Anyone having trouble getting it to use more than 2 threads of your CPU? Guy over at OCN demonstrated that adding a line of code in the config file allowed the game to use all 8 of his threads on his CPU.
Line of code is:
Where the number (8) is the amount of threads your CPU has.
Heres result of adding that bit.
Yeah this is actually old news and worked the same in Rage..
I had Q9450 back then and with jobthread=8 i got higher cuda transcode benchmark even though I was using gpu transcode.
Few extra commands to put in Wolfconfig.cfg
@username/Saved Games/MachineGames/Wolfensetin/Bin
But only with ingame VT_compression - ON,
It uses DXT anyway, very effective no compression Image Quality anomalies., only better detail because of higher texture map size and less streaming popups when you're turning around. Theoretically, it was like so in Rage.
And in Rage I managed to "fix it" completely this way, btw this 16k tweak needs min 1.5gb vram, all 4 at 8k was ok with 1.2gb.
Code:
.
.
.
bind "MOUSE3" "_melee"
bind "MWHEELDOWN" "_weapprevinv"
bind "MWHEELUP" "_weapnextinv"
//========================================
pm_togglesprint "0"
r_mode "16"
r_multiSamples "4"
m_sensitivity "3.360597"
//main settings
com_skipIntroVideo "1"
com_allowconsole "1"
seta com_videoRam "00000000000" //insert your vram size here
seta jobs_numThreads "8"
//Virtual Textures
vt_pageimagesizeuniquediffuseonly2 "16384" // or 8192
vt_pageimagesizeuniquediffuseonly "16384" //or 8192
vt_pageimagesizeunique "16384" // or 8192 or 4096 default ingame 512
vt_pageimagesizevmtr "8192" // same as default ingame
vt_maxaniso "16"
vt_maxlockedpages "2048"
//virtual textures format
vt_qualityHDPLossless "1"
vt_qualityHDPDiffuse "0"
vt_qualityHDPNormal "0"
vt_qualityHDPPower "0"
vt_qualityHDPSpecular "0"
vt_qualityDCTChroma "100"
vt_qualityDCTLuma "100"
vt_qualityDCTNormal "100"
vt_qualityDCTPower "100"
vt_qualityDCTSpecular "100"
seta vt_preload "1"
//cuda settings // for nvidia only, otherwise delete this whole cuda settings part leave vt_maxPPF alone.
vt_cudaRLE "1"
vt_maxPPF "256"
vt_cudaBudget "12"
vt_usecudatranscode "2"
//Render
//vt_minlod "-1"
//vt_lodBias "-1"
//image_lodbias "-1"
r_useNewPrivateLighting "1"
r_useHardwareTextures "1"
r_useSMP "1"
r_noBreakableShadows "0
//Image
seta image_preload "1"
image_blend "GL_LINEAR_MIPMAP_LINEAR"
image_filter "GL_LINEAR_MIPMAP_LINEAR"
//Noskip forceON
md6_generateDamageTextures "1"
r_lazyBindParms "0"
r_lazyBindPrograms "0"
r_lazyBindTextures "0"
//vt_restart
seta r_forceVmtrReload "1" // needs more testing, could cause longer loading between deaths.
set to Read-only after or it can overwrite/erase these extra lines when you die
Again works only with Compression - ON or it can run out of vram really quickly.
Unless you have 3gb+, 2gb isnt enough one 670gtx user confirmed it, he used no compression and with those 3 commands @ 16k and he got 7fps.
When I tested above with VT-compressed I was ok @ 580GTX, usually 55-60fps (min ~ 45fps).
Log comfired it was using 8 &16k
Wolfenstein The New Order\virtualtextures\_vmtr.pages
0.0 seconds for FinishFeedback()
PHYSICAL_PAGES_POOL_VMTR
------------------------
total : 1024
locked : 6
current: 6
free : 1018
100 % empty hash table entries
100 % pages 0 steps down the hash chain
0 % pages 1 step down the hash chain
0 % pages 2 steps down the hash chain
0 % pages > 2 steps down the hash chain
------------------------
0.0 seconds to load and lock 21 pages
S 4096 4096 1 linr DXT5 clmp 16384k _physicalvmtrpages1
S 4096 4096 1 linr DXT1 clmp 8192k _physicalvmtrpages0
S 4096 4096 1 linr DXT5 clmp 16384k _physicalvmtrpages2
S 128 128 1 linr DXT5 clmp 16k _physicaluniquepages1
S 128 128 1 linr DXT1 clmp 8k _physicaluniquepages0
S 128 128 1 linr DXT5 clmp 16k _physicaluniquepages2
S 128 128 1 linr DXT5 clmp 16k _physicaluniquediffuseonlypages1
S 128 128 1 linr DXT5 clmp 16k _physicaluniquediffuseonly2pages1
virtual texture 1:
S 2048 2048 12 nmn RG clmp 10922k _vmtrpagetable