Merge pull request #42 from jbikker/dev

Merging dev into main.

.github/workflows/tinybvh_sdk_ci.yml

@@ -2,9 +2,9 @@ name: tinybvh CI
 
 on:
   push:
-    branches: [ main ]
+    branches: [ main, dev ]
   pull_request:
-    branches: [ main ]
+    branches: [ main, dev ]
 
 jobs:
   build:

README.md

@@ -1,3 +1,6 @@
+# dev
+This is the **development branch** for tinybvh. Potentially unstable code, consider yourself warned.
+
 # tinybvh
 Single-header BVH construction and traversal library written as "Sane C++" (or "C with classes"). The library has no dependencies. 
 
@@ -9,6 +12,8 @@ Single-header OpenCL library, which helps you select and initialize a device. It
 * Vendor and architecture detection and propagation to #defines in OpenCL code
 * ..And many other things.
 
+![Rendered with tinybvh](images/test.png)
+
 To use tinyocl, just include ````tiny_ocl.h````; this will automatically cause linking with ````OpenCL.lib```` in the 'external' folder, which in turn passes on work to vendor-specific driver code. But all that is not your problem!
 
 Note that the ````tiny_bvh.h```` library will work without ````tiny_ocl.h```` and remains dependency-free. The new ````tiny_ocl.h```` is only needed in projects that wish to trace rays _on the GPU_ using BVHs created by ````tiny_bvh.h````. 
@@ -31,6 +36,7 @@ A constructed BVH can be used to quickly intersect a ray with the geometry, usin
 
 The constructed BVH will have a layout suitable for construction ('````WALD_32BYTE````'). Several other layouts for the same data are available, which all serve one or more specific purposes. You can convert between layouts using ````BVH::Convert````. The available layouts are:
 * ````BVH::WALD_32BYTE```` : A compact format that stores the AABB for a node, along with child pointers and leaf information in a cross-platform-friendly way. The 32-byte size allows for cache-line alignment.
+* ````BVH::ALT_SOA```` : This format stores bounding box information in a SIMD-friendly format, making the BVH faster to traverse.
 * ````BVH::WALD_DOUBLE```` : Double-precision version of ````BVH::WALD_32BYTE````.
 * ````BVH::VERBOSE```` : A format designed for modifying BVHs, e.g. for post-build optimizations using ````BVH::Optimize()````.
 * ````BVH::AILA_LAINE```` : This format uses 64 bytes per node and stores the AABBs of the two child nodes. This is the format presented in the [2009 Aila & Laine paper](https://research.nvidia.com/sites/default/files/pubs/2009-08_Understanding-the-Efficiency/aila2009hpg_paper.pdf) and recommended for basic GPU ray tracing.
@@ -57,6 +63,10 @@ The cross-platform fenster-based single-source **bitmap renderer** can be compil
 
 ```g++ -std=c++20 -mavx -O3 -framework Cocoa tiny_bvh_fenster.cpp -o tiny_bvh_fenster``` (on macOS)
 
+The multi-threaded **ambient occlusion** demo can be compiled with
+
+````g++ -std=c++20 -mavx -mwindows -fopenmp -O3 tiny_bvh_pt.cpp -o tiny_bvh_pt```` (on windows)
+
 The **performance measurement tool** uses OpenMP and can be compiled with:
 
 ````g++ -std=c++20 -mavx -Ofast -fopenmp tiny_bvh_speedtest.cpp -o tiny_bvh_speedtest````

external/fenster.h

@@ -199,8 +199,8 @@ static LRESULT CALLBACK fenster_wndproc(HWND hwnd, UINT msg, WPARAM wParam,
     PAINTSTRUCT ps;
     HDC hdc = BeginPaint(hwnd, &ps);
     HDC memdc = CreateCompatibleDC(hdc);
-    auto hbmp = CreateCompatibleBitmap(hdc, f->width, f->height);
-    auto oldbmp = SelectObject(memdc, hbmp);
+    HBITMAP hbmp = CreateCompatibleBitmap(hdc, f->width, f->height);
+	HBITMAP oldbmp = static_cast<HBITMAP>(SelectObject(memdc, hbmp));
     BINFO bi = {{sizeof(bi), f->width, -f->height, 1, 32, BI_BITFIELDS}};
     bi.bmiColors[0].rgbRed = 0xff;
     bi.bmiColors[1].rgbGreen = 0xff;
@@ -249,10 +249,13 @@ FENSTER_API int fenster_open(struct fenster *f) {
   wc.hInstance = hInstance;
   wc.lpszClassName = f->title;
   RegisterClassEx(&wc);
+  RECT desiredRect = {0, 0, f->width, f->height};
+  AdjustWindowRectEx(&desiredRect, WS_OVERLAPPEDWINDOW, FALSE, WS_EX_CLIENTEDGE);
+  int adjustedWidth = desiredRect.right - desiredRect.left;
+  int adjustedHeight = desiredRect.bottom - desiredRect.top;
   f->hwnd = CreateWindowEx(WS_EX_CLIENTEDGE, f->title, f->title,
                            WS_OVERLAPPEDWINDOW, CW_USEDEFAULT, CW_USEDEFAULT,
-                           f->width, f->height, NULL, NULL, hInstance, NULL);
-
+                           adjustedWidth, adjustedHeight, NULL, NULL, hInstance, NULL);
   if (f->hwnd == NULL)
     return -1;
   SetWindowLongPtr(f->hwnd, GWLP_USERDATA, (LONG_PTR)f);
@@ -261,7 +264,10 @@ FENSTER_API int fenster_open(struct fenster *f) {
   return 0;
 }
 
-FENSTER_API void fenster_close(struct fenster *f) { (void)f; }
+FENSTER_API void fenster_close(struct fenster *f) {
+  PostMessage(f->hwnd, WM_CLOSE, 0, 0);
+  (void)f;
+}
 
 FENSTER_API int fenster_loop(struct fenster *f) {
   MSG msg;