ClockBlocker (profiled AMD power-management control)

Code:

SHA1                                              Size  Name
---------------------------------------- -------------  ------------
574DB09D97702887BB7E9AC7DF969FE668AC4791       4252081  YAP_BETA_Setup.exe

YAP 1.0.6 BETA (hotfix):

-More Windows 7 DEP workarounds
-Windows 7 specific OpenCL fixes (helper module -- ClockBlocker)


I understand that Windows 7 is extremely important to have supported with anything for gamers, in that it's one of the best and most used gaming platforms for older games.
Sadly right now most of my dev systems are on 8.1, 10, Server 2012 R2, & Server 2016. So, this is my call out for help, I need more Win7 testers.

If you've an interest in YAP and use Skype, let me know and I'll add you. Whatever can be provided helps, reports of things that work, things that don't work, screenshots of error-popups or the event-log (such as if the server service won't start). --Anything at all really, but especially if you're getting "StackHash" type errors.

 

Just a personal opinion...no hard feelings, you doing great job..

The UI (Colors) are a bit weird (mess?) to me...have u consider work only with 2-3 colors?

1) Backround
2) Font
3) Other critical stuff = red?

 

Bullet point #3 got me thinking that I actually broke my coloring scheme, which is pretty true.
--RED usually means "BAD" in places where I've used it or to make the user think for a second before clicking a button.

For example:
-errors are always red (with the exception of IPC transferred messages -- eg, helper)
-cancel buttons are always red, since canceling could lose work.
^ same applies on "Reload" with server side settings as the reload will wipe any entered data

At the same time the menu-bar is 'RED', Firewall "Defaults" button is not-red (and probably should be), the "Defaults" buttons in all settings pages are not-red.
* Yep, that's good feedback, this will be changed to be more consistent. (When I selected the red menu bar coloring, that was more a tribute to red-team and AMD than anything else)

Probably the inter-process protocol should be updated to allow coloring.


On #1 and #2, I've been going for a mix of light-theme and dark-theme. That probably seems weird to most people, although what I was hoping for was to get people using the tool to choose one or the other theirself (eg, to go in and customize).
--Maybe I can create some presets / unified light or dark themes ready-to-go as starting points.

 

Code:

SHA1                                              Size  Name
---------------------------------------- -------------  ------------
96312E69E7F3958AD63B4CCC67EFC97E07BBE075       4345685  YAP_BETA_Setup.exe

Hopefully getting close to the end of the ugly issues on Windows 7.. rigorously playing monkey with YAP on this end trying to break things and cause crashes. (including damaging the OS install)

YAP 1.0.7 BETA (hotfix):

-Fixed an intermittent startup crash in the server process. (Windows 7 specific)
-Fixed a case where corrupt text was displayed for PIDs under 4 - eg, System process (Windows 7 specific)
-Fixed "Illegal Instruction" exception when run under Wine.
-Added more safety checks for unique instance locks
-Coloring scheme changes:

RED in YAP is supposed to capture the focus of the user. This usually means "critical-stuff", and that somehow got lost over time with YAP.

To keep with the above design goals, the following color changes have been made:
-The menu bar has been changed to a deep-blue.
-The "Apply" button for filters has been changed to green (confirm),
-The firewall "Defaults" button has been changed to red.
-All "DEFAULT" buttons on each scrollable settings page has been changed to red.
-The Blacklist button (Recent Connections) has been changed to green (confirm selection).
-The Refresh button (Recent Connections) has been renamed to Reload, to match other dialogs.

By default, the main status of YAP is now a teal / green theme that's half dark (log, graph) and half light (process list). Please be aware that some color settings are stored in the user's client-config.
---Color settings can be changed to the new scheme selections by using the settings page default buttons.

 

Code:

SHA1                                              Size  Name
---------------------------------------- -------------  ------------
94B9919B6963D8B3579F44AC8A4D580ECFB3852C       4346721  YAP_BETA_Setup.exe

YAP 1.0.8 BETA (hotfix):

-Fixed an issue that prevented YAP from adjusting priority (immediate mode), affinity (immediate mode), or triggering rules (at all) on game processes in the presence of Battleye and XignCode3. (Effected Windows 7 and 8.1)
-Fixed a rare race condition in the YAP client's netcode.


EDIT: Seems the main site is having routing issues ... I've accordingly updated the mirror with YAP's contents for those who have problems reaching it.

 

Code:

SHA1                                              Size  Name
---------------------------------------- -------------  ------------
78CD186D91E9C853BE84314BCEBEEB1427913ED8       4341777  YAP_BETA_Setup_1_0_9.exe

YAP 1.0.9 BETA (hotfix):

-Processor affinity now works on systems with more than 32 logical cores. (AMD EPYC, etc)
-Fixed an issue in the server's evaluator caching which could result in a wrongly ordered rule being applied.
** In the case of multiple rules overlapping this should always be top-down, even-with parallel rules.
-Various "minor" GUI fixes.

 

Code:

SHA1                                              Size  Name
---------------------------------------- -------------  ------------
A4C5E9EABFD7DF9CF0C08B9708AF91EE4F69F01D       4343470  YAP_BETA_Setup_1_0_10.exe

Keep those bug reports coming guys! I don't necessarily think to test things like running games off a RAMDISK.


YAP 1.0.10 BETA (hotfix):

-New methods of querying process path. YAP should now be able to see the full-path of protected games.
-Fixed an issue where process icon is not shown for those launched off a Direct-IO RAMDISK.

 

not matter what I cant get clockblocker to rise my gpu clocks.

it used to worked before, but that was ealier in the year....

chrome keeps causing tdr with out it. mostly when watching youtube videos and web browsing.
gpu clocks keep going up and down.

and clockblocker fixed that issue. but not anymore as the clocks dont rise and stay

tried both yap and clockblocker

 

If all solutions does not work...why u just do not edit your gfx bios a little and write it in?

 

Cause I don't know how.
Need somebody help me with doing so. Finding correct firmware for saphhire 390x nitro,editing and flashing gpu

 

You can define minimum allowed clocks by rightclicking the curve points in Wattman.

 

Changing clocks through vbios not easy as people say.
Changing value and reflashing has no effect.

hex editor needed to editor values.....makes my head hurt...lol

found that OverDriveNtool works...


problem I was having was using chrome to watch youtube and use twitter at same time, card would stay in powerstate 0 and would lead to a TDR.

 

Ive downloaded ClockBlocker long time ago and messed up with settings in a way that system get black screen whenever I try to run this program.

Is there a location to clean settings?

 

Yes, you'll find ClockBlocker's settings at "%USERPROFILE%\Documents\ComroeStudios\ClockBlocker". Wiping out that folder should cause it to be re-created and revert everything to stock.

- Alternatively you can rename the ADL helper (within the program folder), which will prevent applying your settings and permit you to start up ClockBlocker. eg, to disable ADL or otherwise edit your profiles accordingly. After which you would simply rename the ADL helper back.


Or there's the emergency feature in ClockBlocker designed for this purpose:
From the change-log notes,

-"SHIFT" may be held during startup as an emergency halt to using custom clocks. (this will uncheck Enable custom clocks)


Hopefully one of those last two help you not lose everything settings wise.

 

Hello!

I'm trying to use ClockBlocker to stop my R7 250 1Gb GDDR5 (Oland, GCN1) from clocking down when playing certain 2D games such as Spelunky and SteamWorld Dig. With Spelunky my card stays at 300/150 core/mem and on SteamWorld Dig the clock fluctuates between 2D and 3D clocks causing bad frame rate drops.

Testing ClockBlocker I've set the default rule to block. The tray icon turns red, the program generates a 0.1% load on the GPU in Task Manager but the clocks remain at 300/150 in Afterburner. I've tried both the default and alternative compute methods.

Using the ADL method results in an immediate crash with the monitor showing vertical white and grey bands across the whole display, requiring a reboot. This is exactly the same crash I get if I attempt to use the '3D Clocks only' button in the Asus GPU Tweak II software.

When testing the only programs open are ClockBlocker and MSI Afterburner to confirm clocks. MSI Afterburner is set to default with no overclocking.

I don't know if it's relevant but on the tool-tip for the blocking method it says if it doesn't work enable WattMan. The R7 250 doesn't support WattMan and still has the old "Overdrive" page in Radeon Settings. I'm using the newest 19.2.3 drivers.

Thanks!

Code:

2019/68 09:27:15 PM >> Successfully read file "C:\Users\Us\Documents\ComroeStudios\ClockBlocker\ClockBlocker.arc"
2019/68 09:27:15 PM >> Verifying integrity of main and backup config...
2019/68 09:27:15 PM >> Config file integrity looks-good
2019/68 09:27:15 PM >> Bypassing startup version-check
2019/68 09:27:15 PM >> Initialized
2019/68 09:27:15 PM >> Created TrayIcon
2019/68 09:27:15 PM >> INIT_SharedMem():: INIT shared memory
2019/68 09:27:15 PM >> INIT_SharedMem():: Setting up IPC initial params
2019/68 09:27:15 PM >> INIT_SharedMem():: Signaling any resident helpers
2019/68 09:27:15 PM >> INIT_SharedMem():: Unlocked, INIT -- OK
2019/68 09:27:15 PM >> Spawned process-scanner thread
2019/68 09:27:15 PM >> ThreadProcessScanner:: Delaying 4 seconds
2019/68 09:27:15 PM >> Spawned dispatcher thread
2019/68 09:27:15 PM >> Spawned IPC Listener thread
2019/68 09:27:15 PM >> ExecuteProcess [ADL Helper] -- success [0]
2019/68 09:27:16 PM >> Helper-Module [*ADL*] => module version = v1.3.49
2019/68 09:27:16 PM >> Helper-Module [*ADL*] => Spawned coordinator thread
2019/68 09:27:16 PM >> Helper-Module [*ADL*] => ADL adapter querying done...
2019/68 09:27:16 PM >> Helper-Module [*ADL*] => Adapter [0]: Main INIT done...
2019/68 09:27:16 PM >> Helper-Module [*ADL*] => Adapter [1]: Main INIT done...
2019/68 09:27:16 PM >> Helper-Module [*ADL*] => Acquired Lock, Syncing SharedMem State
2019/68 09:27:16 PM >> Helper-Module [*ADL*] => Adapter [0]: Try# [1]: Successfully set normal performance levels
2019/68 09:27:16 PM >> Helper-Module [*ADL*] => Adapter [1]: Try# [1]: Successfully set normal performance levels
2019/68 09:27:16 PM >> Signaled terminate helper container...
2019/68 09:27:16 PM >> Helper-Module [*ADL*] => Destroying threads, cleaning up...
2019/68 09:27:19 PM >> ThreadProcessScanner:: Done waiting, signal main-thread to start timers.
2019/68 09:27:19 PM >> ThreadProcessScanner:: Continuing...
2019/68 09:27:19 PM >> MAIN:: UI-Thread => Starting timers
2019/68 09:27:19 PM >> =+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=
2019/68 09:27:19 PM >> Rule Action: "Block"
2019/68 09:27:19 PM >> Rule Criteria: "DEFAULT"
2019/68 09:27:19 PM >> =+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=
2019/68 09:27:19 PM >> ExecuteProcess [Compute Helper] -- success [0]
2019/68 09:27:20 PM >> Helper-Module [*COMPUTE*] => module version = v1.3.5
2019/68 09:27:20 PM >> Helper-Module [*COMPUTE*] => Spawned coordinator thread
2019/68 09:27:20 PM >> Helper-Module [*COMPUTE*] => Adjusting timer resolution
2019/68 09:27:20 PM >> Helper-Module [*COMPUTE*] => Coordinator: Recognized 1 target-device candidates
2019/68 09:27:20 PM >> Helper-Module [*COMPUTE*] => Coordinator: Rebuilding worker pool
2019/68 09:27:20 PM >> Helper-Module [*COMPUTE*] => Worker pID=0, dID=0: Thread-INIT
2019/68 09:27:20 PM >> Helper-Module [*COMPUTE*] => Worker pID=0, dID=0: Checking...
2019/68 09:27:20 PM >> Helper-Module [*COMPUTE*] => Worker pID=0, dID=0: Method = reliable
2019/68 09:27:20 PM >> Helper-Module [*COMPUTE*] => Worker pID=0, dID=0: Setup successful!
2019/68 09:27:24 PM >> TrayIcon - Open menu

 

Hey there f300,

Taking a quick look through that program log, I do notice a discrepancy with the compute Helper's init in contrast to the ADL's (specifically the two adapters found). Not sure that your R7 is actually being used (despite the load that you're noticing on it), there's a chance that it's not .. which would explain why you're not seeing the clocks get kicked up in testing. I'll admit that I have not recently taken a look at the ADL Helper's functionality, so I would err on the side of caution and prefer to use the 'compute' for now.


For the sake of experimentation, try renaming (or deleting) the OpenCL ICD ("OpenCL.dll") that my installer bundled (from the program directory) and then re-test ClockBlocker's Compute method and see if there's "two" target device candidates this time around -- and or if you see the clocks jump up afterwards. With the local library not found, whatever your installed drivers have provided will be used instead (and the results may be different).

 

Hi everyone!
I'd like to use Clockblocker on my Vega64strix, but with no success at all...I tried many configurations in the software, but each time i try to "block" the clocks, the small tray icon turns to red and nothing happens with the clocks. Does Clockblocker even works with vega64 cards? Anyone managed to make it work?
Have a good day!

 

Yes, ClockBlocker works with Vega (including the frontier edition on pro-drivers where ADL control is locked down).


Please see my above response to f300: I've since then reproduced an issue with the older OpenCL ICD that was bundled (with both YAP and ClockBlocker) which causes the Compute Helper to not work properly in multi-GPU setups. Bear in mind that multi-GPU can include on-processor or motherboard IGP's, and the various hybrid graphics solutions as well. There may be more conditions where devices are not being properly discovered.

As a short term solution, 'remove' OpenCL.dll from the ClockBlocker program dir (which will cause a fallback to any "newer" installed system version). I strongly suggest sticking with the Compute blocking method as the ADL has had no testing on Vega, nor on modern drivers.

 

Thank you very much for helping! Unfortunately, removing the opencl.dll doesn't change anything here, with compute method (or others).
I'd like to make it work because when i set my P7 to minimum in Wattman, i'm not sure the frequency is really my P7, appears more like a p6 (and frequencies differs for each game/benchmark)
Any other idea to make clockblock work?

 

Thank you for your reply A2Razor!

I've tried deleting OpenCL.dll from the ClockBlocker program directory but unfortunately it does not help, the clocks still remain at 2d speeds.

Just to say multi-GPU should not be a factor on this system, I only have one GPU installed (the R7 250) and the CPU (an old Core 2) does not have an IGP and neither does the motherboard (Asus P5K, P35 chipset).