• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

The Radeon RX 7900 XT(X) Owners Thread.

But it is better (7900xt) for RT which is at 3090 level.

Also don’t forget there might be chance FSR 3.0 will lock out 6950xt etc. They seem to be following everything Nvidia is doing so don’t be surprised.

Fair point. RT is a good generation ahead even though it's still a gen behind Nvidia. Shame it's doesn't work for cyberpunk.
 
Sack the ultra preset off immediately - absolutely not worth it to game with on a competitive FPS.

Try using the settings below in the settings file - just replace everythingin options3.cod22.cst with the below (open it with notepad) then see what you get.

0

// system settings version.
Version:0.0 = "1"

//
// Audio
//

// Audio mix
AudioMix:0.0 = "2" // 0 to 9

// Adjusts the volume of the sound effects
EffectsVolume:0.0 = "1.000000" // 0.000000 to 1.000000

// Adjusts the volume of hit markers
HitMarkersVolume:0.0 = "1.000000" // 0.000000 to 1.000000

// Licensed Music Volume
LicensedMusicVolume:0.0 = "1.000000" // 0.000000 to 1.000000

// Adjusts the volume of the music
MusicVolume:0.0 = "0.000000" // 0.000000 to 1.000000

// Sets whether audio should be muted when the game window is out of focus
MuteAudioWhileOutOfFocus:0.0 = "false"

// Adjusts the volume of audio in Telescope (MOTD)
TelescopeVolume:0.0 = "1.000000" // 0.000000 to 1.000000

// Using push to talk to activate microphone.
VoicePushToTalk:0.0 = "true"

// Adjusts the volume for character dialogues and announcer voices
VoiceVolume:0.0 = "1.000000" // 0.000000 to 1.000000

// Adjusts overall volume of the game
Volume:0.0 = "1.000000" // 0.000000 to 1.000000

// War Tracks Volume
WarTracksVolume:0.0 = "1.000000" // 0.000000 to 1.000000

//
// Display
//

// Force specific aspect ratio independent of window aspect ratio
AspectRatio:0.0 = "wide 21:9" // one of [auto, standard, 5:4, wide 16:10, wide 16:9, wide 18:9, wide 19.5:9, wide 21:9, wide 32:9]

// Enables the async compute for specific cases
AsyncCompute:0.0 = "false"

// Constrain mouse to game window
ConstrainMouse:0.0 = "false"

// Color spaces for monitor output
DisplayGamma:0.0 = "BT709_sRGB" // one of [BT709_sRGB, BT709_BT1886]

// Fullscreen mode
DisplayMode:0.0 = "Fullscreen borderless window" // one of [Windowed, Fullscreen, Fullscreen borderless window, Fullscreen borderless extended window]

// Enable focused mode
FocusedMode:0.0 = "false"

// Set overlay opacity for the focused mode
FocusedModeOpacity:0.0 = "0.900000" // 0.000000 to 1.000000

// Monitor name of the monitor used to display the game
Monitor:0.0 = "Dell AW3423DW"

// Enable Nvidia low latency mode. Boost mode request maximum GPU clock frequency regardless of workload
NvidiaReflex:0.0 = "Enabled + boost" // one of [Disabled, Enabled, Enabled + boost]

// Fullscreen mode preference
PreferredDisplayMode:0.0 = "Fullscreen borderless window" // one of [Windowed, Fullscreen, Fullscreen borderless window, Fullscreen borderless extended window]

// Refresh rate of used monitor
RefreshRate:0.0 = "175.000"

// Fullscreen Resolution
Resolution:0.0 = "3440x1440"

// Percentage of window resolution that the 3D scene renders at. value set by the user. can be smaller than the actual.
ResolutionMultiplier:0.0 = "100" // 0 to 200

// Synchronizes framerate with refresh rate to prevent screen tearing issues
VSync:0.0 = "0" // 0 to 4

// Vsync while in menus only
VSyncInMenu:0.0 = "false"

// Window height
WindowHeight:0.0 = "400" // 0 to 32767

// Window is maximized on boot
WindowMaximized:0.0 = "true"

// Window width
WindowWidth:0.0 = "640" // 0 to 32767

// Window X position
WindowX:0.0 = "-32768" // -32768 to 32767

// Window Y position
WindowY:0.0 = "-32768" // -32768 to 32767

//
// Gameplay
//

// Apply custom framerate limit
CapFps:0.0 = "true"

// Adds blur on out of focus areas when aiming down sights
DepthOfField:0.0 = "false"

// Custom maximum frames per second when in a match
MaxFpsInGame:0.0 = "175" // 30 to 300

// Custom maximum frames per second when in menus
MaxFpsInMenu:1.0 = "60" // 30 to 300

// Custom maximum frames per second when game is out of focus
MaxFpsOutOfFocus:0.0 = "30" // 5 to 300

//
// Graphics
//

// User preferred anti-aliasing technique
AATechniquePreferred:0.0 = "Filmic SMAA T2x" // one of [SMAA T2x, Filmic SMAA T2x, XeSS]

// Strength for Contrast Adaptive Sharpening (CAS)
AMDContrastAdaptiveSharpeningStrength:0.0 = "0.750000" // 0.000000 to 1.000000

// Enable AMD FidelityFX Super Resolution
AMDSuperResolution:0.0 = "CAS - Sharpening only" // one of [Off, CAS - Sharpening only, AMD FSR 1.0]

// AMD FidelityFX Super Resolution quality
AMDSuperResolutionQuality:0.0 = "Maximum Quality" // one of [Maximum Performance, Balanced, Maximum Quality, Ultra Quality]

// Absolute target resolution
AbsoluteTargetResolution:0.0 = "none" // one of [540P, 640P, 720P, 900P, 1080P, 1440P, native, none]

// Show bullet impacts
BulletImpacts:0.0 = "true"

// The maximum distance at which clutter models are rendered.
ClutterMaxDist:0.0 = "2500.000000" // 100.000000 to 10000.000000

// Corpses culling threshold
CorpsesCullingThreshold:0.0 = "0.850000" // 0.500000 to 1.000000

// Default SMAA technique
DefaultSMAATechnique:0.0 = "Filmic SMAA T2x" // one of [SMAA T2x, Filmic SMAA T2x]

// Enable deferred physics
DeferredPhysics:0.0 = "Low Quality" // one of [Low Quality, Medium Quality, High Quality, Developer]

// Activate dynamic scene resolution
DynamicSceneResolution:0.0 = "false"

// Target frame time in ms for dynamic scene resolution.
DynamicSceneResolutionTarget:0.0 = "16.670000" // 0.000000 to 100.000000

// Strength of the filmic visual noise filter
FilmicStrength:0.0 = "1.000000" // 0.000000 to 1.000000

// Preferred GPU if multiple GPU system
GPUName:0.0 = "AMD Radeon RX 7900 XTX"

// GTAO Quality level
GTAOQuality:0.0 = "R_GTAO_QUALITY_LOW" // one of [R_GTAO_QUALITY_LOW, R_GTAO_QUALITY_MEDIUM, R_GTAO_QUALITY_HIGH, R_GTAO_QUALITY_ULTRA]

// Set HDR activation mode. Option only takes effect on HDR Display.
HDR:0.0 = "Automatic" // one of [Off, On, Automatic]

// LODs distance quality
ModelLodDistanceQuality:0.0 = "Base Quality" // one of [High Quality, Base Quality]

// Models quality
ModelLodQuality:0.0 = "Base quality" // one of [High quality, Base quality]

// Enable NVIDIA Image Scaling
NVIDIAImageScaling:0.0 = "false"

// NVIDIA Image Scaling quality
NVIDIAImageScalingQuality:0.0 = "Maximum Quality" // one of [Maximum Performance, Balanced, Maximum Quality, Ultra Quality, Custom, Native Quality]

// NVIDIA Image Scaling sharpness
NVIDIAImageScalingSharpness:0.0 = "0.300000" // 0.000000 to 1.000000

// Particle Lighting Quality
ParticleLighting:0.0 = "4" // 0 to 5

// Particle Quality quality level
ParticleQualityLevel:0.0 = "low" // one of [very low, low, medium, high]

// Particle Resolution quality level
ParticuleResolution:0.0 = "true"

// Enable persistent damage layer
PersistentDamageLayer:0.0 = "false"

// Particle Lighting quality level
PixelPerLightmapTexel:0.0 = "24.000000" // 1.000000 to 256.000000

// Enable half resolution reflection probes
ReflectionProbeHalfResolution:0.0 = "false"

// Reflection probe relighting update stages
ReflectionProbeRelighting:0.0 = "4" // 1 to 4

// SMAA Quality level
SMAAQuality:0.0 = "SMAA_QUALITY_MEDIUM" // one of [SMAA_QUALITY_LOW, SMAA_QUALITY_MEDIUM, SMAA_QUALITY_HIGH, SMAA_QUALITY_ULTRA]

// Screen-space ambient occlusion method
SSAOTechnique:0.0 = "Off" // one of [Off, GTAO, MDAO, GTAO & MDAO]

// Screen-space reflection mode
SSRMode:0.0 = "Deferred LQ" // one of [Off, Deferred LQ, Deferred HQ]

// Screen Space Shadow quality level
ScreenSpaceShadowQuality:0.0 = "Off" // one of [Off, Low, High]

// Select the shader quality setting
ShaderQuality:0.0 = "Low" // one of [Default, Medium, Low]

// Shadow resolution quality level
ShadowMapResolution:1.0 = "Normal" // one of [Very Low, Low, Normal, High, Extra, Ultra]

// Adjust size of Spot Shadow Cache
SpotShadowCacheSize:0.0 = "2" // 0 to 3

// Quality level of spot shadow
SpotShadowQualityLevel:0.0 = "Medium" // one of [Low, Medium, High]

// Static sunshadow moment clipmap resolution
StaticSunshadowClipmapResolution:0.0 = "512" // 0 to 2147483647

// Catmull Clark subdivision level
SubdivisionLevel:0.0 = "3" // 0 to 8

// Quality level of shadows from the sun at a distance
SunShadowCascade:0.0 = "High (2-3 cascades)" // one of [Low (1 cascade), Medium (1-2 cascades), High (2-3 cascades)]

// Tessellation quality level
Tessellation:0.0 = "0_Off" // one of [0_Off, 1_Near, 2_All]

// Texture filtering quality level
TextureFilter:0.0 = "TEXTURE_FILTER_ANISO16X" // one of [TEXTURE_FILTER_NEAREST, TEXTURE_FILTER_LINEAR, TEXTURE_FILTER_ANISO2X, TEXTURE_FILTER_ANISO4X, TEXTURE_FILTER_ANISO8X, TEXTURE_FILTER_ANISO16X, TEXTURE_FILTER_CMP]

// Texture quality level, high to low ( higher number means lower resolution )
TextureQuality:0.0 = "1" // 0 to 3

// Ui Quality
UiQuality:0.0 = "Auto" // one of [1080P, 4K, Auto]

// Sets which memory mode to use for physical textures. 0:Extra small (11x11, 121 slots), 1:Small (16x16, 256 slots), 2:Medium (23x23, 529 slots), 3:Large (32x32, 1024 slots), 4:Extra large (45x45, 2025 slots).
VirtualTexturingMemoryMode:0.1 = "Large" // one of [Extra Small, Small, Medium, Large, Extra Large]

// Volumetric quality
VolumetricQuality:0.0 = "QUALITY_LOW" // one of [QUALITY_LOW, QUALITY_MEDIUM, QUALITY_HIGH]

// Select water caustics mode
WaterCausticsMode:0.0 = "Off" // one of [Off, Low Quality, High Quality]

// Select weather grid volumes quality
WeatherGridVolumesQuality:0.0 = "Medium" // one of [Ultra, High, Medium, Low, Off]

// World streaming quality option
WorldStreamingQuality:0.0 = "Low" // one of [Low, High]

// XeSS quality
XeSSQuality:0.0 = "Balanced" // one of [Maximum Performance, Balanced, Maximum Quality, Ultra Quality, Custom]

//
// Interface
//

// Skip introduction movie that plays when game is started
SkipIntro:0.0 = "false"

//
// Mouse and Gamepad
//

// Set minimum delay in milliseconds between valid mouse wheel inputs
WeaponCycleDelay:0.0 = "0" // 0 to 5000

//
// System
//

// Allow campaign savegame to be synced in with the cloud
ConfigCloudSavegameEnabled:0.0 = "true"

// Allows non-hardware settings to be synced in with the cloud
ConfigCloudStorageEnabled:1.0 = "true"

// Indicates whether recommended settings have been set, will reset settings to recommended if set to 0
RecommendedSet:0.0 = "true"

// Thread count for handling the job queue
RendererWorkerCount:0.0 = "8" // -1 to 16

// Set a target fraction of your PC's video memory to be used by the game
VideoMemoryScale:0.0 = "0.800000" // 0.000000 to 2.000000

// Adjusts the strength of the visual noise applied in-game
FilmGrain:0.0 = "0.000000" // 0.000000 to 1.000000

// Use large physical textures
VirtualTexturingLargeMemory:0.0 = "true"

// Use spot shadow cache to improve framerate
CacheSpotShadow:0.0 = "Enabled" // one of [Disabled, Enabled, Preserve Updates]

// Use cache sun shadow
CacheSunShadow:0.0 = "true"

// Enable weather grid volumes
WeatherGridVolumes:0.0 = "false"
Just joined the forum to say thx mlord.
 
Someone explain to me what this is and how it works, quite the claim!


Clickbait for a 12 minute technical video on RT, i got to 6 minutes in of him rambling the technicalities of RT, he actually said 3 times at that point that he didn't want to go in to the technicalities of RT for this bite size video, and then said "anyway" and goes back to rambling mind numbing technicalities repeating the same thing in different ways over and over again.

Everything is about the 10 minute length Youtube algorithm so they can squeeze 16 15 second adverts in it (if you're watching it on a telly where no adblock exists) and make anything and everything unwatchable.

Now completely brain dead and frustrated i'm none the wiser as to the purpose of the video or the rumour.

 
Last edited:
Clickbait for a 12 minute technical video on RT, i got to 6 minutes in of him rambling the technicalities of RT, he actually said 3 times at that point that he didn't want to go in to the technicalities of RT for this bite size video, and then said "anyway" and goes back to rambling mind numbing technicalities repeating the same thing in different ways over and over again.

Everything is about the 10 minute length Youtube algorithm so they can squeeze 16 15 second adverts in it (if you're watching it on a telly where no adblock exists) and make anything and everything unwatchable.

Now completely brain dead and frustrated i'm none the wiser as to the purpose of the video or the rumour.


This is the conclusion i came to, i wanted someone to explain it to me cos it was just a complete mess of rambling jargon that made no sense to me lol.
 
From what i can gather AMD have been R&D'ing a new way to calculate RT and there is a huge uplift in performance, but its all for the future, an unknown future.

A 12 minute video to say that, he calls what is a full CH5 documentary (once you've taken all the adverts and recaps out) a "bite size" video, this is why i unsubscribed from him, he rambles and rambles on just to fill 9 minutes of a 10 minute+ video because the Youtube algorithm doesn't promote anything that's less than 10 minutes long.
 
Last edited:
Problem is that few people can afford cards priced over £1,000.

The cost per frame on the 4090 is poor. You don't get 2x the performance of the RX 7900 XT, even assuming this card costs £800, which it doesn't, more like £700 (far better than the RTX 3090 TI price). If you did, the price might seem less ridiculous.

It's safe to say that the RTX 4090 TI is pretty unlikely to offer 2x the performance of the RX 7900 XT also.

Average_4K_o-p.webp


You'd need a card that could offer 1% lows of ~160 to get double the performance, the 4090 doesn't come close.
 
Last edited:
Its literally the only game with RT I own (unless you count Hogwarts and thats broken for all users) and too many nvidia users are reporting issues with Hogwarts because their vram is is inadequate and thats more useful. The worship of the gods of RT is overdone.
 
Last edited:
A good rule of thumb that people sometimes forget, is to try to buy something that is equal or a bit better than the top card of the last generation (AMD or NVIDIA).

The RTX 4080 should've been the price competitive card to beat AMD's 7900 series cards, but presumably they just can't produce enough.

meanwhile, there's RTX 4070 TIs selling for £800 or more... This is clearly the card that Nvidia wants to sell to consumers / mass produce. Also the RTX 4070.

On the Nvidia side, lots of people seem to be thinking this whole generation is one to skip. There haven't been the necessary price reductions at the high end to make them sufficiently appealing.
 
Last edited:
A good rule of thumb that people sometimes forget, is to try to buy something that is equal or a bit better than the top card of the last generation (AMD or NVIDIA).

The RTX 4080 should've been the price competitive card to beat AMD's 7900 series cards, but presumably they just can't produce enough.

meanwhile, there's RTX 4070 TIs selling for £800 or more... This is clearly the card that Nvidia wants to sell to consumers / mass produce. Also the RTX 4070.

On the Nvidia side, lots of people seem to be thinking this whole generation is one to skip. There haven't been the necessary price reductions at the high end to make them sufficiently appealing.


When AMD and Nvidia initially placed their orders with TSMC for the chips there was a scarcity of supply from TSMC but that is no longer the case. Nvidia wanted to lower their production but TSMC did not let them out of the contract. There is supply for the silicon if it is needed and the ram is also low in price and available, if AMD or Nv they wanted to produce more cards then they easily could. The pandemic production constraints are not really an issue anymore.

Prices have dropped but not to the point where they are enticing the millions of gamers on 1060, 1650, 2060 to upgrade and yes many will skip this gen.
 
Clickbait for a 12 minute technical video on RT, i got to 6 minutes in of him rambling the technicalities of RT, he actually said 3 times at that point that he didn't want to go in to the technicalities of RT for this bite size video, and then said "anyway" and goes back to rambling mind numbing technicalities repeating the same thing in different ways over and over again.

Everything is about the 10 minute length Youtube algorithm so they can squeeze 16 15 second adverts in it (if you're watching it on a telly where no adblock exists) and make anything and everything unwatchable.

Now completely brain dead and frustrated i'm none the wiser as to the purpose of the video or the rumour.


This is the conclusion i came to, i wanted someone to explain it to me cos it was just a complete mess of rambling jargon that made no sense to me lol.

I'll try to get through the remaining 6 minutes to see if there is a point to it...

Simply put- BVH transversal is a way to help accelerate ray tracing by grouping objects together based on their location.

The RT system then uses this list to find objects in that area that will possibly come into contact with the approximate path of the ray (narrows down the search). If it's decided a ray will hit that boundary, it'll split that boundary and keep doing until it finds the primitive (poly) it actually intersects with.

Short: BVH transversal makes ray tracing faster. It basically makes an index for the lighting system so it can find where the rays hit faster than just raw dogging it.

Nvidia has hardware acceleration for this because BVH transversal is very computationally expensive, you need to generate this index based on whats currently on screen/where the camera is pointed. this changes often too making it harder.

AMD uses a different method using texture data instead, amds looking into implementing their own method of BVH transversal, and in some tests it increases performance by 1.5x.
 
Last edited:
Simply put- BVH transversal is a way to help accelerate ray tracing by grouping objects together based on their location.

The RT system then uses this list to find objects in that area that will possibly come into contact with the approximate path of the ray (narrows down the search). If it's decided a ray will hit that boundary, it'll split that boundary and keep doing until it finds the primitive (poly) it actually intersects with.

Short: BVH transversal makes ray tracing faster. It basically makes an index for the lighting system so it can find where the rays hit faster than just raw dogging it.

Nvidia has hardware acceleration for this because BVH transversal is very computationally expensive, you need to generate this index based on whats currently on screen/where the camera is pointed. this changes often too making it harder.

AMD uses a different method using texture data instead, amds looking into implementing their own method of BVH transversal, and in some tests it increases performance by 1.5x.

Glad you put that into more plain english for us folk! Be interesting to see how this plays out then, it will be quite the jump in RT performance if they manage to implement it properly.
 
Simply put- BVH transversal is a way to help accelerate ray tracing by grouping objects together based on their location.

The RT system then uses this list to find objects in that area that will possibly come into contact with the approximate path of the ray (narrows down the search). If it's decided a ray will hit that boundary, it'll split that boundary and keep doing until it finds the primitive (poly) it actually intersects with.

Short: BVH transversal makes ray tracing faster. It basically makes an index for the lighting system so it can find where the rays hit faster than just raw dogging it.

Nvidia has hardware acceleration for this because BVH transversal is very computationally expensive, you need to generate this index based on whats currently on screen/where the camera is pointed. this changes often too making it harder.

AMD uses a different method using texture data instead, amds looking into implementing their own method of BVH transversal, and in some tests it increases performance by 1.5x.

Thanks for the layman's version :)
 
Clickbait for a 12 minute technical video on RT, i got to 6 minutes in of him rambling the technicalities of RT, he actually said 3 times at that point that he didn't want to go in to the technicalities of RT for this bite size video, and then said "anyway" and goes back to rambling mind numbing technicalities repeating the same thing in different ways over and over again.

Everything is about the 10 minute length Youtube algorithm so they can squeeze 16 15 second adverts in it (if you're watching it on a telly where no adblock exists) and make anything and everything unwatchable.

Now completely brain dead and frustrated i'm none the wiser as to the purpose of the video or the rumour.

You did well as soon as I hear this bloke I want to fall asleep
 
When AMD and Nvidia initially placed their orders with TSMC for the chips there was a scarcity of supply from TSMC but that is no longer the case. Nvidia wanted to lower their production but TSMC did not let them out of the contract. There is supply for the silicon if it is needed and the ram is also low in price and available, if AMD or Nv they wanted to produce more cards then they easily could. The pandemic production constraints are not really an issue anymore.

Prices have dropped but not to the point where they are enticing the millions of gamers on 1060, 1650, 2060 to upgrade and yes many will skip this gen.
I think they'd probably rather commit themselves to product refreshes / improvements, if they are going to spend on more capacity. They would make more money that way.
 
Back
Top Bottom