Maintenance for the week of December 22:
• NA megaservers for maintenance – December 22, 4:00AM EST (9:00 UTC) - 8:00AM EST (13:00 UTC)
• EU megaservers for maintenance – December 22, 4:00AM EST (9:00 UTC) - 8:00AM EST (13:00 UTC)
• NA megaservers for maintenance – December 22, 4:00AM EST (9:00 UTC) - 8:00AM EST (13:00 UTC)
• EU megaservers for maintenance – December 22, 4:00AM EST (9:00 UTC) - 8:00AM EST (13:00 UTC)
Multi-thread improvement actually works
ZeroXFF
✭✭✭✭✭
✭
✭
I did a quick test that basically consisted of me looking at the same scene on both PTS and live with the same in-game settings. Haven't made any videos, but it's in Elden Root near the wayshrine, standing on the stairs on the way to the crafting stations, but turned towards the wayshrine (in both cases the place is almost empty).
My specs
CPU: AMD R7 1700@3.775 GHz on all cores
RAM: 16GB DDR4-3200MHz CL16
GPU: AMD HD 7850 2GB (I know, time to upgrade, blame the crypto-miners)
Settings used
Here are my observations:
Live
FPS: ~42
CPU load: ~13% (with 1 core maxed out)
GPU load: ~60%
PTS
FPS: ~53
CPU load: ~25% (1 core still maxed out)
GPU load: ~75%
I used Process Explorer to see the load of individual threads. I also observed that while on live only 7 threads ever reach more than 1% load (which is about 15-20% of a single logical core on an 8 core CPU with SMT), on PTS there were at least 14 such threads (didn't count them, but all the ones visible in the window were over 1%).
Conclusion: noticeable improvement in hardware utilization, resulting in an FPS increase of about 25%, but I'm still CPU bottlenecked with a slightly overclocked high-end modern CPU and a potato graphics card at stock configuration, so there certainly still is a lot of room for improvement.
My specs
CPU: AMD R7 1700@3.775 GHz on all cores
RAM: 16GB DDR4-3200MHz CL16
GPU: AMD HD 7850 2GB (I know, time to upgrade, blame the crypto-miners)
Settings used
Here are my observations:
Live
FPS: ~42
CPU load: ~13% (with 1 core maxed out)
GPU load: ~60%
PTS
FPS: ~53
CPU load: ~25% (1 core still maxed out)
GPU load: ~75%
I used Process Explorer to see the load of individual threads. I also observed that while on live only 7 threads ever reach more than 1% load (which is about 15-20% of a single logical core on an 8 core CPU with SMT), on PTS there were at least 14 such threads (didn't count them, but all the ones visible in the window were over 1%).
Conclusion: noticeable improvement in hardware utilization, resulting in an FPS increase of about 25%, but I'm still CPU bottlenecked with a slightly overclocked high-end modern CPU and a potato graphics card at stock configuration, so there certainly still is a lot of room for improvement.
2
-
Kadoin✭✭✭✭✭
✭✭I did a quick test that basically consisted of me looking at the same scene on both PTS and live with the same in-game settings. Haven't made any videos, but it's in Elden Root near the wayshrine, standing on the stairs on the way to the crafting stations, but turned towards the wayshrine (in both cases the place is almost empty).
My specs
CPU: AMD R7 1700@3.775 GHz on all cores
RAM: 16GB DDR4-3200MHz CL16
GPU: AMD HD 7850 2GB (I know, time to upgrade, blame the crypto-miners)
Settings used
Here are my observations:
Live
FPS: ~42
CPU load: ~13% (with 1 core maxed out)
GPU load: ~60%
PTS
FPS: ~53
CPU load: ~25% (1 core still maxed out)
GPU load: ~75%
I used Process Explorer to see the load of individual threads. I also observed that while on live only 7 threads ever reach more than 1% load (which is about 15-20% of a single logical core on an 8 core CPU with SMT), on PTS there were at least 14 such threads (didn't count them, but all the ones visible in the window were over 1%).
Conclusion: noticeable improvement in hardware utilization, resulting in an FPS increase of about 25%, but I'm still CPU bottlenecked with a slightly overclocked high-end modern CPU and a potato graphics card at stock configuration, so there certainly still is a lot of room for improvement.
What happens when you put shadows on high though ? I'm pretty sure that shadows are the TRUE bottleneck in the game. Can you try it with shadows off too and report the difference? I'm just curious since I got 100 KB/s download speed
0 -
Ashtaris✭✭✭✭✭
✭✭Actually what I witnessed was a FPS decrease when playing on the PTS, going from about 80 fps on live down to 30 fps. So I need to do some checking to see what may have changed. I’ll go ahead and delete the PTS UserSettings.txt file and have the game recreate it to see if that will make any difference first, and then go from there.0 -
Try changing to full screen instead of windowed. I'm on Windows 10, and that oftentimes drastically increases performance on all my games (thanks for that Microsoft, I so love having to alt tab all the time to use my 2nd monitor). Right now on live I'll get 30fps in fullscreen-windowed, and 85fps when switched to fullscreen.0
-
Sweetare✭✭✭The Performance gain IS REAL. Holy smoly.
I think its the first time my PC has some work to do while playing ESO, could hear my fans hitting 100% a few times where a lot of people where around and CPU was mostly at 40%, spiking up to 75%. Amazing
Running an i7 6700 + GTX970 here0 -
ZeroXFF✭✭✭✭✭
✭Jimbostein wrote: »Try changing to full screen instead of windowed. I'm on Windows 10, and that oftentimes drastically increases performance on all my games (thanks for that Microsoft, I so love having to alt tab all the time to use my 2nd monitor). Right now on live I'll get 30fps in fullscreen-windowed, and 85fps when switched to fullscreen.
There is no way this would cause a 200% fps difference if there are no settings changed anywhere else. This has probably something to do with some settings/bugs in your graphics drivers setting your graphics card into some form of power saving mode if you aren't running anything in full screen. I compared full screen and windowed mode before, and the difference, if there was any, couldn't be measured without extensive benchmarking (meaning it was under 10%). This could affect some games more than others, but in ESO the difference is negligible if you don't have other problems.0 -
Smaxx✭✭✭✭✭Not really sure about the changes. I still see hardly more than one core used by the game on an i7 4770K outside loading screens. GPU load is at ~60%, so that shouldn't be the bottleneck either.0 -
Reorx_Holybeard✭✭✭✭✭I didn't see any change in CPU usage or FPS gain in a quick test between Live and PTS. Will do another one today and look more closely.Reorx Holybeard -- NA/PC
Founder/Admin of www.uesp.net -- UESP ESO Guilds
Creator of the "Best" ESO Build Editor
I'm on a quest to build the world's toughest USB drive!0 -
Without using an overlay like rivatuner plus having a higher end GPU you're not going to tell whether the multi core enhancements are giving any sort of advantage.
I'm currently downloading the PTS client at a snail's pace, if someone can beat me to it and get a video runthrough (ideally with a 1070 or above & a cpu with 12+ threads) it would be able to show what's actually happening.0 -
ZeroXFF✭✭✭✭✭
✭infraction2008b16_ESO wrote: »Without using an overlay like rivatuner plus having a higher end GPU you're not going to tell whether the multi core enhancements are giving any sort of advantage.
I'm currently downloading the PTS client at a snail's pace, if someone can beat me to it and get a video runthrough (ideally with a 1070 or above & a cpu with 12+ threads) it would be able to show what's actually happening.
You are describing the reason I tried it at relatively low settings and wrote down the GPU and CPU utilization. A better GPU with the given settings and my CPU wouldn't do diddly-squat. On an 8700k@5.3GHz it would probably cause a GPU bottleneck, but with my setup it's a clear CPU bottleneck that appears below 60 fps, which is exactly the problem with the game. And my CPU has 16 threads, so this part certainly applies.
But a proper test with more than a single scene would certainly be better to make an adequate picture of it. It's just that the client finished downloading at 1am and all I had time for was what I posted.0
