(12-15-2013 02:39 PM)ToastyX Wrote: [ -> ]No workaround other than getting a newer video card.
Oops, I guess I should have mentioned that this is occurring on my spanking new 780 Ti. Currently it sits at 384 MHz and 27 C idle. As soon as I change the refresh rate it jumps to 849 MHz and 47 C with the default fan curve.
I'm actually really glad this problem is listed in this thread. It was driving me nuts trying to figure out what was going on. It was sheer luck that I noticed the GPU clock being way too high at idle, then decided to try using CRU instead of a custom resolution in the NVIDIA Control Panel. I was afraid something was broken.
Regardless, I'm guessing there's still nothing that can be done, but I wanted to at least mention it. If this doesn't occur with other 600 and 700 series cards, it appears it crept back in either the 780 Ti or driver version 331.82.
I haven't heard of newer cards being affected by this limit. Does it clock down at 404.00 MHz pixel clock but not 404.01 MHz? If the limit is not exactly 404 MHz, then it's an unrelated problem.
Make sure the vertical total is high enough. The video card might not clock down when idle if the vertical total is too low because it needs enough time between frames to avoid screen flicker when changing memory clocks. I don't know what the lower limit is for NVIDIA cards, but I think 1475-1500 is enough. AMD/ATI cards require the "LCD standard" vertical total to clock down, which would be 1525 at 120 Hz, but that raises the pixel clock to near 500 MHz.
The video card also won't clock down if you have two monitors connected.
(12-15-2013 11:02 PM)ToastyX Wrote: [ -> ]Make sure the vertical total is high enough. The video card might not clock down when idle if the vertical total is too low because it needs enough time between frames to avoid screen flicker when changing memory clocks.
This was exactly the issue. I had it set to 1450 like I did on my previous video cards. I played around and it looks like 1458 is the cutoff. Any lower and it won't down clock. Thank you very much for the tip. Sending a beer your way.
I didn't even realize there was a cutoff like that. Is there a guide you'd recommend for setting up manual timing? I'd like to make sure I know what I'm doing and there's no other nuance I'm missing.
Hi ToastyX,
Would like to know if it's ok to update Nvidia gpu driver to 331.82. I have updated to 331.65 and it's alright. But playing BF4 with latest BF4 patch/updates now there is this annoying prompt before game starts, saying recommended gpu driver 331.82 or later, very annoying. And want to confirm if your NVidia Pixel Clock patcher works with 331.82? So, what's your take on this? I am using EVGA GTX 690.
Thank you.
Yes, it works with 331.82, and it should work with 331.93 as well.
(12-15-2013 11:18 PM)Sycobob Wrote: [ -> ]I didn't even realize there was a cutoff like that. Is there a guide you'd recommend for setting up manual timing? I'd like to make sure I know what I'm doing and there's no other nuance I'm missing.
That's more of a video card limitation rather than a monitor timing issue. There isn't anything else to worry about. Just use whatever works with the monitor.
You can try tweaking the horizontal total if you want to reduce the bandwidth a little more. The lowest horizontal total is usually a multiple of 8 - 4. The Catleap/Overlord monitors can usually handle a horizontal total of 2700 at 120 Hz, and I think the QNIX/X-Star monitors can handle 2652.
The only other thing to consider is the total = active + front porch + sync width + back porch, so if you wanted to reduce the vertical total below 1449, you'd have to reduce the front porch and/or sync width to fit within the total. Those monitors can handle as low as 1443 with the front porch and sync width set to 1, but some video cards can't handle less than 1445.
(12-15-2013 11:18 PM)Sycobob Wrote: [ -> ] (12-15-2013 11:02 PM)ToastyX Wrote: [ -> ]Make sure the vertical total is high enough. The video card might not clock down when idle if the vertical total is too low because it needs enough time between frames to avoid screen flicker when changing memory clocks.
This was exactly the issue. I had it set to 1450 like I did on my previous video cards. I played around and it looks like 1458 is the cutoff. Any lower and it won't down clock. Thank you very much for the tip. Sending a beer your way.
I didn't even realize there was a cutoff like that. Is there a guide you'd recommend for setting up manual timing? I'd like to make sure I know what I'm doing and there's no other nuance I'm missing.
Thank you guys! I was trying to figure out why my cards(770s) weren't at there usual idle clock speeds for awhile now. I forgot I had adjusted the pixels to 1450 to stop a few lines from popping up onscreen when I upped the refresh to 120Hz. I changed it to 1458 and my cards idle where they should now.
I followed your guide and had my monitor oc'd to 120hz for my 2 - 7950's. I just got 2 -770's and I'm having some problems. Cards install fine (using 331.93 drivers) and everything is good until I try and patch/add new refresh rates with CRU. Now even after restoring the patch to default, I'm getting black screen freezes, as if there is no input to the monitor. This is the case even at the default refresh rate of 60hz.
Any suggestions? I've tried your 1.2.3 patch and as soon as I did that was when the problems began.
Did you use the full patch? It's required for SLI to function properly.
(01-01-2014 11:24 PM)ToastyX Wrote: [ -> ]Did you use the full patch? It's required for SLI to function properly.
Yes I did. I'm pretty sure I have a bad card. I got the monitor overclocked with the first card on its own, but when I added the second it crashed immediately. I took out the first card and ran just the second one on its own and it crashed with 331.93 drivers. I tried booting again and it wouldn't even output video to the 1440p monitor. I could output through HDMI to a 1080p monitor, but only at a default 800x600 resolution. The 331.93 drivers weren't being used either, and I couldn't access the nvidia control panel even if I tried reinstalling the drivers.
Fairly certain at this point that it's the card.
EDIT: I decided to try the cards out using driver 331.82 and so far so good when using your full patch. I hope I don't have the same issues I was having yesterday and hopefully this fixed it.
After running 3dMark11 for a while, it crashed again. I had hoped that 331.93 was just buggy and that 331.82 would fix it. Once again I'm fairly certain it's the card.