Discussion in 'Videocards - AMD Radeon' started by OnnA, Oct 3, 2016.
So far, nobody has gotten this to work with a DisplayPort adapter.
Has anybody encountered the monitor losing sync (going black) but now found a way to work around that
my freesync hdmi monitor really a b***h , only lower it by 1 Hz, and it starts black flickering in-game :/
48-72 Hz freesync range
OK, just tried freesync on my CRT using an HD Fury Nano GX adapter. Set my freesync range to 60-85 and enabled it in Crimson. Tried a couple games (Mirror's Edge Catalyst and PvZ Garden Warfare 2) Anytime I dipped below 85fps in a game, my monitor blanked out. Point my mouse towards the sky (to get above 85fps), picture would return. My desktop resolution was 1600x1200@85hz.
This is what I thought would happen. If anybody else can get different results, I'd love to hear about it.
I Have HDMI->VGA Active Vivanco Converter with ~300MHz Pixel Clock
I Have Fury (Fiji) GPU so it have only DP and HDMI ports, i need converter.
Im waiting for OLED or HDR Monitors (I need Colour and Contrast )
Try 57-85 (but in CRU set 57-120) then try again some games.
My CRT Handles this very good TBF (i don't see any fliping at all)
If i have this -> Then FS will be OFF for sure ;-)
Thanks, I'll try that tomorrow.
Cool side note, I realized that when you add the HDMI block to a VGA connection, you can use VSR! Very useful since in many games my crossfire 380x setup can exceed my maximum refresh rate for a given resolution on my CRT.
I can't do that :bang:
Tell me what resolution i need to put in CRU to have VSR?
I know it uses 16:9 & 16:10 but for my 4:3 or 5:4 is a no go
For this to work on DP you would need a monitor with DP 1.2a (supports the Variable Vblanking header). The scaler/TCON vendors ships their product with the header field intact rather than having two SKUs. If monitor vendor didn't disable the header value or complete remove the headers then VVblanking should be able to work. AMD shouldn't be policing this on their side either because of the way monitor and video card interactions are spec'd to work. That is to say the monitor reports it's capabilities to the video card via EDID. Notably this is why Nvidia G-Sync is a product that is not following established specs. Anyways, AMD's Freesync implementation *should* accept the monitor capabilities without question. There could be some negatives though. A non-Freesync branded(ie AMD certified) monitor will not have any specific tweaking to a model of monitor. You could get aberrations like ghosting because overdrive compensation isn't employed correctly to account for the VVblanking intervals.
Also, this should be really easy to verify if this will work on a monitor. You just load up one of the many EDID reader programs. You can see the header support. Another thing I just thought of is that you can make your own .infs with many of these programs. If the VVBlanking header was present, but disabled theoretically you could just flip the bit to enable and set any supported range related fields in a custom .inf. Install the .inf and voila!
you mean this?
Are you talking about the "Vertical Total" value that defaults to 1125 at 60hz, 1133 at 100hz, 1144 at 120hz? (1920x1080p resolution?).
I know the Benq XL Z monitors support this, but the monitor will report strange refresh rates if the VT matches another established VT sometimes. In fact the Benq Z series monitors can go up to VT 1502, however once you exceed VT 1170, you will get an "Out of Range!" error, but the screen will still be visible. Then once you go past VT 1250, the error will disappear. But some VT's will cause stuttering in the higher ranges, while maybe +1 higher or lower will be fine.
Once the VT is past 1380, the "Out of Range!" error appears again with the screen still showing up. Once you go past VT 1410, the monitor black screens with an error. Then at VT 1481, the screen reappers with the OOR error again, blacks out at VT 1486, then suddenly appears with no error at all from VT 1497-VT 1502.
I think this VT blanking range is used by Lightboost mode (3d mode itself) as if you are using Benq Blur Reduction, the "Crosstalk" amount on the screen when using VT 1500, Strobe Phase=000, and Strobe Duty 009, is identical to Lightboost mode, literally pixel for pixel (at the bottom of the screen).
I don't remember any sync errors with a 'low' vertical total (VT 1125 to VT 1160) on my XL2720Z, however enabling blur reduction with a refresh rate that has a vertical total that matches a different refresh rate will cause the monitor to report the "incorrect" refresh rate, with strange strobing errors--e.g. Double strobing below strobe phase 040, then single strobing at strobe phase >041.
If the Vertical Total does NOT match a known vertical total value for an established refresh rate, the monitor reports back as 60hz, and the "strobe persistence values" (strobe duty) revert to 0.167ms per pulse width. (16.7 divided by 100).
Did I lose anyone here?
I think the Benq can accept some strange vertical totals...depends. a few cause problems.
1920x1080@100hz ->default VT=1133.
1920x1080@85hz ->default VT=1124
1920x1080@120hz: default VT=1144.
Using 1920x1080@100hz, VT 1125 ->monitor scaler reports 1920x1080@100. (NOT THE OSD! the factory menu which is unlocked by powering off, holding button 3 and 4 for 5 seconds, and powering on with the buttons still held <script id="gpt-impl-0.13568881218555945" src="http://partner.googleadservices.com/gpt/pubads_impl_100.js"></script>down. This is not the service menu either (the strobe settings are in the service menu).
1920x1080@100hz: VT 1124 ->scaler reports 1920x1080@85hz or 100hz (random sometimes 1 or the other)
1920x1080@100hz: VT 1125 ->1920x1080@100hz
1920x1080@100hz, VT 1133 ->monitor scaler reports 1920x1080@100hz.
1920x1080@100hz: VT 1138 ->3d mode if unlocked, otherwise 1920x1080@100hz
1920x1080@100hz: VT 1144 ->monitor scaler reports 1920x1080@100hz
1920x1080@100hz: VT 1149 ->monitor scaler reports 1920x1080@100
1920x1080@100hz: VT 1155 -> 1920x1080@100hz
1920x1080@100hz: VT 1117 -> 1920x1080@60hz <--this VT cannot be identified--reverts to 60hz pulse widths.
1920x1080@100hz: VT 1098 ->1920x1080@144hz
1920x1080@100hz: VT 1354 ->1600x1280@60
1920x1080@100hz: VT 1400 -> "Out of Range" ->screen appears ->scalar: 1792x1344@60hz
1920x1080@100hz: VT 1415-1480: Black screen OOR
1920x1080@100hz: VT 1481-1485: Out of range ->1920x1440@60 (Screen)
VT 1486-1496: black screen.
VT 1497-VT 1502 -> 1280x1440@60 (Lightboost range?)
Now for 120hz things get interesting.
1920x1080@120hz: VT 1133; --> scaler 1400x1050@120 ??
1920x1080@120hz: VT 1125; --> scaler 1400x1050@120hz ??
1920x1080@120hz: VT 1144; --> scaler 1920x1080@120hz
1920x1080@120hz: VT 1124 --> "Black screen out of range" or 1400x1050@120hz (random).
The higher VT's act the same, reporting 60hz and strange scalar resolutions: vertical size increases, horizontal size is bizarre.
This is all I can give right now.
Yes. The most common hack is to use EDID overrides to overclock the monitor refresh.
I haven't tried anymore configurations, but when I had it working my DSR resolutions were 16:10. When I tried it again with a lot of 4:3 resolutions, it wasn't working. I'm still gonna mess with it some more though.
Ok i finally got it working. But im still slightly confused: why would the monitor go blank on certain games only?
So far ive set my freesync range to 39-73 and this works wonderfully in Witcher 3 and Arkham Knight but for some reason GTA V just blanks out from time to time at around the same fps.
I've made sure none of these games ever go past 60 and my settings have been dialed way down to ensure fps high enough for freesync to kick in.
I blank out wherever I set the range to be. If its 60-75 hz, i'll blank out there. if its 34-60, i'll blank out in menus when the game only hits 60..
there must be something i'm missing here.
Proof of FreeSync working on the Qnix QX2710 single DVI input (old PLS panel):
Also tested playing Rise of the Tomb Raider on the Fury at max settings no AA (for ~40-66FPS) and 2xSSAA (for 27-50FPS) and FreeSync works wonderfully (it shows the framerates for what they are with no tearing whatsoever and no VSync stutter).
Is that the actual native resolution/refresh of your monitor before doing anything with CRU?
It's a 2560x1440 60Hz monitor that overclocks to 117Hz. I leave it at 110Hz for games. That's over dual-link DVI.
wow that is impressive overclocking there.. ok. i was wondering if it was a 1080 monitor by chance... So.. Can you use 110hz and use freesync still?
Have you pushed further than 66hz?
No. A few posts back, ToastyX explained that dual-link DVI operation is not supported over (what looks like, to the driver) HDMI by the driver. My findings match his experience. I cannot enable dual-link DVI operation after an HDMI block is added. The connection becomes a single-link DVI connection. That is despite checking the box enabling dual-link DVI operation in the HDMI block. It seems to strictly be a driver limitation.
That would mean that no, I cannot use 2560x1440 110Hz with FreeSync (yet).
I didn't try pushing much further than 66Hz because the monitor was flashing green lines from time to time, indicating that we are already pushing the limits of the GPU's TMDS transmitter / DVI-D cable / monitor bypass PCB over that single-link.
Do note that I had only 2h30m to test all of this. Install the Radeon R9 Fury, connect monitor, patch driver, add HDMI block, add FreeSync range, unplug and plug in, enable FreeSync, test Windmill demo, play Rise of the Tomb Raider, record footage of the Windmill demo running FreeSync. Then my friend left with his card as I had to go somewhere else. I have cards supporting FreeSync (eventually) on their way over to me, so I'll be testing much, much further.
On my 7970 CF, I generally run 110Hz at a 422MHz pixel clock over dual-link which would indicate a 211MHz pixel clock over single link. 66Hz over single-link on the Fury meant a ~250MHz pixel clock, which is quite something.
If AMD add FreeSync support over DVI, this whole conundrum is solved immediately, I believe.