Another case of: #1318
Unfortunately there is nothing we can do.
Side note: Please pay attention to formatting when you make issues on GitHub and put the config on something like Pastebin instead of just dumping it here. It's not readable and the logs tell us which config option you've set anyway.
You could try to patch the Skyrim EXE and set the large address aware flag:
https://www.techpowerup.com/forums/threads/large-address-aware.112556/
No idea if this works with Skyrim but it's something you could try.
I just saw your other issue, maybe keep ReduceSystemMemoryUsage enabled after all.
You could try to patch the Skyrim EXE and set the large address aware flag:
https://www.techpowerup.com/forums/threads/large-address-aware.112556/No idea if this works with Skyrim but it's something you could try.
I just saw your other issue, maybe keep
ReduceSystemMemoryUsageenabled after all.
Well the main problem is when ReduceSystemMemoryUsage set to true, my skyrim have terrible stutters. With your patch it has no stutter, but game run out of memory. Also i found that if i delete my textures folder, i can load my save, so as i understand dxvk patch have problem with vram memory allocation,
Did you apply the LAA patch?
If that didn't help, there's nothing we can do. Just play without DXVK then.
Did you apply the LAA patch?
If that didn't help, there's nothing we can do. Just play without DXVK then.
Yes it didnt help. But when i use dxvk mem patch in stalker COP it works good. Why not on Skyrim LE?
Because different games do different things and manage memory differently. Apparently this is mostly caused by the ENB mod.
DXVK_HUDx1 2021-01
When i try to use DXVK memory patch it says DxvkMemoryAllocator: Memory allocation failed.
TESV_d3d9.log
Here my dxvk.conf settings:
Create the VkSurface on the first call to IDXGISwapChain::Present,
rather than when creating the swap chain. Some games that start
rendering with a different graphics API may require this option,
or otherwise the window may stay black.
Supported values: True, False
dxgi.deferSurfaceCreation = False
d3d9.deferSurfaceCreation = False
Enforce a stricter maximum frame latency. Overrides the application
setting specified by calling IDXGIDevice::SetMaximumFrameLatency.
Setting this to 0 will have no effect.
Supported values : 0 - 16
dxgi.maxFrameLatency = 0
d3d9.maxFrameLatency = 0
Override PCI vendor and device IDs reported to the application. Can
cause the app to adjust behaviour depending on the selected values.
Supported values: Any four-digit hex number.
dxgi.customDeviceId = 0000
dxgi.customVendorId = 0000
d3d9.customDeviceId = 0000
d3d9.customVendorId = 1002
Report Nvidia GPUs as AMD GPUs by default. This is enabled by default
to work around issues with NVAPI, but may cause issues in some games.
Supported values: True, False
dxgi.nvapiHack = False
Override maximum amount of device memory and shared system memory
reported to the application. This may fix texture streaming issues
in games that do not support cards with large amounts of VRAM.
Supported values: Any number in Megabytes.
dxgi.maxDeviceMemory = 16318
dxgi.maxSharedMemory = 16318
Overrides synchronization interval (Vsync) for presentation.
Setting this to 0 disables vertical synchronization entirely.
A positive value 'n' will enable Vsync and repeat the same
image n times, and a negative value will have no effect.
Supported values: Any non-negative number
dxgi.syncInterval = -1
d3d9.presentInterval = -1
True enables the mailbox present mode in case regular Vsync is disabled.
This should avoid tearing, but may be unsupported on some systems
or require setting dxgi.numBackBuffers to a higher value in order
to work properly.
False enables the relaxed fifo present mode in case regular Vsync is enabled.
This should result in tearing but reduce stutter if FPS are too low,
but may be unsupported on some systems.
Please do not report issues with this option.
Supported values: Auto, True, False
dxgi.tearFree = True
d3d9.tearFree = True
Declares vertex positions as invariant in order to solve
potential Z-fighting issues at a small performance cost.
Supported values: True, False
d3d11.invariantPosition = False
d3d9.invariantPosition = true
Sets number of pipeline compiler threads.
Supported values:
- 0 to automatically determine the number of threads to use
- any positive number to enforce the thread count
dxvk.numCompilerThreads = 16
Toggles raw SSBO usage.
Uses storage buffers to implement raw and structured buffer
views. Enabled by default on hardware which has a storage
buffer offset alignment requirement of 4 Bytes (e.g. AMD).
Enabling this may improve performance, but is not safe on
hardware with higher alignment requirements.
Supported values:
- Auto: Don't change the default
- True, False: Always enable / disable
dxvk.useRawSsbo = True
Toggles early discard.
Uses subgroup operations to determine whether it is safe to
discard fragments before the end of a fragment shader. This
is enabled by default on all drivers except RADV and Nvidia.
Enabling this may improve or degrade performance depending
on the game and hardware, or cause other issues.
Supported values:
- Auto: Don't change the default
- True, False: Always enable / disable
dxvk.useEarlyDiscard = True
Sets enabled HUD elements
Behaves like the DXVK_HUD environment variable if the
environment variable is not set, otherwise it will be
ignored. The syntax is identical.
dxvk.hud = 1
Reported shader model
The shader model to state that we support in the device
capabilities that the applicatation queries.
Supported values:
- 1: Shader Model 1
- 2: Shader Model 2
- 3: Shader Model 3
d3d9.shaderModel = 3
Evict Managed on Unlock
Decides whether we should evict managed resources from
system memory when they are unlocked entirely.
Supported values:
- True, False: Always enable / disable
d3d9.evictManagedOnUnlock = True
DPI Awareness
Decides whether we should call SetProcessDPIAware on device
creation. Helps avoid upscaling blur in modern Windows on
Hi-DPI screens/devices.
Supported values:
- True, False: Always enable / disable
d3d9.dpiAware = True
Strict Constant Copies
Decides whether we should always copy defined constants to
the UBO when relative addressing is used, or only when the
relative addressing starts a defined constant.
Supported values:
- True, False: Always enable / disable
d3d9.strictConstantCopies = False
Strict Pow
Decides whether we have an opSelect for handling pow(0,0) = 0
otherwise it becomes undefined.
Supported values:
- True, False: Always enable / disable
d3d9.strictPow = True
Lenient Clear
Decides whether or not we fastpath clear anyway if we are close enough to
clearing a full render target.
Supported values:
- True, False: Always enable / disable
d3d9.lenientClear = False
Max available memory
Changes the max initial value used in tracking and GetAvailableTextureMem
Value in Megabytes
Supported values:
- Any int32_t
d3d9.maxAvailableMemory = 32768
My video: RX 6900xt (latest driver), total ram 32 gb 5800x Ryzen.
Pls some help me to fix this problem. TY!