GPU if you prefer to use legacy or third party GUI skins with no such button. You may use it to completely remove some GPUs from MSI Afterburner's GUI, including both hardware control and hardware monitoring sections. The functionality is intended for hiding integrated iGPUs
Relaxing preemptive sleep delay is now excluded from cumulative polling time displayed by performance profiler on Overdrive8 capable graphics cards to avoid confusing beginners
Name of voltage rail mapped to AUX or AUX2 voltage slider (e.g. VDDCI, MSVDD, PEXVDD or 1V8) is now displayed in context help of AUX or AUX2 GUI controls and inside MSI Afterburner system info window ("i" button)
Added internal database format improvements, allowing us to remap AUX/AUX2 GUI controls to different physical voltage rail if multiple optional controllable voltage rails are available
Empty compact names of some iGPUs in GPU selection panel are now replaced with abstract "AMD GPU" / "NVIDIA GPU" / “Intel GPU" names
Fixed GUID, PCI location and registry key info displayed in MSI Afterburner system info window ("i" button) for secondary Intel GPUs
I2C cache of secondary GPUs is now saved inside proper hardware profile file. Previously it was always written to and read from the primary GPU's profile
Asynchronous fan control buttons are no longer displayed in GUI on GPUs with no fan control support (e.g. iGPUs)
Now MSI Afterburner reinitializes skin scaling engine on DPI scaling change events to prevent cases when GUI looks cut off in some cases (e.g. after switching between display resolutions with different DPI scaling settings)
Now MSI Afterburner is forcibly uninstalling IO driver on OS shutdown to prevent the driver from being hibernated and reloaded due to fast startup till the next clean system boot event
Fixed instances enumeration for some performance counters with no localized names (e.g. GPU related Windows performance counters) in PerfCounter.dll plugin
Added experimental support for "Beta : Use Unicode UTF-8 for global language support" option enabled in administrative regional OS settings. Now each localization description file contains additional "Codepage" field, defining runtime ANSI to UTF8 conversion rule for selected language pack
Refactored CPU info database format provides more simplified and unified future database updates for MSI Afterburner monitoring HAL, CPU.dll sample plugin and RivaTuner Statistics Server’s own monitoring HAL
Added experimental support for AMD Ryzen 9xxx CPUs
Added experimental support for Intel Arrow Lake CPUs
Added alternate Voltage/Frequency point locking mode inside Voltage/Frequency curve editor. Now you may repeatedly press <L> key inside V/F curve editor to switch between voltage locking and max frequency locking mode
Optimized GPU context switching implementation in HAL. This provides slightly faster GPU selection in GUI and slightly reduced CPU polling time for the systems with multiple GPUs
Now MSI Afterburner uses extended 32KB RivaTuner Statistics Server’s text slot instead of previously used 4KB slot to display On-Screen Display. Due to this improvement legacy MSI Afterburner On-Screen Display can fit up to 8x more text items without hitting the limit
Updated HwInfo.dll plugin. Now the plugin can automatically spawn new versions on HwInfo, which now store path to installed application in slightly different installer’s registry key
Added Bulgarian localization
Seriously revamped German and Ukrainian localizations
Fixed ACL bypass vulnerability and NULL ptr dereferencein RTCore driver (reported by Andres Roldan)
RivaTuner Statistics Server has been upgraded to v7.3.7:
New version’s changes list is huge, it includes more than 150 compatibility enhancements, changes and new features including programmable conditional layers support, PresentMon, LibreHardwareMonitor and NVIDIA Reflex integration

ცალკე RTSS ახალი ვერსიის ჩეინჯლოგი უზარმაზარია:

Version 7.3.7

Improved hypertext parser:
Now bar range graphs are rendered when the range width is less than one pixel
Improved bar embedding <B> hypertext tag syntax. Now the third tag parameter can specify rounding radius in pixels to allow you to render solid color filled bar with rounded corners. This feature was added to simplify implementation of overlay layouts similar to minimalist overlay designed by Devyn Johnston. You no longer need to implement similar rounded background with embedded images, now you may render such primitives directly from hypertext.
Added new <FNT> hypertext tag support. New tag is allowing hypertext parser to ignore the font selected by user in RivaTuner Statistics Server application profiles and always use specified font, font size and overlay zooming ratio. New tag is allowing OverlayEditor plugin to render overlay layouts with original font and zooming out of box and without manually applying master font settings
Added late initialization of Vulkan device function pointers aimed to improve Vulkan hooking reliability in the applications dynamically loading vulkan-1.dll on the systems with multiple installed third party Vulkan layers
Added support for 64-bit hook library renaming. Now you may rename RTSSHooks64.dll and instruct RivaTuner Statistics Server to use renamed hook library instead of original one via configuration file. Power users may use this feature to diagnose and debug compatibility with various third party software products, which try to detect RivaTuner Statistics Server presence by hardcoding hook library name (e.g. bugged module name detection inside AMD 24.6.1+ OpenGL driver causing DaVinci Resolve to crash)
Improved “Inject NVIDIA Reflex latency markers” option compatibility with 3D applications, performing runtime Direct3D11 device recreation (e.g. RetroArch when Direct3D11 rendering backend is selected)
Improved render submission stage start detection for NVIDIA Reflex latency markers injected into Direct3D11 applications. Now render submission stage is detected inside ID3D11DeviceContext::ClearRenderTargetView hook as it was initially intended instead of doing it in ID3D11DeviceContext::RSSetState hook before
Added new options to main RivaTuner Statistics Server window, allowing you to center overlay horizontally or vertically
Reduced cumulative committed memory usage due to switching from statically to dynamically allocated hypertext parser context
Improved displayed 3D API information for Vulkan/OpenGL applications, rendering RivaTuner Statistics Server overlay at DXGI interoperability swapchain level. Now 3D API for such applications is displayed as OpenGL/DXGI or Vulkan/DXGI instead of displaying interoperability DXGI 3D API only (Direct3D11 or Direct3D12)
"Display maximum frametime" option in RivaTuner Statistics Serve properties is no longer affected by minimum overlay refresh period. Previously maximum frametime spikes inside RefreshPeriodMin period (10ms by default) could be skipped in this mode
Added workaround for recent EA App launcher’s bug, which unintentionally broke RivaTuner Statistics Server's compatibility with some games like Battlefield I or Battlefield V. The problem was caused by blocking loading of legacy d3dx11_42.dll runtime from EA side. RivaTuner Statistics Server depended on this library for runtime shader compilation and backward compatibility with older operating systems. Now RivaTuner Statistics Server supports two optional shader compilation paths and can switch to alternate compiler runtime if loading of d3dx11_42.dll is blocked
Added compatibility with NVIDIA Smooth Motion frame generation technology:
Now RivaTuner Statistics Server ignores presentation calls for invisible windows. This change is necessary to provide compatibility with NVIDIA Smooth Motion frame generation in Direct3D11 games
Added alternate Direct3D12 command queue detection mode to provide compatibility with NVIDIA Smooth Motion frame generation technology, which hooks Direct3D12 presentation calls similar to RivaTuner Statistics Server. Alternate detection mode is automatically enabled at 3D application level when RivaTuner Statistics Server detects NVIDIA Smooth Motion hook module (nvpresent64.dll) in context of hooked application, but for testing purposes it can also be enabled globally by changing QueueDetection field to 2 inside [RendererDirect3D12] section of application profile
Now RivaTuner Statistics Server forcibly enables "Use Microsoft Detours API hooking" compatibility option for processes, which load NVIDIA Smooth Motion hook module (nvpresent64.dll). This change is required to improve compatibility with hooks installed by Smooth Motion runtime and fix overlay flickering in Direct3D11 applications when Smooth Motion frame generation is enabled
Improved OverlayEditor plugin:
Now you may press <Ctrl>+<A> inside the editor window to select all overlay layers or press <Ctrl>+<Shift>+<A> to deselect all selected layers
Now new layout creation command deletes all data sources and text tables, instead of deleting just overlay layers before
Save prompt is now displayed when you try to load new overlay layout without saving the previously modified overlay
Added new “Merge” command to “Layouts” Now you can easily combine multiple overlay layouts or copy data sources, layers or text tables between the layouts. All layers of source layout are automatically selected after merging to allow you to move them to new position with a single mouse gesture
Added new "Dark background" option to “View”. You may uncheck it to switch editor's workspace to light mode, if you need to edit your own overlays containing dark background elements
Improved NVIDIA Reflex metrics access functionality. Previously NVIDIA Reflex metrics were represented by hardcoded internal "NVIDIA Reflex XXX timespan" and "NVIDIA Reflex XXX latency" groups of data sources. Such internal data sources are deprecated since this version, now you can use more robust functions for accessing NVIDIA Reflex data:
New reflexlatency(markerFrom, markerTo) function can be used to calculate a latency between two arbitrary NVIDIA Reflex latency markers related to the current frame. You can specify one of the following marker identifiers: 0 – input_sample, 1 – simulation_start, 2 – simulation_end, 3 – rendersubmit_start, 4 – rendersubmit_end, 5 – present_start, 6 – present_end, 7 – driver_start, 8 – driver_end, 9 – osrenderqueue_start, 10 – osrenderqueue_end, 11 – gpurender_start, 12 – gpurender_end, 13 – virtual gpurender_start (gpurender_end minus GPU active time). You may also set markerFrom to -1 to specify the very first marker and set markerTo to -1 to specify the very last marker if necessary. This function allows you to calculate all NVIDIA Reflex latencies previously represented by hardcoded internal "NVIDIA Reflex XXX latency" data sources and implement additional custom latencies (e.g. a latency between simulation_start and present_start markers)
New reflextimespan(markerFrom, markerTo, base) function can be used to calculate and encode a timespan between two arbitrary NVIDIA Reflex latency markers related to the current frame. This function allows you to calculate all NVIDIA Reflex timespans previously represented by hardcoded internal "NVIDIA Reflex XXX timespan" data sources and implement additional custom timespans (e.g. a timespan between simulation_start and present_start markers). Also the previous "NVIDIA Reflex XXX timespan" data sources calculated timespan using fixed base 100 (meaning that the timeband graphs visualizing the timespan had just 100 gradations), now you may specify custom base to implement more detailed timebands
New reflextime(id) function can be used to access absolute times reported by NVIDIA reflex for the current frame. You may set time identifier to 0 to get GPU active or 1 to get GPU frametime
Built-in reflex.ovl overlay layout has been modified to demonstrate new reflextime/reflextimespan/reflexlatency functions usage instead of deprecated "NVIDIA Reflex XXX timespan" and "NVIDIA Reflex XXX latency" data sources. Please apply similar modifications to your custom overlays, if you used deprecated data sources there
Updated helper PresentMonDataProvider.exe application:
Added PresentMon V2 service support. Please take a note that due to major changes between PresentMon V1 and V2 service API, service API V1 is no longer supported
Added PresentMon V2 console application support. Unlike the service, console mode is backward compatible with the previous versions of PresentMon, so you may switch to the previous versions of console PresentMon if necessary via the configuration file. The most recent versions of console PresentMon V1 and V2 applications are included into distributive
Added PresentMon V2 metrics support. The most of PresentMon V1 metrics are also reconstructed in realtime from V2 metrics for your convenience and for backward compatibility with previously existing PresentMon based overlay layouts. Please take a note that PresentMode metric encoding changed between V1 and V2 versions of PresentMon service API, so now OverlayEditor use new encoding too. All PresentMon based overlay layouts have been updated to reflect new PresentMode encoding. Please apply the same changes to your overlays if you decode PresentMode there
PresentMon V1 service API specific hardware telemetry is no longer available in PresentMon data provider
Custom version of console PresentMon v2.3.1 is now bundled with PresentMonDataProvider. PresentMon-2.3.1-x64-DLSS4.exe is an official version with additional changes from NVIDIA side, which finally add flip metering support for msBetweenDisplayChange metric. Please take a note that PresentMonDataProvider can consume data from both service and console versions of PresentMon and the service has higher data streaming priority. So to ensure that you stream desired data from PresentMon-2.3.1-x64-DLSS4.exe please ensure that the service is not installed
Added configuration file setting allowing power users to specify additional command line parameters for console mode (e.g. -v1_metrics for forcing condole PresentMon V2 to report V1 metrics or -track_input for enabling experimental input tracking functionality in PresentMon V1)
Added incomplete frames parsing support in console mode
Slightly refactored frames parsing implementation in console mode to simplify integration of future console versions of PresentMon
NaN values for dropped frames are now replaced with zeroes in service client mode to provide unified data format for both service client and console modes
Dropped frames are now excluded from FramerateDisplayed metric calculation
Added burst mode to PresentMonDataProvider's data consumption loop for both service and console versions of PresentMon. Burst mode is activated when PresentMonDataProvider detects that PresentMon's reporting lag increases due to way it is batching frames in applications with extremely high framerate. In this mode PresentMonDataProvider starts consuming data batches from PresentMon without any relaxing sleep periods between batches to compensate increasing reporting lag
Added ovl overlay layout:
The layout demonstrates usage of PresentMon V2 specific data sources and keyboard based overlay transformations. You may press <F> to switch between present-to-present (PresentMon V1) and start-to-start (PresentMon V2) frametime sampling modes in realtime
The layout demonstrates PresentMon V2 ClickToPhotonLatency data source usage to render input events graph and calculate averaged latency for the last 512 frames
Added ovl overlay layout:
The layout demonstrates usage of improved <B> tag for rendering background with rounded colors
The layout also allows you to switch between RivaTuner Statistics Server’s own framerate or framerate sampled with PresentMon via pressing <F> This allows you to use this layout either in traditional application hooking mode or in combination with Desktop Overlay Host to display framerate in applications like Destiny 2, which globally restrict hooking
Now PresentMon data provider calculates timestamps of different stages of rendering pipeline using PresentMon V2 metrics, when at least one PresentMon data source is added to layout, and allows accessing PresentMon timestamps in a way similar to accessing NVIDIA Reflex latency markers:
New presentmonlatency(markerFrom, markerTo) function can be used to calculate a latency between two arbitrary PresentMon latency markers related to the current frame. You can specify one of the following marker identifiers: 0 – input_sample, 1 – cpu_busy_start, 2 – cpu_busy_end, 3 – present_start, 4 – present_end, 5 – gpurender_start, 6 – gpurender_end, 7 – virtual gpurender_start (gpurender_end minus GPU active time), 8 - display. You may also set markerFrom to -1 to specify the very first marker and set markerTo to -1 to specify the very last marker if necessary
New presentmontimespan(markerFrom, markerTo, base) function can be used to calculate and encode a timespan between two arbitrary PresentMon latency markers related to the current frame. It behaves similar to reflextimespan function documented above
Added newovl overlay layout. This layout replicates NVIDIA Reflex Latency analyzer using PresentMon’s latency markers, so you may use it as alternate solution in cases when NVIDIA Reflex Latency analyzer is not available (e.g. on Intel/AMD GPUs or in Direct3D9 or OpenGL application, where Reflex is not supported)
Added new ovl overlay layout. New layout allows you to track frametimes at both rendering pipeline input (present-to-present intervals) and rendering pipeline output (display-to-display interval) sides. In conjunction with custom PresentMon-2.3.1-x64-DLSS4.exe it helps you to monitor framepacing performed by flip metering unit in DLSS4 frame generation mode
Added buf(name, position) function support for accessing desired data source’s ring buffer contents. You may use this function to implement different forms of data filtering. For example PresentMon V2 DisplayLatency data source is designed to report zero for dropped frames, so you may apply (x > 0) ? x : buf(x,-1) formula to it to filter such data and report the previously sampled values instead
Added support for string output format for external data sources. Some external data providers (e.g. AIDA64) export some sensors with plain string info (e.g. IP address, mainboard name etc), which couldn't be connected to overlay editor's numeric data sources. Now you may specify string output format (%s) for such data sources, so OverlayEditor will be able to display it in a text format as is
Added new LibreHardwareMonitor data provider:
New provider was added special for those who want to integrate mainboard specific sensors into overlay layouts. Previously it was strongly recommended to use HwInfo as an ideal free solution for this task, but sadly shared memory became paid feature since v7.0 so if you don't want to purchase it or reenable shared memory support there each 12 hours, you may try to use LibreHardwareMonitor data provider instead
Standalone LibreHardwareMonitor GUI application doesn't provide native shared memory support, so out of box it cannot effectively export sensor data to running applications in realtime like HwInfo, AIDA64 or MSI Afterburner. Instead of that redistributable LibreHardwareMonitorLib.dll .NET library is available and it can be integrated directly into third party products. OverlayEditor wraps LibreHardwareMonitorLib.dll usage into separate LHMDataProvider.exe process, which is effectively a GUI-less LibreHardwareMonitor 0.9.3, providing data to OverlayEditor via shared memory interface. Please take a note that LibreHardwareMonitor installs legacy copy of WinRing0 driver, which can be detected and blocked by some security related software products (e.g. AVAST family products)
Now "with Radeon Graphics" text pattern is excluded by OverlayEditor from %CPU% / %CPUShort% macroses
Added experimental support for AMD Ryzen 9xxx CPUs
Added experimental support for Intel Arrow Lake CPUs
Added quad fan monitoring support for reference design NVIDIA GeForce RTX 50x0 series GPUs
Fixed compact CPU name macro formatting implementation for 16-Core CPUs
Slightly improved algorithm for MPERF based base clock and bus clock calculation implementations. Now the algorithm ensures that both MPERF performance counter and timestamp counter are sampled inside the same scheduler's time slice and filters the values if sampling was interrupted by scheduler
Added new “Lock user settings” option to "Overlay properties" When this option is enabled, OverlayEditor uses new <FNT> tag to override your custom font and zoom ratio settings and use overlay creator’s master settings when rendering this overlay layout. You can uncheck this option and save overlay layout to use your own font or change zoom ratio like before. It is up to overlay creator to decide how to ship overlay now - you may either ship it with "Lock user settings" option disabled to allow user to select a different font and zooming ratio instead of your master settings, or you may enable this option to make your overlay look as intended out of box
Bundled DesktopOverlayHost has been upgraded to v1.4.1:
Added Direct3D10, Direct3D12 and Vulkan rendering backends. New backends do not affect the primary application’s functionality, however now you may use DesktopOverlayHost as a built-in overlay rendering test for all supported 3D APIs in x86 applications
Added optional 64-bit version of DesktopOverlayHost. You may use it as a built-in overlay rendering test for all supported 3D APIs in x64 applications
Added keyboard shortcut for rendering backend restarting. Now you may press <Ctrl>+<Shift>+<R> when DesktopOverlayHost window is in focus to perform full restart of currently selected rendering backed, indicated by short background color flash. Rendering backend restarting feature can be used as universal vendor agnostic overlay stability test, allowing you to check how overlay handles full renderer restart from 3D application side
Added power user oriented config file switch allowing using idle based rendering loop instead of default timer based rendering loop
Updated DesktopOverlayHostLoader helper application can launch both 32-bit and 64-bit versions of DesktopOverlayHost
Added built-in profile for Counter Strike 2
Added built-in profile for "Doom : The Dark Ages". The profile forces RivaTuner Statistics Server overlay to be rendered at DXGI interoperability swapchain level to bypass native Vulkan overlay renderer's incompatibility with Vulkan AMD FSR frame generation implementation. The profile is not necessary if you don't use AMD FSR frame generation, so you may safely remove .\ProfileTemplates\DOOMTheDarkAges.exe.cfg on NVIDIA hardware if you prefer to see overlay rendered at native Vulkan API level
Updated profiles list

ესეც გადმოსაწერი ლინკი: https://www.guru3d.com/download/msi-afterburner-beta-download/ მანდ წინა ფინალური ბილდიცაა და ახალიც, სწორათ უნდა დააკლიკო ზემოთ, ცალკე ახალი RTSS ბილდის გამოდწერა აქედან შეიძლება: https://www.guru3d.com/download/rtss-rivatuner-statistics-server-download/ ეგეც ფინალური ვარიანტია მაინც, ბეტა აღარაა

[George]

MSI Afterburner v4.6.7 beta 1 ბილდი გამოვიდა 22 ოქტომბერს შემდეგი ჩეინჯლოგით:

- Default size of V/F nodes is now adjustable via configuration file, so you can make nodes bigger if default node size is too small and uncomfortable for your eyes
- Now you may zoom work area of V/F curve editor window with mouse wheel in 100%-500% range. Size of V/F nodes is also zoomed in this mode.
- Now you may pan work area of V/F curve editor window by pressing right mouse button and dragging work area with mouse cursor.
- Now you may restore default view mode and reset zoom and pan position of V/F curve editor with <V> keyboard shortcut
- Improved functionality of linear voltage curve interpolation mode with fixed anchor point (<Ctrl> + node dragging). Now this mode can be applied to selection, the farthest node inside selected range is used as anchor point in this mode.
- Added arbitrary linear interpolation mode. Now you may select a source node and press <I> key to activate arbitrary interpolation mode, source node will start flashing to indicate started interpolation process. After selecting the source node simple click destination node to finalize the process, it will result in generating linearly interpolated frequency offsets between source and destination nodes.
- Improved functionality of V/F curve movement with presvered per-point offsets (<Alt> + node dragging). Now this mode can be applied to selection.

ესეც გადმოსაწერი ლინკი: https://download-eu2.guru3d.com/afterburner/[Guru3D]-MSIAfterburnerSetup467Beta1Build16870.rar