All groups and messages. #3. At some point I have to perform the following loops to visit the colors in a certain radius : for (int. 2 US English only for live captions, and voice access. A flow-control statement determines at run time which block of HLSL statements to execute next. Adding the unroll attribute to loops does not guarantee that the compiler is able to legalize the code. void main () { return ; } However, a return statement can return one or more values. If we wish to add source annotations to more grammatical elements in the future we will encounter more ambiguities because the : character has other meanings in C and modern C++ as well. For example, you can use HLSL to write a vertex shader, or a pixel shader, and use those shaders in the implementation of the renderer in your Direct3D application. In HLSL, the unroll count is the upper limit of number of loops, not the number of unrolled iterations per loop in output IR, as it means in LLVM. Q&A for work. A flow-control statement determines at run time which block of HLSL statements to execute next. Probably your shader model version requieres to have a compile-time constant iteration count, eg. Direct3D 9 shaders can be designed using shader model 1, shader model 2 and shader model 3; Direct3D 10 shaders can only be. . The attributes come in the form of type- and declaration attributes:In general the HLSL compiler likes to unroll loops whenever it can, which is typically when the number of iterations is known at compile time. Use the clip HLSL intrinsic function to simulate clipping planes if each component of the x parameter represents the distance from a plane. I would like for this code to be dynamic so as to reduce the instruction count. It contains an implementation (copyrighted by Nvidia) of a parallel prefix sum algorithm. In your case though you're mostly conflating the [] operator. The compiler simulates loops by default to evaluate whether it can unroll them. This example returns a literal value: float main ( float input : COLOR0) : COLOR0 { return 0; } This example returns the scalar result of an expression. Remarks. Saved searches Use saved searches to filter your results more quicklyJussi said: Another thing to check would be if the Unity Editor reports any errors for the URP/Lit shader when you select it in the Editor. Geometry Shader. Late response to this, but all I can suggest is leveraging the per-component math of HLSL better in your shaders. I would like for this code to be dynamic so as to reduce the instruction count. Oct 24, 2007. soramamenatan. ARB_compute_shader. HLSL中用于优化的指令，[branch],[flatten],[loop],[unroll] branch，shader会根据判断语句只执行当前情况的代码; flatten，shader会执行全部情况的分支代码，然后再根据判断条件获得结果; unroll，for循环是展开的，直到循环条件终止; loop，for循环不展开，Exit the surrounding loop (do, for, while). It takes an array like { 1, 2, 1, 2 } and outputs the accumulative sum from left to right either exclusive { 0, 1, 3, 4 } or inclusive { 1, 3, 4, 6 }. However in general we cannot garantee that we can unroll loops when the bound requires the compiler doing arbitrary math. The regular for loops does also give different results from the unrolled version of the same loop, even on the same local computer. Append(o); 引数として渡した構造体を現在のストリームに出力しています。In HLSL, we can replace the manual computation with the intrinsic function call rcp. Teams. I have encountered a similar issue and I believe it is because you are assigning data to an array inside the loop. It's also settled that a variable coming out of the vertex shader stage for example is varying (HLSL doesn't need this keyword at all!). hlsl. Stop executing the current loop ( do, for, while ), update the loop conditions, and begin executing from the top of the loop. So for every vertex you give the GPU, this code will be. Where details are missing, C++ functionality can be assumed. . Resources Developer Site; Xilinx Wiki; Xilinx GithubNew subject: [PATCH v4 1/1] vkd3d-shader/hlsl: Ignore "unroll" attribute for loops. There were other "safe fix" patches I noticed after 1. In general the HLSL compiler likes to unroll loops whenever it can, which is typically when the number of iterations is known at compile time. That means the compiler will have to create a copy of the body of the loop for each iteration so that there is no loop anymore. loop. If it unrolls the loop and your array is marked as constant (doesn't come from a buffer. First, the reason why the second one works, is because it is automatically unrolled. So after i calculate the filter size in the vertex shader, i pass it to the pixel shader and calculate the around pixel position This is my code to calculate the offset for each pixel around the current pixel. The Overflow Blog Build vs. New subject: [PATCH v8 1/1] vkd3d-shader/hlsl: Ignore "unroll" attribute for loops. Many of them are very similar or identical to C++ counterparts. hlsl","contentType":"file. h","path":"NIS/NIS_Config. As far as I know, you have to declare arrays with a predefined length ( Array syntax needs a positive integer as size ). And unrolling loops is one of those directives. I have another shader of similar size and compilation time is 20 seconds. Compute Shader. vs","contentType":"directory"},{"name":"DXUT","path":"DXUT","contentType. Now this flow control can be set statically in compilation time, or it can be. 前回の記事でジオメトリシェーダーでの頂点分割を試みました。. Low fps on Arm Mali GPU. Sc4Freak. So we’re just left with the min/max component selection, which we can easily unroll and — boom: the. Like. I could go debug HLSL, that is obtainable from context, but I found some more errors, caused by angle in other parts of application. See also. gitignore","path":". That's a no no. Submit and view feedback for. Tool adoption does. a < 128; mip++) { indirection = tex2Dlod. [unroll] turns into spv::LoopControlUnrollMask [loop] turns in. I was able to piece together several other sources. 2 HLSL validation No Yeshlsl; directive; or ask your own question. Default volume texture: _texture = new VolumeTexture(device, size. . True if all components of the x parameter are non-zero; otherwise, false. Direct3D High-level shader language (HLSL) Article 08/04/2021 4 contributors Feedback HLSL is the C-like high-level shader language that you use with. Array reference can not be used as an l-value. non-streaming or streaming) as well as the state of the. Stack Exchange network consists of 181 Q&A communities including Stack Overflow, the largest, most trusted online community for developers to learn, share their knowledge, and build their careers. As far as I´m aware with any branching in HLSL both sides of the branch will be calculated and one will be thrown away. stream. Note: When you use Shader Model 2. Incident update and uptime reporting. Thanks for the reference. Loop aren't automatically unrolled, if you are. In HLSL, #pragma directives are a type of preprocessor directive. For example, the following pragma is equivalent to the /Os compiler option:. HLSL to SPIR-V feature mapping manual. It is available in the latest compiler code, but is not the default. 5. If you're stuck with DX9 then your only alternative is to store your lighting parameters in a texture. 1, + 6. You need to declare blurLength as a static int for this to work. Very much same case as: #1645 In my case problem is scoped to vertex shader, yet I'm expecting similar cases on every-else shader, save fragment. e. Type Description1 Answer. The OpenGL Shading Language is a C-style language, so it covers most of the features you would expect with such a language. {"payload":{"allShortcutsEnabled":false,"fileTree":{"OIT_DX11/AOIT Technique":{"items":[{"name":"AOIT. In this article. Add test for sample mask bulk load/store. @AusSkiller: Putting the declarations outside the for-loop, good idea!GLSL and SPIR-V define implicit derivatives as undefined in the specification in contrast to HLSL which seems to like these cases (e. gitignore. 0 HDRP (windows) and getting warnings. Where HLSL mostly fails (for AnKi) is in the SPIR-V backend of DXC. Note that we use function texCUBElod()—setting the mipmap level explicitly to 0—instead of texCUBE(), because texCUBE(), like tex2D(), would force the DirectX 9 HLSL compiler to unroll the dynamic loop (Sander 2005). The OpenGL Shading Language is a C-style language, so it covers most of the features you would expect with such a language. In HLSL, #pragma directives are a type of preprocessor directive. 5. basic recursive-decent language grammar: declarations, expressions, statements, loops, if-else, functions. Unity outputs this error: Shader error: can't unroll loops marked with loop attribute. 2. So, if you use dynamic branching excessively when you target these profiles, you can receive compilation errors. The transformation can be undertaken manually by the programmer or by an optimizing compiler. OpenGL has vender specific unrolling and stuff. Width, size. don't know what this #pragma is. hlsl. Optionally, you can specify the maximum number of times the loop can execute. So you need to pass a the array length to the shader. g. Append(o) stream. In conclusion . Minimum Shader Model. " Code (CSharp): const float minLayers = 8; const float maxLayers = 32; float numLayers = lerp (maxLayers. You can put #pragma directives anywhere in your HLSL code, but it is a. Note that the following images were captured using the Vehicle Game Sample/Template/Demo which is made. 1 is available in Direct3D 10. Hello fellow Unreal Engine users! This is not really a tutorial in that I guide you through every single step. HLSL shaders can be compiled at author-time or at runtime, and set at runtime into the appropriate pipeline stage. Clang supports a number of AArch64-specific attributes to manage state added by the Scalable Matrix Extension (SME). buy doesn't matter. Use flow-control statements in the compiled shader; do not unroll the loop. 8. Use the [unroll(n)]. Featured on Meta Update: New Colors Launched. The regular for loops. For example you can write the following (dummy) shader program test. Replace the sigma and radius parameters with literal values to let the Unity shader compiler produce efficient code. CEO update: Giving thanks and building upon our product & engineering foundation. But although they should be available the texture object PostProcessInput_0_Texture and sampler object PostProcessInput_0_Sampler defined in PostProcessMaterialShaders. What does it mean? “If” flow controls in HLSL come in two flavors: “branch” and “flatten. To compute a smooth interpolation between 0 and 1, use the smoothstep HLSL intrinsic. This could be a problem for us. x が有限の. 1 Answer. Or in cases where both sides should be shown, you can turn culling off. Teams. HLSL tile shader. It uses glslang for parsing GLSL and compiling SPIR-V. So i +1 and accept the answer. GCN3 introduced 16 bit register support. HLSL 2021 introduces new language features to improve and facilitate the shader authoring experience. vertical path hookup through to SPIR-V generation. Q&A for professional and independent game developers. In this case, the compiler should have emitted a warning, stating that the attribute was unrecognized. you are not allowed to have dynamic variables in the 'for'. The any function determines if any components of the specified value are non-zero, while the all function determines if all components of the specified value are non-zero. Additionally, SHADER_TARGET_GLSL is defined when the target shading language is GLSL (always true for OpenGL/GLES platforms). 在着色器阶段之间传递的所有变量都需要语义。. Prefix sum is also called "prefix scan". ll and is processed by flex to produce C code. The compiler has to be able to fully unroll the loop. More info See in Glossary compiler that isn’t covered by other types of preprocessor directive. This is not supported in legacy GL/GLES targets, so to support this, varying structs are flattened. generic preprocessor (same as GLSL) scanner (keywords, identifiers, etc. このため、今回は…. これに指定された幅の値を掛けると、太さ分に該当する重. vs","path":". D3DCreateLinker. January 19, 2009 10:59 PM. I'm not sure but what I know is that the hardware schedule and process fragments by block of 2x2 (for computing derivatives). The simplest return statement returns control from the function to the calling program; it returns no value. This results in additional clock cycles to enter and exit the loops. 添加了loop标签的for循环不能展开，流式控制每次. There were other "safe fix" patches I noticed after 1. . dynamic does not exist as a valid attribute. Reload to refresh your session. Example code: 1 #pragma unroll <N> 2 for (int i = 0; i < M; ++i) { 3 // Some useful work 4 } In this example, N specifies the unroll factor, that is, the number of copies of the loop that the HLS compiler generates. New subject: [PATCH v5 1/1] vkd3d-shader/hlsl: Ignore "unroll" attribute for loops. 1 I didn't pick any of them. Dynamically Uniform Expressions and Uniform Control Flow" from the GLSL specs I linked above elaborates on how different instances of the shader are executed at the same time, and the flow is "uniform" only when the same instruction is called in all instances of the shader. I believe so, yes. Problems building or running the engine, queries about how to use features etc. . 추가 정보. If you used say tex2dlod (or a texture reading funciton that accepts the ddx and ddy parameters) it would be able to use a real loop. g Sample in non-uniform control flow) and try to unroll the loop. #pragma token-string. Remarks. Qiita Blog. HLSL has compiler hints for unrolling and special branching and such. However sometimes that's not really an option, like when unrolling a loop makes a shader more than 65k instructions, or some other both artificial and spec limitations. You loop a constant amount of iteration, so the compiler can silently unroll without side effects. Or you could use HLSL to write a compute shader, perhaps to implement a physics. New subject: [PATCH v5 1/1] vkd3d-shader/hlsl: Ignore "unroll" attribute for loops. In your case it's doing this, and since you don't provide an upper. It contains detailed information on semantics, syntax, supported features and extensions and much more and is a must-read. yes (pixel shader only), but you must use the legacy compile option when compiling. h","contentType":"file"},{"name":"AppSettings. It seems like a very basic algorithm for parallel computing, so. Sorry. okay, I converted it to tex2Dgrad. 7. } Here, it's possible for the compiler to prove that the count can't be more than 4, so it would be safe to unroll the loop 4x, with. hlsl This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. 0 and all shader types (vertex, pixel, geometry, hull, domain, and compute). fastopt. #2. ll and is processed by flex to produce C code. it will take one cycles to enter the loop and one cycle to exit the loop copy_A. {"payload":{"allShortcutsEnabled":false,"fileTree":{"OIT_DX11/AOIT Technique":{"items":[{"name":"AOIT. When the unroll attribute is applied on a loop and if the loop-trip–count can't be ascertained at compile-time or if it's above a certain threshold, then by default, DXC fails to compile shaders such as the following. Remarks. Typically, pixels outnumber vertices in a scene, so pixel shaders execute more often than vertex shaders. 生成使用流控制执行循环的每个迭代的代码。 与 [unroll] 属性不兼容。 fastopt: 缩短编译时间，但生成较少主动的优化。 如果使用此属性，编译器将不会展开循环。 此属性仅影响支持 中断 指令的着色器模型目标。aah yes, you're right, webgl is quite limited there. Compilation time is with fxc (Windows 8 SDK) about 90 seconds. This allowed the developer to easily and effortlessly compile a single shader into many versions of itself, where each. The #pragma unroll directive however can be used to control unrolling. Writing HLSL Shaders . Unroll / loop attributes as seen in other compilers are always seen as hints. When using DXC to compile the HLSL produced by spirv-cross I get the following error: error: use of undeclared identifier '_189' ui. 4. Re: [PATCH v8 0/1] MR150: vkd3d-shader/hlsl: Ignore "unroll" attribute for loops. So if your early exit can make a whole warp terminate sooner (for example, if threads 0 to 31 all take the early exit), then yes, it is worth it, because the hardware can schedule another warp to execute and that reduces the overall kernel run. hlsl","path. ワイヤーフレームの幅. It works with vs_3_0 and ps_3_0. hlsl', which I can see is there in the. 3. This adds infrastructure suitable for any front end to create SPIR-V loop control flags. The loop can be fully or partially unrolled to create enough hardware to consume the additional data in a single clock cycle. HLSL is the C-like high-level shader language that you use with programmable shaders in DirectX. - lighting and everything off - which Unity seems to want to compile for all possible scene conditions. I try adjustments and recompile one shader but that seems to recompile ALL shaders making. Remarks When the compiler uses the branch method for compiling an if statement it will generate code that will evaluate only one side of the if statement depending on the given condition. Creates a linker interface. these are hints for the hlsl compiler. The transformation can be undertaken manually by the programmer or by an optimizing compiler. In that case it's going to unroll loops for you in unexpected ways because the [] notation is really syntactic sugar. Example code: 1 #pragma unroll <N> 2 for (int i = 0; i < M; ++i) { 3 // Some useful work 4 } In this example, N specifies the unroll factor, that is, the number of copies of the loop that the HLS compiler generates. It is executed concurrently on many GPU cores. Thanks!HLSL也支持for, while和do while循环。和条件语句一样，它可能也会在基于运行时的条件值判断而产生动态分支，从而影响程序性能。如果循环次数较小，我们可以使用属性[unroll]来展开循环，代价是产生更多的汇编指令。用法如下：I am working on a Direct3D rendering application that uses a lot of texture RAM. In your case the number of iterations is fixed at 16, and so its likely that the compiler will unroll the loop. New subject: [PATCH v4 1/1] vkd3d-shader/hlsl: Ignore "unroll" attribute for loops. The following information only applies to. hth Jack Cancel SaveThe solution to this issue is either to make sure that the proper side is always facing the camera. HLSL compilation speed. bool invertSource = false, invertDepth = false; float2 sourceUVC = inputUV, depthUVC = inputUV; // these need to be in the function itself; if they were consts outside then they would be part of the CBUFFER and. generic preprocessor (same as GLSL) scanner (keywords, identifiers, etc. Would a 50-iteration for loop have a significant performance impact during a pixel shader? I tried, just out of curiosity, setting the float4 lights array to have 250 elements, and the shader wouldn’t compile, saying. 0. Posts: 198. I have encountered a similar issue and I believe it is because you are assigning data to an array inside the loop. 3. What you get depends on the kind of conversion. This function is similar to the any HLSL intrinsic function. This means that it has to unroll your loop to support the code you wrote. The shader assembly now has the ability to jump/branch to a new statement of execution, which is noticeable in the compiled code. I mean, it may happen that the shader works for a small part of the image, anything that doesn't take 1024 iterations to render a single pixel (that's crazy, by the way)@user2464424: Yes, NV does have quite a few proprietary GLSL #pragma directives. When unrolling a loop, the iteration count has to be a compile-time constant; without the static on there, the compiler has to assume you might change the value of the parameter at runtime. I notice that if I allocate a Pool. My hlsl shader cannot unroll a loop? 0. Feedback. #2 - External Code Editing and # include. It is executed concurrently on many GPU cores. 2. Shader Model 4. Sc4Freak. branch, flatten -> if statement. Optionally, you can specify the maximum number of times the loop can execute. Compilation time is with fxc (Windows 8 SDK) about 90 seconds. Unroll the loop until it stops executing. You signed out in another tab or window. Ah, very good, thanks for the closure on this one. Note This function is part of the HLSL shader linking technology that you can use on all Direct3D 11 platforms to create precompiled HLSL functions, package them into libraries, and link them into full shaders at run time. It is a revisited itoa() function for HLSL producing resulting ascii codes in an array of uint. You can now read the entire book for free! GDC Talks. This could be a reason that fxc try to unroll the for loop so that the shader program is executed in lockstep mode. Also did you try to use [loop] attribute for generating code that uses flow control? -Microsoft’s next-gen HLSL compiler-Open sourced in January 2017-Based on LLVM/Clang •Spiregg: HLSL to SPIR-V compilation using DXC-Google contributing SPIR-V CodeGen since April 2017-Share front-end parsing, HLSL validation-Recommended DXC for HLSL to SPIR-V compilation Careers. Hi everybody, I am currently encountering a problem with some HLSL custom node. Loops in the C/C++ functions are kept rolled by default. hlsl in front of other paths, or just simply put #include "Common. To turn this failure into a warning, set the language version to 2016. hlsl". しかし粗いポリゴンのモデルにひびを入れる場合は分割数を増やしたいなど、分割具合をもっと柔軟に設定できるようにしたいです。. Zebediah Figura (＠zfigura) Release for 1. Imperfect loop nests cannot be flattened by the tool. Rendering Materials , Shaders , material-functions , hlsl , custom-nodes , question , unreal-engine The HLSL parser is built in two parts: the lexer and the parser. Shader Model 3 (DirectX HLSL) yes (pixel shader only) Shader Model 2 (DirectX HLSL) yes (pixel shader only)New subject: [PATCH v8 1/1] vkd3d-shader/hlsl: Ignore "unroll" attribute for loops. This includes code to declare and initialize variables, write. Reload to refresh your session. It contains detailed information on semantics, syntax, supported features and extensions and much more and is a must-read. 1 is available in Direct3D 10. Color. The source file is hlsl_lexer. Re: [PATCH v8 0/1] MR150: vkd3d-shader/hlsl: Ignore "unroll" attribute for loops. hlsl","path":"GDeflate/shaders/GDeflate. Well that is the thing, I do want to change these at runtime. Was able to get rid of the compiler optimizing by adding an [unroll] before the loop which allowed the setting of the walls variable and subsequent parsing of the array in the vertex shader. buy doesn't matter. 8. Started by 51mon October 18, 2007 10:40 AM. Settings. 相关主题. Preprocessor directive that provides machine-specific or operating system-specific features while retaining overall compatibility with the C and C++ languages. 1 or higher. HLSL semantics. ワイヤーフレームの幅. Use the [unroll(n)] attribute to force an exact higher number. Decluttering your inbox has never been so easy, or looked so good. The reason it's forced to unroll that loop is because of that texture load and the way D3D handles automatic mip level calculations. The particular non-portable pragma you want is #pragma optionNV (unroll all). hatenablog. If you're stuck with DX9 then your only alternative is to store your lighting parameters in a texture. Here is the source code. The decoder ring also has a translation table for concepts and terms used in. This state includes the runtime mode that the processor is in (e. g. int count = min (iterCount, 4); for (int i = 0; i < count; ++i) {. October 23, 2023 12:45. When you design shader algorithms, keep the. Tour Start here for a quick overview of the site Help Center Detailed answers to any questions you might have Meta Discuss the workings and policies of this siteTour Start here for a quick overview of the site Help Center Detailed answers to any questions you might have Meta Discuss the workings and policies of this siteThe Graph Settings tab is important to set up the graph correctly if you’ve selected the Blank Shader Graph or want to switch to a different graph type. Hi, how do I manage to compile the shaders on Ubuntu linux. We don't want to have to do this level of analysis in Naga. glslcc: Cross-compiler for GLSL shader language (GLSL->HLSL,METAL,GLES,GLSLv3) @septag. Does the ATI or nVidia driver unroll the loops (SM 2. Zebediah Figura (＠zfigura) Release for 1. Shader has ~700 lines, which are compiled into ~3000 instructions. Language Syntax (DirectX HLSL) - Programming shaders in HLSL requires that you understand the language syntax, that is, how you write HLSL code. Shader Model 4. Note This function is part of the HLSL shader linking technology that you can use on all Direct3D 11 platforms to create precompiled HLSL functions, package them into libraries, and link them into full shaders at run time. 0 and 3. NEW! GDC 2021. Loop Unrolling (unroll Pragma) The Intel® HLS Compiler supports the unroll pragma for unrolling multiple copies of a loop. HLSL-only or GLSL-only compute shaders. In short: You can't return an array of floats defined in the function in HLSL. Type Description. We don't want to have to do this level of analysis in Naga. ) basic types: all int/uint/float/double scalar/vector/matrix. in advance. 20,295. roots_n[i] is a dynamic access, and it's apparently only because the HLSL compiler is unrolling the loop that it recognizes that this expression will definitely perform an out-of-bounds access. This is a design decision. The downsample shader uses a nested 'for' loop to traverse the uv coordinates. New Here? Create an AccountHLSL is the native shading language for DirectX, so certain HLSL features do not have corresponding mappings in Vulkan, and certain Vulkan specific information does not have native ways to express in HLSL source code. Fragment Shader. For information about writing HLSL itself, see Using HLSL in Unity. All i did was saying #define light cookies instead of #pragma multi_compile, as you mentioned in the 2nd post, in the urp lit shader code and the variants went from 1500M to 250M. ret radians ( x)The rule can be to check whether the texture2D call is inside a loop or when inside a loop of more than N iterations. The DirectX 12 compiler DXC has received a PR about a month ago that adds template support to HLSL. {"payload":{"allShortcutsEnabled":false,"fileTree":{"NIS":{"items":[{"name":"NIS_Config. Get Cell Values 🔗︎. The joys of each vendor implementing. Some flow-control statements support static control that is specified at compile time; others offer. HLSL will let you do it, but when the shader is compiled all constant register accesses have to be explicit (static at compile-time). TextureCubeArray is available in Shader Model 4. "3. #pragma token-string. float xCoordMultiplication : register (c13); float yCoordMultiplication : register (c14); Make that a float2. Example. Well you have a few options: -Use a loop/ifs like you have. For clarification, if considering my Python results as the correct solution, unrolling the loop in HLSL gives me better results than what not unrolling gives. I have one relatively complicated shader, which I want to compile. TLDR: use the attribute [branch] in front of your if-statement. Image Effect Shaders have culling off by default if you’d like to see an example. With these requirements, I put the following constraints: #pragma HLS LATENCY min=500 max=528 // directive for FUNCT #pragma HLS UNROLL factor=1 // directive for L0 loop However, the synthesized design results in function latency over 3000 cycles and the log shows the following warning. You can insert the Custom expression from the Custom category in the Material palette or from the right-click search menu in the Material Graph. fx shader file uses some uniform parameters for a function: VSOutputTxEnvMap ComputeEnvMapVSOutput (VSInputNmTx vin, float3 normal,. I'm coming from a DX9 background, using the now outdated FX framework, where uniform variables can be used as a sort of preprocessor value that changes the compilation of a shader. -Write multiple shaders/techniques/passes to handle the different combinations. CEO update: Giving thanks and building upon our product & engineering foundation. Defines the number of threads to be executed in a single thread group when a compute shader is dispatched (see ID3D11DeviceContext::Dispatch ). More info. That means you're sampling a texture using tex2D in a variable length loop.