digitalmars.D - [GSoC 2026] DCompute Vulkan Backend
- Quang Ha (17/17) May 09 Hi everyone,
- Quang Ha (46/46) May 24 Hi everyone, here is my progress update for **Week 1** of the
- Quang Ha (47/47) Jun 01 ### My GSoC Progress: Week 2 Update
- Quang Ha (29/29) Jun 09 ### My GSoC Progress: Week 3 Update
- Quang Ha (34/34) Jul 08 ### My GSoC Progress: Back from Exams & A Surprise Bug!
- FinalEvilution (2/7) Jul 08 Interesting. What's the bug? If you don't mind me asking.
- Quang Ha (49/50) Aug 21 ### My GSoC Progress: Fixing LLVM and Running Vulkan Compute! 🚀
- Richard (Rikki) Andrew Cattermole (2/2) Aug 21 Very nice, I'm sure having to fix a bug in LLVM wasn't in your plan for
- Richard (Rikki) Andrew Cattermole (2/4) Aug 21 Err GSoC not SAoC oops.
- Quang Ha (4/6) Aug 27 yes, it is ^^, I just spent a few days writing a resume and
- FinalEvilution (35/54) Aug 23 Don't worry about it. I figured you were busy and i didn't take
- Quang Ha (5/6) Aug 27 Thank you so much for your interest! I just spent a few days
- FinalEvilution (6/12) Aug 27 Good luck with your application!
Hi everyone, I am making this thread to share my weekly updates for the **DCompute Vulkan Backend** project. Here are the 3 main parts of my project (Mentored by Nicholas Wilson): * **LDC Compiler:** Connect D's data types with Vulkan's memory rules by adding a wrapper kernel. I will also fix pointer sizes to make correct SPIR-V files. * **Testing:** Check the compiled `.spv` files with Vulkan tools to make sure the memory works right. * **DCompute Runtime:** Build the host-side system to start the Vulkan device and run the code. I will post my progress, problems, and questions as replies to this thread every week. This helps keep everything in one place. I am excited to share my work with you all! GitHub: [https://github.com/gnavdev28](https://github.com/gnavdev28)
May 09
Hi everyone, here is my progress update for **Week 1** of the DCompute Vulkan Backend project. * Squashed [Nicholas's PR](https://github.com/ldc-developers/ldc/pull/4958) to thoroughly analyze the required implementation details in `targetVulkan.cpp` and `dcomputecodegenerator.cpp`. * Investigated the `TODO` in `dcomputecodegenerator.cpp` and implemented support for parsing the `vulkan-{version}` target value by adapting the existing OpenCL (OCL) and CUDA logic. * Updated `cl_options.cpp` to include Vulkan in the `--mdcompute-targets` flag's description. * Addressed the hardcoded 64-bit pointer assignment in `targetVulkan.cpp` to fix the layout issue for different pointer sizes. * Refactored relevant files by updating the LLVM version macro checks from `LLVM_LDC_VER >= 2100` to `LLVM_VERSION_MAJOR >= 23`. * **The .spv file output problem:** I used the `--mdcompute-targets` flag to make a `.spv` file, but it was in COFF format, not real SPIR-V. It was very strange. When I added a logging line just to print things out, it made correct SPIR-V code. After that, I built the newest LLVM 23 (on May 24), and it went back to COFF format. Nicholas and I didn't understand why. Finally, I fixed it. I copied the "HACK" part from `DComputeTarget::emit` to `DComputeTarget::writeModule`. Now it works normally. * **Build error:** I got a linker error when building. But it was easy to fix. I just changed the paths in the `CMakeLists.txt` file. * **The TODO part:** There is a `TODO` for `spirv.Layout` in `targetVulkan.cpp`. It is very difficult to understand. So, I haven't touched it yet. I need more time to study it. * **Look into spirv.Layout:** I will spend time studying the `TODO` about `spirv.Layout` in `targetVulkan.cpp` because I didn't finish it this week. * **Write more tests:** I will write some more test files to make sure the compiler always makes the correct `.spv` files. * **Update my PR:** I will wait for Nicholas to finish reviewing my PR. Then, I will update the code and fix any issues he points out. * Commits/PRs: https://github.com/ldc-developers/ldc/pull/5132 * My Personal Blog: Will update soon! Feel free to leave any feedback or suggestions!
May 24
Hi everyone, here is my progress update for **Week 2** of the
DCompute Vulkan Backend project.
* **Updated `target.cpp`:** I added `gABI = abi; gIR = _ir;
gTargetMachine = targetMachine;` to `writeModule` to help the
compiler write the correct file format.
* **Added missing tags:** I added ` CompileFor.deviceOnly` to
`test_vulkan.d` so it can emit the module. I also added the
` kernel()` tag to the test functions so the compiler knows the
entry point.
* **Fixed 3 big crashes in `targetVulkan.cpp`:** This was my
most important work this week. I updated the LLVM intrinsic calls:
* Deleted `i1false` from the `handlefrombinding` call to
fix the bad signature.
* Added `i32zero->getType()` and changed `i32one` to
`i32zero` to fix the `getpointer` invalid index.
* Wrote a `CreateStructGEP` loop to get each `fieldTy` and
load them one by one. This replaced the old `CreateAlignedLoad`
for the big struct and fixed the element type error.
* **Checked `isWritable`:** I used `spirv-dis` to check the
`TargetExtType`. I saw `OpMemberDecorate %_struct_20 0
NonWritable`. I am researching if the number should be `0` or `1`.
* **The COFF file problem:** At first, my compiler made a COFF
file, not a `.spv` file. Sometimes it was right, sometimes wrong.
It was very strange until I updated `writeModule`.
* **Missing Kernel and "not set" spam:** The `.spv` file did
not have `OpEntryPoint` and `_kernel`. Also, when I put a log
inside `addKernelMetadata`, my screen was full of `"->
dcomputetarget: not set"`. I found out I just missed the
` CompileFor.deviceOnly` and ` kernel()` tags.
* **The Compiler Crashes:** Right after the file emitted, the
compiler crashed 3 times in a row! First, it had a bad signature.
Second, it had an invalid index. Third, LLVM did not let me load
a big struct at the same time. I had to debug and fix all of them
in `targetVulkan.cpp`.
* **Update my PR:** I will wait for Nicholas to check my pull
request and I will update my code.
* **Find `isWritable` answer:** I will find the final answer
for the `isWritable` number (0 or 1).
* **Write more tests:** I will write more test files to make
sure everything works perfectly.
* Commits/PRs: https://github.com/ldc-developers/ldc/pull/5132
* My Personal Blog: Updating soon!
Jun 01
Hi everyone, this is my update for Week 3. This week, I did not
write a lot of new code for the compiler. Instead, I focused on
testing to make sure everything works correctly in real life!
* **Added a new test file:** I wrote another test file for the
compiler to generate the `.spv` file.
* **Cloned a test project:** I cloned a repository called
`vulkan_minimal_compute` from GitHub.
* **Tested the Vulkan API:** I changed some code in the
`vulkan_minimal_compute` project. I want to use it to run my
generated `.spv` file. This helps me check if the Vulkan API can
read and use my file correctly.
* **Understanding the real environment:** It takes time to
read and understand the `vulkan_minimal_compute` repo. Setting up
a real Vulkan environment to test the `.spv` file is a bit slow
and requires a lot of checking.
* **Finish the API test:** I will finish modifying the
`vulkan_minimal_compute` project and fully test my `.spv` file.
* **Fix bugs (if any):** If the Vulkan API finds any problems
with my `.spv` file, I will go back to the compiler and fix them.
* **Wait for PR review:** I am still waiting for Nicholas to
review my pull request so I can update my code.
* Commits/PRs: https://github.com/ldc-developers/ldc/pull/5132
* My Personal Blog: https://gnavdev28.github.io (nothing special
yet :D)
Jun 09
Hi everyone, this is my new update. I took a short break from the project because I had to study for my university final exams. Now, I am back to work! * **Fixed the Linker Error:** First, I updated my LDC and LLVM to the newest versions. But then, I got a very bad linker error and could not build the compiler. Finally, I found a trick! I added a new entry in CMake: `CMAKE_MSVC_RUNTIME_LIBRARY` with the value `MultiThreaded`. This fixed the build. * **Continued ABI Testing (from Week 3):** I went back to my unfinished work from Week 3. I used the `vulkan_minimal_compute` project to do ABI testing and check if my `.spv` file works with the Vulkan API. * **Found an LLVM Bug!:** This is the most interesting part of the week. While I was doing the ABI testing, I found a bug. But it was not my bug, it was a bug inside LLVM! So now, I am also trying to fix this LLVM bug. * **Getting back on track:** Taking a break for exams made me forget some small details. It took me a little time to remember how the ABI testing works from Week 3. * **Fixing LLVM:** Finding a bug in my own code is normal, but finding a bug inside LLVM is a big surprise. The LLVM codebase is very huge, so fixing it is a big challenge for me. * I will focus on fixing the LLVM bug I found. * I will finish the ABI testing with the `vulkan_minimal_compute` project. * I will clean up my code and wait for my mentor to review my pull request. * Commits/PRs: https://github.com/ldc-developers/ldc/pull/5132 * My Personal Blog: https://gnavdev28.github.io
Jul 08
On Thursday, 9 July 2026 at 00:39:55 UTC, Quang Ha wrote:* **Found an LLVM Bug!:** This is the most interesting part of the week. While I was doing the ABI testing, I found a bug. But it was not my bug, it was a bug inside LLVM! So now, I am also trying to fix this LLVM bug.Interesting. What's the bug? If you don't mind me asking.
Jul 08
Interesting. What's the bug? If you don't mind me asking.First of all, I want to sincerely apologize to FinalEvilution and everyone for the long silence. I have been deep in the debugging trenches for a while, but I am finally back with the ultimate update! To answer FinalEvilution's question: *"Interesting. What's the bug? If you don't mind me asking."* — The short answer is that the LLVM SPIR-V Backend was creating an invalid instruction when compiling shaders for Vulkan. You can find the full technical details of how I fixed this in the LLVM section below! Now, onto the best news: the project is successfully done! I am just waiting for the final reviews and merges to officially wrap up GSoC. After this, my next big stop is DConf 2026! * **Fixed Bugs in LLVM SPIR-V Backend:** * When compiling shaders for Vulkan, LLVM created an invalid instruction (`OpBitcast` on pointers). I fixed it by using `OpAccessChain` instead. * Added support for `PhysicalStorageBuffer` in Vulkan so pointer addresses work correctly. * Made the code clean and generic so any language can use it, not just D. * Added a new test file in LLVM (`physical-storage-buffer-load-store.ll`) and opened a pull request to LLVM. * **Fixed LDC ABI Issues:** * Fixed how D passes pointers to Vulkan compute shaders so the GPU can read the buffer directly. * Updated the target detector in LDC to recognize `spirv`, `spirv32`, and `spirv64`. * **Tested on Real GPU:** * Tested with our C++ Vulkan test program. * Both the minimal test (`67.36f`) and Vector Addition (A + B = C) ran successfully and gave the correct output! * **Understanding LLVM Codebase:** LLVM is very large, so finding where the bug happened took a lot of time and debugging. * **Clean Code:** My mentor, Nicholas Wilson, helped me review the code to make sure it follows all LLVM coding rules. * Follow the LLVM pull request and reply to reviewers. * Fix the CI issue in the LDC repository. * Get ready for my upcoming online talk: "Vulkan Compute Support in DCompute and LDC" at DConf 2026 this September! * Write more compute test cases in D for Vulkan. * **LDC PR:** https://github.com/ldc-developers/ldc/pull/5132 * **LLVM PR:** https://github.com/llvm/llvm-project/pull/216919 * **My Personal Blog:** https://gnavdev28.github.io
Aug 21
Very nice, I'm sure having to fix a bug in LLVM wasn't in your plan for SAoC!
Aug 21
"Richard (Rikki) Andrew Cattermole" <richard cattermole.co.nz> On 22/08/2026 9:25 AM, Richard (Rikki) Andrew Cattermole wrote:Very nice, I'm sure having to fix a bug in LLVM wasn't in your plan for SAoC!Err GSoC not SAoC oops.
Aug 21
On Friday, 21 August 2026 at 21:25:46 UTC, Richard (Rikki) Andrew Cattermole wrote:Very nice, I'm sure having to fix a bug in LLVM wasn't in your plan for SAoC!yes, it is ^^, I just spent a few days writing a resume and proposal to apply for SAoC too!
Aug 27
On Friday, 21 August 2026 at 16:37:45 UTC, Quang Ha wrote:First of all, I want to sincerely apologize to FinalEvilution and everyone for the long silence. I have been deep in the debugging trenches for a while, but I am finally back with the ultimate update!Don't worry about it. I figured you were busy and i didn't take any offense, and I'm sure everyone else feels the same.To answer FinalEvilution's question: *"Interesting. What's the bug? If you don't mind me asking."* — The short answer is that the LLVM SPIR-V Backend was creating an invalid instruction when compiling shaders for Vulkan. You can find the full technical details of how I fixed this in the LLVM section below!Even the experts over at LLVM get it wrong sometimes. I have my own (unrelated) LLVM bug i need to create an issue for. I tried building your LLVM pr but it failed with. ``` llvm-project/llvm/lib/Target/SPIRV/SPIRVEmitIntrinsics.cpp:945:40: error: ‘SrcTy’ was not declared in this scope; did you mean ‘STy’? 945 | if (auto *STy = dyn_cast<StructType>(SrcTy)) { | ^~~~~ | STy ``` Is there perhaps a commit that hasn't been pushed to the repo yet? Ps. You left a little vietnamese in llvm/lib/Target/SPIRV/SPIRVLegalizePointerCast.cpp:233 I also took a look at the ldc pr and i was able to build it and one of the test kernels as well. I don't have a vulkan test environment setup yet, but i have a cuda project i would like to port to D and i think that it could be a good exercise. The only issue i was able to find were that a build without vulkan support (LLVM 22) prints ``` Error: LDC was not built with Vulkan DCompute support. Error: Unrecognised or invalid DCompute targets: the format is ocl-xy0 for OpenCl x.y and cuda-xy0 for CUDA CC x.y, and vulkan-xy0 for Vulkan. Valid version strings for OpenCl are ocl-{100, 110, 120, 200, 210, 220}. Valid version strings for Vulkan are vulkan-{VULKAN_VALID_VER_INIT}. ```Now, onto the best news: the project is successfully done! I am just waiting for the final reviews and merges to officially wrap up GSoC. After this, my next big stop is DConf 2026!Congratulations on your success! Figuring out both LLVM and SPIR-V is no small feat!* Follow the LLVM pull request and reply to reviewers. * Fix the CI issue in the LDC repository. * Get ready for my upcoming online talk: "Vulkan Compute Support in DCompute and LDC" at DConf 2026 this September! * Write more compute test cases in D for Vulkan.Looking forward to the talk!
Aug 23
Looking forward to the talk!Thank you so much for your interest! I just spent a few days writing my resume and proposal to apply for SAoC, and I hope the organization will continue to trust me. I wonder if you've still encountered the issues you mentioned over the past few days? Please let me know!
Aug 27
On Friday, 28 August 2026 at 02:45:24 UTC, Quang Ha wrote:Good luck with your application! There have been no new commits or force pushes to ether pr since i did my tests, and therefore the issues are still there. If you want or need any help testing things get in touch. I'm not much of a programmer but i can compile things :)Looking forward to the talk!Thank you so much for your interest! I just spent a few days writing my resume and proposal to apply for SAoC, and I hope the organization will continue to trust me. I wonder if you've still encountered the issues you mentioned over the past few days? Please let me know!
Aug 27