AMD/ATI Pixel Clock Patcher

[Black Octagon]

Ah ok understood! Haven't actually downloaded CRU yet (screen is still in customs) so carry on.

Yet I wonder why some people appear to be using the non-ToastyX version of CRU at all?!

[ToastyX]

(12-13-2012 02:11 PM)Randolla Wrote:  I have read your explanation but I do not fully understand it. What is the goal to reach with each number? I know we want to end at 120Hz but what about Pixel clock? Should we try for a lower number or a higher one?
How about horizontal scan? You use 174 in your example. Is that what we should shoot for or should it relate to another number in some way?

The goal is to reduce the pixel clock and horizontal scan rate as low as the monitor can handle. Reducing the horizontal and vertical totals will reduce the pixel clock, and reducing the vertical total will reduce the horizontal scan rate.

Because active + blanking = total and we want to reduce the totals without changing the active resolution, we have to reduce the blanking.

Because front porch + sync width + back porch = blanking, we have to reduce one or more of those values to reduce the blanking.

Front porch, sync width, and back porch can be any values the monitor will accept. The Catleap 2B and the Tempest OC monitors don't seem to care what the individual values are. Reducing the total in my program reduces the back porch because that's what NVIDIA's control panel does.

I mentioned reducing the vertical total to 1450 because that's the simplest most effective change you can make, and it gives you an exact 120 Hz timing. It's possible to tweak it down further, but not by much.

If you're getting intermittent flickering colored horizontal lines, that's a signal quality issue. If reducing the vertical total to 1450 doesn't help, try a better cable. Tweaking it down further probably won't help because you can't go much lower with those monitors, but you can try anyway.

The lowest horizontal total that has been stable for me is 2704. Horizontal values should be a multiple of 8 because they seem to get rounded anyway. 2700 will work, but that just gets rounded to 2704. In this case, the default front porch (48) and sync width (32) don't need to be changed because the total can accommodate them.

If you set the vertical front porch and sync width to 1, the monitor can handle a vertical total as low as 1443, but some video cards can't handle lower than 1445.

[Randolla]

(12-13-2012 06:08 PM)ToastyX Wrote:  

(12-13-2012 02:11 PM)Randolla Wrote:  I have read your explanation but I do not fully understand it. What is the goal to reach with each number? I know we want to end at 120Hz but what about Pixel clock? Should we try for a lower number or a higher one?
How about horizontal scan? You use 174 in your example. Is that what we should shoot for or should it relate to another number in some way?

The goal is to reduce the pixel clock and horizontal scan rate as low as the monitor can handle. Reducing the horizontal and vertical totals will reduce the pixel clock, and reducing the vertical total will reduce the horizontal scan rate.

Because active + blanking = total and we want to reduce the totals without changing the active resolution, we have to reduce the blanking.

Because front porch + sync width + back porch = blanking, we have to reduce one or more of those values to reduce the blanking.

Front porch, sync width, and back porch can be any values the monitor will accept. The Catleap 2B and the Tempest OC monitors don't seem to care what the individual values are. Reducing the total in my program reduces the back porch because that's what NVIDIA's control panel does.

I mentioned reducing the vertical total to 1450 because that's the simplest most effective change you can make, and it gives you an exact 120 Hz timing. It's possible to tweak it down further, but not by much.

If you're getting intermittent flickering colored horizontal lines, that's a signal quality issue. If reducing the vertical total to 1450 doesn't help, try a better cable. Tweaking it down further probably won't help because you can't go much lower with those monitors, but you can try anyway.

The lowest horizontal total that has been stable for me is 2704. Horizontal values should be a multiple of 8 because they seem to get rounded anyway. 2700 will work, but that just gets rounded to 2704. In this case, the default front porch (48) and sync width (32) don't need to be changed because the total can accommodate them.

If you set the vertical front porch and sync width to 1, the monitor can handle a vertical total as low as 1443, but some video cards can't handle lower than 1445.

Thanks Toasty,

I will try this when I get a chance along with a different cable and another port on my 680s.

Thank you again.

[reinhardrudi]

hello again

ok-
i test the beta11 with NET4 - no working
i test the beta11 without NET4 - working

i dontknow why-but itis so.

greez

[Randolla]

(12-13-2012 07:01 PM)Randolla Wrote:  

(12-13-2012 06:08 PM)ToastyX Wrote:  

(12-13-2012 02:11 PM)Randolla Wrote:  I have read your explanation but I do not fully understand it. What is the goal to reach with each number? I know we want to end at 120Hz but what about Pixel clock? Should we try for a lower number or a higher one?
How about horizontal scan? You use 174 in your example. Is that what we should shoot for or should it relate to another number in some way?

The goal is to reduce the pixel clock and horizontal scan rate as low as the monitor can handle. Reducing the horizontal and vertical totals will reduce the pixel clock, and reducing the vertical total will reduce the horizontal scan rate.

Because active + blanking = total and we want to reduce the totals without changing the active resolution, we have to reduce the blanking.

Because front porch + sync width + back porch = blanking, we have to reduce one or more of those values to reduce the blanking.

Front porch, sync width, and back porch can be any values the monitor will accept. The Catleap 2B and the Tempest OC monitors don't seem to care what the individual values are. Reducing the total in my program reduces the back porch because that's what NVIDIA's control panel does.

I mentioned reducing the vertical total to 1450 because that's the simplest most effective change you can make, and it gives you an exact 120 Hz timing. It's possible to tweak it down further, but not by much.

If you're getting intermittent flickering colored horizontal lines, that's a signal quality issue. If reducing the vertical total to 1450 doesn't help, try a better cable. Tweaking it down further probably won't help because you can't go much lower with those monitors, but you can try anyway.

The lowest horizontal total that has been stable for me is 2704. Horizontal values should be a multiple of 8 because they seem to get rounded anyway. 2700 will work, but that just gets rounded to 2704. In this case, the default front porch (48) and sync width (32) don't need to be changed because the total can accommodate them.

If you set the vertical front porch and sync width to 1, the monitor can handle a vertical total as low as 1443, but some video cards can't handle lower than 1445.

Thanks Toasty,

I will try this when I get a chance along with a different cable and another port on my 680s.

Thank you again.

Ok Now I have a new problem. It seems that whatever timings I choose, I cannot get 120Hz to stay at 32 bit color. It either reverts back to 16 bit or it reverts back to 115hz. Any suggestions? It seemed to work before.

[ToastyX]

I've only seen that happen when adding custom resolutions through NVIDIA's control panel. You have to choose the resolution under Custom instead of PC.

[Randolla]

(12-14-2012 02:10 AM)ToastyX Wrote:  I've only seen that happen when adding custom resolutions through NVIDIA's control panel. You have to choose the resolution under Custom instead of PC.

I think I fixed it. Don't ask me how. I did something last night that worked.Thanks for the help.

[walley]

Not sure if this will help but mine kept jumping back to 16 bit color as well. What I noticed was that the last thing I had placed my cursor was on 32bit. Then I would have to go into my screen resolution to set the desired hz and it would jump back to 16. I was scrolling it by accident. If you click on something else or close out of nvidia control panel you can't do that. It was doing this for every resolution I made till I saw what I was doing. Pretty simple ,but not very obvious.

[Black Octagon]

Hi ToastyX,

I just hooked up my Tempest OC. However, I'm having problems with the patcher. I'm running the latest Catalyst 12.11 Beta 11 drivers, with .4 NET support (plus the latest CAP profiles).

The patcher itself runs fine, but when I then reboot, Windows BSODs during startup, and continues doing so until I boot in Safe Mode and unpatch.

Event Viewer gives the following info for the BSOD:

Log Name: Application
Source: Microsoft-Windows-Winlogon
Date: 18/12/2012 9:33:19 PM
Event ID: 6000
Task Category: None
Level: Warning
Keywords: Classic
User: N/A
Computer: Oli-PC
Description:
The winlogon notification subscriber <GPClient> was unavailable to handle a notification event.
Event Xml:
<Event xmlns="http://schemas.microsoft.com/win/2004/08/events/event">
<System>
<Provider Name="Microsoft-Windows-Winlogon" Guid="{DBE9B383-7CF3-4331-91CC-A3CB16A3B538}" EventSourceName="Wlclntfy" />
<EventID Qualifiers="32768">6000</EventID>
<Version>0</Version>
<Level>3</Level>
<Task>0</Task>
<Opcode>0</Opcode>
<Keywords>0x80000000000000</Keywords>
<TimeCreated SystemTime="2012-12-18T20:33:19.000000000Z" />
<EventRecordID>17840</EventRecordID>
<Correlation />
<Execution ProcessID="0" ThreadID="0" />
<Channel>Application</Channel>
<Computer>Oli-PC</Computer>
<Security />
</System>
<EventData>
<Data>GPClient</Data>
<Binary>D9060000</Binary>
</EventData>
</Event>

I don't understand any of this...don't suppose you can suggest a solution?
Try the latest WHQL drivers, perhaps?

Many thanks!

[ToastyX]

That's weird. I haven't heard of that happening to anyone else.

That event information is for something else. Try Blue Screen View to get information about the BSoD: http://www.nirsoft.net/utils/blue_screen_view.html

Use that to get the bug check string/code and the caused by driver/address information.