Page 2 of 2
Posted: 2013-12-24 15:30
by Hulabi
What about the drop kit geos, a friend told me he used to have much higher fps before those were implemented.
Granted it was a long time ago, but still worth a look imo.
Re: Performance Test III - pr_fpstest3
Posted: 2013-12-24 15:46
by Rudd
[R-COM]Hulabi wrote:What about the drop kit geos, a friend told me he used to have much higher fps before those were implemented.
Granted it was a long time ago, but still worth a look imo.
Certainly wouldn't want the old drop kits back, but something simplier than current might be good, but I would expect that would require reexporting every single kit? big big job.
Re: Performance Test III - pr_fpstest3
Posted: 2013-12-24 17:44
by Spush
The custom kit geos have one texture sheet, granted it's a large texture sheet, but with small uv islands would decrease the quality of the textures even more so no to that. The geos have decent lods last I checked, I'm pretty sure the last time I checked Chuc did the newer vietnam ones. But everything to decrease loads would have to decrease some where. And it's all a very big big job for those who can. Personally I see it as a much needed move to gain better performance. Decreasing textures (not size), better mesh optimization and lods.
Re: Performance Test III - pr_fpstest3
Posted: 2013-12-24 20:27
by DankE_SPB
Decreasing textures (not size), better mesh optimization and lods.
Quite often problem are not the textures themselves (many textures on one model), but the way the texture is assigned to the mesh - i.e. same texture has different mat. ID on one mesh thus increasing drawcalls.
Also, did you guys implement the Rhino's way of LODing weapons? (last lod - simple box with alpha transparency to make weapon's silhouette)
Re: Performance Test III - pr_fpstest3
Posted: 2013-12-24 22:06
by Spush
DankE_SPB wrote:Also, did you guys implement the Rhino's way of LODing weapons? (last lod - simple box with alpha transparency to make weapon's silhouette)
I believe there are some especially for the newer weapons, but don't really remember to be honest. I know that Tim was at one point working on the AR-15 series.
Re: Performance Test III - pr_fpstest3
Posted: 2013-12-24 22:13
by Adriaan
DankE_SPB wrote:Quite often problem are not the textures themselves (many textures on one model), but the way the texture is assigned to the mesh - i.e. same texture has different mat. ID on one mesh thus increasing drawcalls.
Also, did you guys implement the Rhino's way of LODing weapons? (last lod - simple box with alpha transparency to make weapon's silhouette)
A few do, but most don't have the final alpha'd plane. Once the last lod is down to 100 or so tris the difference is pretty much neglegible, especially compared to the extra effort it takes.
Re: Performance Test III - pr_fpstest3
Posted: 2013-12-25 06:18
by ShockUnitBlack
These are all GPU concerns though. Pretty much all the tests are pointing towards CPU usage being the problem, not GPU usage.
Re: Performance Test III - pr_fpstest3
Posted: 2013-12-25 07:53
by Arab
6 Localization could not be found. Please reinstall Project Reality.
Off to re-download and re-install then will test.
Update: Fixed it. Reinstall was useless. Problem was the same folder had another 'pr_fpstest3' so simply cutting that folder and pasting it to /mods worked!
Dalian Plant
-As USMC, couldn't get the pilot kit from helicopter or jet spawned as riflemen
-As Chinese, could get a pilot kit from cow spawned as riflemen, but not for other kits.
Strike at Karkand
Loading Time 2:50 (Fresh Start/Optimizations)
Tested at Spawn Point.
Looking South: 91.8 - 99 FPS
Looking East: 92 - 99 FPS. Sometimes goes lower due to smoke
Looking North facing smoke: Big Spike Drop. Stays 93 then goes to 80, then 70. Lowest is 60 FPS
Looking West: 99.9 to 98. Spikes to 80 and 70 but stays stable.
So in order to least affected to most affected:
South, East, West, North.
Zatar Waterland:
Near Perfect FPS of 98-99 all directions!
East of MEC: A perfect 102.4 FPS Stable
Worst FPS: As MEC, place a c4 on the car, wait 19 seconds then drive into jet and explode. It took me from 97 FPS to 40 and lower FPS when the explosions and sound were happening.
-Bugs-
Karkand
-Getting into the M249 MG Emplacement and BGM-71 TOW will make the game think you are exiting the boundaries and gives you 10 seconds.
-M249 MG doesn't have it's name on the bottom-left hud. Not a gameplay bug, but a visual hud one. Not really sure if the vehicle bugs are only in the fpstest.
Dalian Plant
-Chinese Machine Gun Kit in Spawn Selection at last clip doesn't fire in deploy mode, but fires the remaining rounds in undeployed mode.
Zatar Waterlands
USMC can use the helicopter.
Again, I'm not sure if the vehicle name and entering the helicopters are vehicle bugs. Should we also report those as well?
Unrelated:
The FN FAL has a UV Map error from what I've read by a user who wanted to create his own model for the Brazilian Forces.
Re: Performance Test III - pr_fpstest3
Posted: 2013-12-25 11:40
by Prevtzer
[R-DEV]AfterDune wrote:Hmmm.... what happens when you set sounds to low? Less sounds at the same time or just less quality?
If the latter, I wonder what would happen if, for testing purposes, we would downgrade all sounds to something of lower quality and filesize.
I use 'Creative X-Fi' and 'Ultra High' with EAX 5.0 so all of the processing is done by the sound card. I still have terrible performance, so the sound isn't it.
Re: Performance Test III - pr_fpstest3
Posted: 2013-12-25 12:34
by Arab
Sound Quality doesn't affect FPS at all. I find that EAX does improve performance a little with me more than with it off though.
Re: Performance Test III - pr_fpstest3
Posted: 2013-12-25 14:52
by Spush
The only thing I see sounds being a problem is loading a heavy files, which anders has optimized before 1.0 was released to more compressed files. I'm sure there can be more done in this field but don't think it's a large issue atm to be honest. If anything it's the more gpu extensive content that will impact performance more than anything.