EQ1, 2k primary monitor, 4 instances, and lag issues
G'day, all.
Just to detail hardware here:
GA-990FX-GA
FX-8350
16GB DDR3 1600Mhz RAM
1TB SSD (game files are running off this)
Evga GTX970 4GB (primary)
AUS GTX960 4GB (secondary)
150MB down/15 up network connection
I had up until recently been using a 3 monitor (3x1080p, 27") setup which worked fine - all 4 instances of EQ were set to run off my primary (center) monitor, and I never had any problems with it from an EQ standpoint - everything ran fine, I had set each instance to be bound to 2 CPU's (ie. instance 1 had CPU1/2, instance 2 had CPU 3/4, etc), with the GTX970 only serving the single central monitor via HDMI, and the two side monitors driven by a GTX960. My window layouts were all full screen windowed on the primary monitor - again, with no issues, with the background FPS set to 30 with the primary (foreground) set to 60 FPS. I am using VideoFX to show on the primary instance items from the other three (spell bars, the single hotbar menu, character health, etc). The (usual) background instances have been set to use stickfigures as models, with most particle/lighting/etc. turned off or set at the bare minimums, although I do have their memory usage set to Balanced instead of Minimum, and the primary instance having a better visuals (particles, lighting, higher clip plane, etc) turned on.
I've since switched the center monitor to a 27" 2k monitor (2560x1440) - as it only runs Dual-DVI, and the other two 27" are HDMI, I had to share one of the side monitors on the GTX970 as well, with the other working off the GTX960. Now, while running 4 instances at 2560x1440 on the primary monitor, plus the one 1080p monitor off the GTX970, I'm getting horrible response from EQ instances 2-4, with widely varying amounts of lag between them (usually one to two instances are fine, running between 130-180ms, but the other two range anywhere from 600-1570ish ms, which makes it unplayable).
I did check my video card/cpu load - with all 4 instances running, my FX-8350 hits 50% used (averaged across the CPU, but at most single cores reach 65-70% used at any period of time). and the GTX970 uses 70% VRAM usage (and it's load usually only gets to at most 80% used), so it didn't seem like I was GPU/CPU pinned at all. I wouldn't expect the network to have issues (although who knows, it's Windows 10) as I wouldn't expect changing the display side of things hasn't increased the amount/# of packets being sent to SOE/EQ, which is why I didn't think it was a network problem.
I was going to pick up a DVI to HDMI adapter to make sure both side monitors are running off the GTX960 only today to see if it helps, but wanted to post if anyone else has suggestions for me.
(I'll upload a diagnostics of it in case that would help, but unfortunately I'm at work right now, bleh.)
Thanks!
Just to detail hardware here:
GA-990FX-GA
FX-8350
16GB DDR3 1600Mhz RAM
1TB SSD (game files are running off this)
Evga GTX970 4GB (primary)
AUS GTX960 4GB (secondary)
150MB down/15 up network connection
I had up until recently been using a 3 monitor (3x1080p, 27") setup which worked fine - all 4 instances of EQ were set to run off my primary (center) monitor, and I never had any problems with it from an EQ standpoint - everything ran fine, I had set each instance to be bound to 2 CPU's (ie. instance 1 had CPU1/2, instance 2 had CPU 3/4, etc), with the GTX970 only serving the single central monitor via HDMI, and the two side monitors driven by a GTX960. My window layouts were all full screen windowed on the primary monitor - again, with no issues, with the background FPS set to 30 with the primary (foreground) set to 60 FPS. I am using VideoFX to show on the primary instance items from the other three (spell bars, the single hotbar menu, character health, etc). The (usual) background instances have been set to use stickfigures as models, with most particle/lighting/etc. turned off or set at the bare minimums, although I do have their memory usage set to Balanced instead of Minimum, and the primary instance having a better visuals (particles, lighting, higher clip plane, etc) turned on.
I've since switched the center monitor to a 27" 2k monitor (2560x1440) - as it only runs Dual-DVI, and the other two 27" are HDMI, I had to share one of the side monitors on the GTX970 as well, with the other working off the GTX960. Now, while running 4 instances at 2560x1440 on the primary monitor, plus the one 1080p monitor off the GTX970, I'm getting horrible response from EQ instances 2-4, with widely varying amounts of lag between them (usually one to two instances are fine, running between 130-180ms, but the other two range anywhere from 600-1570ish ms, which makes it unplayable).
I did check my video card/cpu load - with all 4 instances running, my FX-8350 hits 50% used (averaged across the CPU, but at most single cores reach 65-70% used at any period of time). and the GTX970 uses 70% VRAM usage (and it's load usually only gets to at most 80% used), so it didn't seem like I was GPU/CPU pinned at all. I wouldn't expect the network to have issues (although who knows, it's Windows 10) as I wouldn't expect changing the display side of things hasn't increased the amount/# of packets being sent to SOE/EQ, which is why I didn't think it was a network problem.
I was going to pick up a DVI to HDMI adapter to make sure both side monitors are running off the GTX960 only today to see if it helps, but wanted to post if anyone else has suggestions for me.
(I'll upload a diagnostics of it in case that would help, but unfortunately I'm at work right now, bleh.)
Thanks!