ClockBlocker (profiled AMD power-management control)

To be honest I didn't tested your tool, but thanks for the post.
I didn't realized that there are problems with corruptions/artefacts with those cards.
And it worries me, as I planned to buy fury card...ATM Im using only iGPU with Benqa xl2720x so upgrade is must, I want to see that 144Hz...
That case I might turn to 980, but I rather wouldn't do that.


Sry 4 english

 

Hello,

unable to get it to clockblock (great name)

Windows 10 X64
r9 290x
Driver Crimson 15.11.1 (9.14.10.01162)

I have tested with firefox setup for the firefox.exe
I also tried it with NBA2K16.exe and also just with the gaming running full screen.

I tried setting the short cut to run under windows 8 compatibly mode with no luck.


I am using MSI afterburner 4.20 to monitor my GPU usage on my second monitor.

Thanks keep up the great work. let me know if there something else I could try.

BIG_FREAK

 

Hello, great idea for this APP. Greatly needed haha but it is not working on my end.

Trying to get it to "clockblock" AC Syndicate but my core is still bouncing all over the place.

I have the Sapphire Fury Tri-X btw

 

Hi,
just to report that this tool works with my R9 270 (non-x) on Win7. I want to ask did this tool can use 3D clock/profile from AB?
http://imgur.com/DsREuek

 

Agree.

Ppl with less than optimal cooling solutions or in locations who have high ambient temp should at least use an AB fan temp customized to their local specific conditions if they want to block (lock) his GPU timings with this tool or AB profiles.

Relying only on thermal throttling to lower temp in all cases is not good at all for the GPU.

@A2Razor

Thanks to spend your time in this tool and share it with all of us.

As PR said the need of tools like this one or RadeonMod expose how little effort AMD put in giving us basic control over our GPUs and instead spend a lot of time delivering fancy graphics drivers GUI, social connectors and a worst gameplay/multimedia playback in the pursuit of lower watt usage.

I guess AMD consider his "green" crazyness a top priority for his GPUs.

I think the customers who buy these GPUs normally use them to play games and want performance to be top priority over watts spend...or at least have the option to have this performance and stability back deactivating the "green fetures" right from the fancy drivers GUI.

 

Having a hard time seeing the clock graphs from that thumbnail shot. If the card went to full clocks and pegged for a few seconds, then likely my CL workload is actually being generated and had some-type of impact (briefly). That's at least a good sign.


It looks like there's only the two default rules in there, I'm assuming the tray icon was RED for this entire time -- including the time after the clock bump where the card downclocked. So, based on that I'd say the drivers didn't respond to this workload the same for your systems as they do for mine.

Which OS is this? And which driver version?


--I'm currently working on an alternate trick (to the OpenCL one) that achieves a very similar effect. Hopefully it'll work on systems where the current trick does not.

 

Of curiosity, is this a TH1 or TH2 install of Windows 10?

--If it doesn't work for you on the desktop, with the DEFAULT profile edited. (looks like you tried setting the default to block already) Then there's not much you can do with the current version.


I see no errors in the log at least, so the CL setup was successful. I'll have to think on this a bit, but you'll probably have to wait for the next release and alternate trick that I'm working on.

EDIT: Of curiosity, how many 'device candidates' does the log show? (given this is off the screen)

 

Indeed, but on this note it's a good practice for anyone to monitor their temps, especially due to the current driver "issues" that AMD has had with fan controls (on the stock ramp). I'd advise everyone [using this tool or not] to keep an eye on their temperatures. Thermal throttling happens at very high temps, and it's certainly not good to let the card run at those temps to begin with.


Every graphics card owner owes it to theirself and their hardware to get a tool that can monitor temps. If you do, you have a chance to notice premature failure or problems before the card actually breaks. It's far easier (and cheaper) to replace a fan for example than to replace a 500$+ graphics card.

But yeah, this tool won't allow you to exceed power limits or anything like that to reitterate.


Far as power-management. I actually do think it's important to have working power-management, yet AMD is getting it wrong in that they're being far too restrictive: I don't think anyone necessarily wants to run their cards at maximum clocks when idle on the desktop, surfing the web, or while doing work, but the thing is that AMD needs to understand that there's many uses for a PC. People have different priorities at different times and under different usage scenarios (like gaming competitively), user-choice really needs to be allowed.

Another way to look at it is from the competitive standpoint, AMD is giving its users far less options than their competition (and this really needs to change). NVidia has had driver provided power-profile support for years.

AMD's overall intentions may be good, but power users and gamers all love choice. It's a great selling point for products that I hope they don't overlook.

 

Yes, you can use custom clock profiles in Afterburner. That shouldn't cause any problems at all to use the two tools together.


If it does break something, please let me know. At least in principle I don't think it will.

 

I'm locking the core clock with afterburner and somehow my top card hit 94 degrees without me knowing it. After a while i checked afterburner and the core clocks still throttled to maintain a steady 94. So locking the core clock won't have any effect on the thermal throttling. At least not for my card.

 

Great idea, but unfortunately it's not working on my system.
Windows 10

 

Sorry for being so briefly.
Yes the tray icon stay red after the clock downclocked
Im on Windows 10 Pro x64 TH2
The driver is Crimson 15.11.1 Update 11/30/2015

 

you make sure u are using a modified fan profile?

also this is kinda neat, i did some tests with power saving stuff and counter strike

https://www.youtube.com/watch?v=Y7OKN_gn_bQ

 

Not working for me

Windows 10 pro x64 th1.

When you say "red icon" what do you mean by that ? The Icon I get is like a black round thing with a red cross over it.

 

The black part of the tray-icons is a silhouette of a rooster. When blocking, the background color around the silhouette changes to red, with a cross (X) in a light blue through the center. This is intended to symbolize blocking.

I'll add images of the tray-icons in to the guide later on if this wasn't too clear, or maybe add a tooltip description of the status in addition to the image.



Also not given up on you guys that this isn't working for. I'm taking a deep look at what I'm doing, and also the way the AMD's drivers respond to CL and DirectCompute workloads (to see if it's something simple that I can fix).

-This may take me some time though, especially on the alternate method.

 

Hey, thanks for the bigger image. That's great and I can see everything now.


-Are those spikes in your clockspeed when you toggle the tool off and then back on (by repeated editing of the DEFAULT rule)? Or, for instance did you re-open the program on each of them?

 

That is indeed good to read.

We have here only one user who experienced this kind of problem (on nano). But in general many users had/have issue where image on screen jumps up a bit, shows corruption and stays that way till you change resolution or re-plug cable to monitor/graphics card. (frequency of this issue was very rare, maybe once a month in average)
It did happen to me few times, but it is fine for very long time. Either driver fixed it or it was caused by bad connector connection and therefore fixed by one of re-plugs.

 

Depends on the people, some have it rarely, in a weekend (while the pc is on 24h a day) i'll get it a least twice.
Nothing to do with cable. I think it's more to do with the minimum voltage, I suspect it's going below 0.900 Mv and that interferes with the HBM, with all dies not being equal, i'm thinking it's happening more often to others due to that.

But i'm in the phase of testing my theory with GPU-Z, to see what the gpu does exactly when the corruption occurs.

 

1.0-5 released. Same download link as before.

--Changelog on web-documentation and also in the main forum post here.

This is an interim release while still experimenting with a DirectCompute approach. If anyone is feeling adventurous there are some improvements in here that might help on machines that didn't work prior (hard to say).

 

I have tested the new release (1.0-5). I have a Fury Tri-X.

OS is Windows 10 Pro x64, build 10586.29.
Driver is the "Crimson" hotfix.

1) This release solved the issue people in this thread were reporting (would run at max clocks for few moments then downclock back) with few oddities:
1.a) It doesn't really honor overclocked settings. It would fluctuate between stock clocks and overclocked settings but never going below stock clocks.
1.b) Stock clock works just fine. Solid 1000Mhz at all situations.
1.c) I honestly think there's nothing to fix before people says fluctuating clocks is an issue. Not downclocking below stock clocks is more than enough for me.

2) Situations where driver would downlock for no reason (GTA V Online, when you enter/leave elevator in an high-end apartment) and FPS drops dramatically. This tool completely solves the issue.

3) Fury X owners, would this tool keep a solid 1050Mhz core clock? GTA V Online if someone plays it, it's the king of downclocking because of the zillion of the CPU-Bound situations we have, ranging from traffic jams to hacked lobbies where you would find dozens of jumbo aircrafts laid down the streets.