DX11 Locking lowest mip levels of a DXT texture

This topic is 2601 days old which is more than the 365 day threshold we allow for new replies. Please post a new topic.

Recommended Posts

Hi, I'm writing a cross platform engine (GL, GL ES, DX9, DX10, DX11), I've got a Texture class with a Lock and Unlock method, in directX, depending on the pool & usage some textures can be locked directly and some cannot and must use a temporary "staging" texture ( or surface for locking a single mip level ), this works perfectly, except with DX9...

In (more or less) pseudo code this look like this (Considering only the particular case of a DXT texture in D3DPOOL_DEFAULT which cannot be locked directly):

Texture::Lock( u32 _mipLevel ){      u32 mipWidth, mipHeight;      GetMipMapDimensions( _mipLevel, mipWidth, mipHeight );      #ifdef TRY_FORCING_MINIMUM_STAGING_TEX_SIZE                 //DX9 complains about creating DXT textures less than 4x4 in size     	 mipWidth = Max( mipWidth, 4 );     	 mipHeight = Max( mipHeight, 4 );      #endif      pDevice->CreateTexture( mipWidth, mipHeight,                                            1,//mip maps 		           0,//usage		           format,	                           D3DPOOL_SYSTEMMEM,                                          &m_pStagingTexture,  NULL );      m_pStagingTexture->Lock( ... );}Texture::Unlock( u32 _mipLevel ){	if( m_pStagingTexture )	{		m_pStagingTexture->UnlockRect( 0 );		IDirect3DSurface9* pStagingSurface;		IDirect3DSurface9* pDestSurface;		m_pStagingTexture->GetSurfaceLevel( 0, &pStagingSurface );		((IDirect3DTexture9*)m_pTex)->GetSurfaceLevel( _mipLevel, &pDestSurface );		//Here I also tried to call UpdateSurface with various parameters for pSourceRect and pDestinationPoint instead of NULL, with no success :(		pDevice->UpdateSurface( pStagingSurface, NULL, pDestSurface, NULL ); 		SAFE_RELEASE( m_pStagingTexture )	}}

This works nicely for uncompressed formats, but this breaks when locking the lowest mip levels (1x1 and 2x2) because of the DXT 4x4 block size

if TRY_FORCING_MINIMUM_STAGING_TEX_SIZE is NOT defined I get "Direct3D9: (ERROR) :DXT Formats require width/height to be a multiple of 4. CreateTexture failed." while creating the staging texture

if TRY_FORCING_MINIMUM_STAGING_TEX_SIZE is defined I get "Direct3D9: (ERROR) :pRect doesn't fit inside the surface" while calling UpdateSurface

I think I could get away by creating a staging texture with the full mip chain instead of only the mip level I'm locking and then using UpdateTexture instead of UpdateSurface but this would be very sub-optimal as well as making my code kind-of ugly

Do you know a solution to this problem ? ( I could eventually use D3DPOOL_MANAGED textures and lock them directly, but I'd rather avoid that solution if possible )

Nobody ?

Share on other sites
What if you specify NULL for the rect and point in UpdateSurface?

• 9
• 10
• 11
• 9
• 16
• Similar Content

• I wanted to see how others are currently handling descriptor heap updates and management.
I've read a few articles and there tends to be three major strategies :
1 ) You split up descriptor heaps per shader stage ( i.e one for vertex shader , pixel , hull, etc)
2) You have one descriptor heap for an entire pipeline
3) You split up descriptor heaps for update each update frequency (i.e EResourceSet_PerInstance , EResourceSet_PerPass , EResourceSet_PerMaterial, etc)
The benefits of the first two approaches is that it makes it easier to port current code, and descriptor / resource descriptor management and updating tends to be easier to manage, but it seems to be not as efficient.
The benefits of the third approach seems to be that it's the most efficient because you only manage and update objects when they change.

• hi,
until now i use typical vertexshader approach for skinning with a Constantbuffer containing the transform matrix for the bones and an the vertexbuffer containing bone index and bone weight.
Now i have implemented realtime environment  probe cubemaping so i have to render my scene from many point of views and the time for skinning takes too long because it is recalculated for every side of the cubemap.
For Info i am working on Win7 an therefore use one Shadermodel 5.0 not 5.x that have more options, or is there a way to use 5.x in Win 7
My Graphic Card is Directx 12 compatible NVidia GTX 960
the member turanszkij has posted a good for me understandable compute shader. ( for Info: in his engine he uses an optimized version of it )
Now my questions
is it possible to feed the compute shader with my orignial vertexbuffer or do i have to copy it in several ByteAdressBuffers as implemented in the following code ?
the same question is about the constant buffer of the matrixes
my more urgent question is how do i feed my normal pipeline with the result of the compute Shader which are 2 RWByteAddressBuffers that contain position an normal
for example i could use 2 vertexbuffer bindings
1 containing only the uv coordinates
2.containing position and normal
How do i copy from the RWByteAddressBuffers to the vertexbuffer ?

(Code from turanszkij )
Here is my shader implementation for skinning a mesh in a compute shader:

• Hi, can someone please explain why this is giving an assertion EyePosition!=0 exception?

It looks like DirectX doesnt want the 2nd parameter to be a zero vector in the assertion, but I passed in a zero vector with this exact same code in another program and it ran just fine. (Here is the version of the code that worked - note XMLoadFloat3(&m_lookAt) parameter value is (0,0,0) at runtime - I debugged it - but it throws no exceptions.
and here is the repo with the alternative version of the code that is working with a value of (0,0,0) for the second parameter.

• Hi, can somebody please tell me in clear simple steps how to debug and step through an hlsl shader file?
I already did Debug > Start Graphics Debugging > then captured some frames from Visual Studio and
double clicked on the frame to open it, but no idea where to go from there.

I've been searching for hours and there's no information on this, not even on the Microsoft Website!
They say "open the  Graphics Pixel History window" but there is no such window!
Then they say, in the "Pipeline Stages choose Start Debugging"  but the Start Debugging option is nowhere to be found in the whole interface.
Also, how do I even open the hlsl file that I want to set a break point in from inside the Graphics Debugger?

All I want to do is set a break point in a specific hlsl file, step thru it, and see the data, but this is so unbelievably complicated