So I got my shiny new ultra wide 3440x1440 monitor, set it up, and of course I had to fire up ED for a spin.
Test 1: gaming box, GTX960 2GB, connected by DP. Monitor reports 60Hz capable but the card itself is a bit weak at this resolution. On ultra settings averages around 50fps (vsync off). I also noticed while running GPU-Z reported about 2GB of vram was being used. Might that impact performance? Let's try my other box then...
Test 2: old main box, R9 280X 3GB, connected by HDMI. Monitor only offers 50Hz here. The card is made by XFX and a quick look on their webpage says it is "HDMI 1.4a ready". As far as I understand, that supports 4k at 30Hz only. Scale the pixel pushing numbers, and you get 50Hz at my monitor's resolution. There's the problem. Shame, with vsync off it was averaging 60fps, but it felt unsmooth. I believe my panel is 60 Hz native, so if it is getting info at 50Hz, I guess it has to use irregular updates to get them to match. So the 960 felt smoother than the 280X. Oh the 280X peaked at 2.2GB vram usage during the short time I was running ED.
Why not swap connections? The 280X only has mini-DP, and the cable that came with the monitor has a full size connector which the 960 has. So I can't unless I get a new cable.
I think I'll run at 30Hz as a short term workaround as that should then be a multiple of native refresh and look smoother, even with fewer updates. I have to speculate if mismatched refresh rates between panels and GPUs may be a contributing factor in reported cases of micro-stutter.
Unrelated problem: I have a DVI KVM so I can connect the above two boxes to the same monitor. I'm keeping my old main monitor as a secondary display as it has a wider colour gamut, so the KVM still switches that between them. While I was doing all these updates, I thought I'd plug in a backlit gaming keyboard I recently bought. I've used it by itself already, but it just wouldn't work through the KVM. I'm guessing the backlight might be taking too much power from the KVM even when set to minimum. I didn't try off, which would rather defeat the point of buying a backlit keyboard in the first place.
Test 1: gaming box, GTX960 2GB, connected by DP. Monitor reports 60Hz capable but the card itself is a bit weak at this resolution. On ultra settings averages around 50fps (vsync off). I also noticed while running GPU-Z reported about 2GB of vram was being used. Might that impact performance? Let's try my other box then...
Test 2: old main box, R9 280X 3GB, connected by HDMI. Monitor only offers 50Hz here. The card is made by XFX and a quick look on their webpage says it is "HDMI 1.4a ready". As far as I understand, that supports 4k at 30Hz only. Scale the pixel pushing numbers, and you get 50Hz at my monitor's resolution. There's the problem. Shame, with vsync off it was averaging 60fps, but it felt unsmooth. I believe my panel is 60 Hz native, so if it is getting info at 50Hz, I guess it has to use irregular updates to get them to match. So the 960 felt smoother than the 280X. Oh the 280X peaked at 2.2GB vram usage during the short time I was running ED.
Why not swap connections? The 280X only has mini-DP, and the cable that came with the monitor has a full size connector which the 960 has. So I can't unless I get a new cable.
I think I'll run at 30Hz as a short term workaround as that should then be a multiple of native refresh and look smoother, even with fewer updates. I have to speculate if mismatched refresh rates between panels and GPUs may be a contributing factor in reported cases of micro-stutter.
Unrelated problem: I have a DVI KVM so I can connect the above two boxes to the same monitor. I'm keeping my old main monitor as a secondary display as it has a wider colour gamut, so the KVM still switches that between them. While I was doing all these updates, I thought I'd plug in a backlit gaming keyboard I recently bought. I've used it by itself already, but it just wouldn't work through the KVM. I'm guessing the backlight might be taking too much power from the KVM even when set to minimum. I didn't try off, which would rather defeat the point of buying a backlit keyboard in the first place.