15089cf7b5I did the thing, I moved all textures into one big descriptor array. Now each frame looks up a texture in an array. But this has absolutely destroyed performanceAndreew Gregory2026-02-14 22:35:47 +03:00
2367ce1e9dMargaretImgAllocation now stores a direct pointer to MargaretImgAllocatorOneBlock. And both margaretImgAllocator and margaretBufAllocator blocks store pointers to allocators that created them. Memory allocators (buffer, images) are now stored on heap. Now I can do a simplification of allocations management: I don't need to pass allocator to allocation method. Though I hadn't refactored that yet...Andreew Gregory2026-02-14 03:00:03 +03:00
bca52758ccSplit 0a pipeline set into light config set and texture set. Replaced light array ubo with light array storage buffer. But while writing this I realized that my current approach to staging buffer handling is trash, so I will need to rewrite everything I wrote yesterday :(Andreew Gregory2026-02-13 18:51:23 +03:00
8cc0b43a5dWow! I removed std140 padding of matrices in geom.h and nothing brokeAndreew Gregory2026-02-12 21:05:22 +03:00
5b8ecd8020Little refactoring of alice: Added Abigail urility (transfer_in_mainloop.h). Buffers/textures that get copied to device local memory and then never modified occured many times in alice and lucy and was generally worth moving into a separate class.Andreew Gregory2026-02-12 20:58:23 +03:00
71f73964cfStashing progress... Don't take this commit seriouslyAndreew Gregory2026-02-12 17:54:11 +03:00
8c928841adFixed dumb color blending bug in margaret. Discovered it when set non-black wallpaperAndreew Gregory2026-02-11 19:12:49 +03:00
1e6097ac2aI wrote program that renders sekelton.glb. But what's the point now? It does not draw as expected. Spine and jaw are gone. Hierarchy makes no sense. Plus my game erendering class sucks. And as if it couldn't already be worse, HSE just dropped like 7 homeworks and tomorrow I can't afford to be late for stupid fucking classes. How is it that I hate gltf rendering 10 much more than homework. Fuck this fucking shit. Fuck prob-theory. Fuck caos. Fuck every single zoom lesson. Fuck ML. Fuck khronos group. Fuck blender. Fuck , fuck, fuck. I fucking hate glb format. Fuck this shit. And I am sick of C already
parsing-scenes
Andreew Gregory2026-02-09 01:25:43 +03:00
795f675e6aSaving progres on glb parsing. All checks of structures are doneAndreew Gregory2026-02-08 17:00:48 +03:00
24a8be3a75Okay, we have all data. Now what? (tldr: I marshalled all the json data from gltf. I did not verified it. I did not tested it. I did not imported it. Nothing is done)Andreew Gregory2026-02-08 02:41:07 +03:00
2dd4ad0d7bSaving progress. Parsing of nodes and scenes arrays of glb objectAndreew Gregory2026-02-06 17:44:36 +03:00
09df9f4e2dSyntax error fix + partially wrote GltfNode and GltfScene structures. Sometimes I do wonder if writing 'Gltf' look good or not...Andreew Gregory2026-02-05 13:15:53 +03:00
f0634d08a1OMG OMG IT WORKS. Достигнута цель для КТ1: симуляция одного тёрдого тела в невесомости. This entire quaternion thing turned out to be irrelevant
physics-app
Andreew Gregory2026-02-02 22:34:47 +03:00
53dd36ab7cWrote a cute little program where you can shoot bullets at ONE FOURTH OF A CYLINDER. Too bad they still don't rotate the cylinderAndreew Gregory2026-02-02 02:48:49 +03:00
f99e673d79Refactored most of l1 using brand new VecU8_append_fmt method and brand new %r VecU8_fmt specifierAndreew Gregory2026-02-01 21:05:43 +03:00
0c41f3dd3eSaving... I was in the process of writing my physics sim when I noticed that Ozon delivered my new headphones. Which would also imply that I am already too lateAndreew Gregory2026-02-01 03:01:01 +03:00
26a2badb94I wrote a little bit of tests. But that is it, I have to hurry up for another cool university thing. Look what our supervisor said about this: Е"Если вам это не удобно, то я вас поздравляю, вы неудачник". You don't need to be genius to know that this shit is realAndreew Gregory2026-01-27 02:55:58 +03:00
78fd9f0816Just wrote basic json support. Had not tested yet. In fact, I haven't even finished debugging marie_triangulation algorithm. I am truly running out of timeAndreew Gregory2026-01-26 18:47:49 +03:00
9a9a5b1b0fFixed VkDevice initialization mistake. Single queue family case is now handled correctly. Fixed Lucy bug, removed a lot of bloat, now LucyImages are stored in a vector each slot taking one descriptor array element. It should be noted that my previous"lucy fix" from prev commit is a complete crap and got removed. Also I downloaded some cool fonts yay :]Andreew Gregory2026-01-01 21:07:55 +03:00
284b0b711bFound a serious flaw in Lucy design. A very very serious flaw. Gonna rewrite everythingAndreew Gregory2026-01-01 15:29:05 +03:00
238a3653e8Опять налажал с текстурами. И аллокатор изображений опять поломался. Из хорошего: в r4 можно ходить по полюAndreew Gregory2025-12-30 18:54:47 +03:00
b1c5fca4b1Wrote r4. Not a full game, but we have an engine showcase in haskellAndreew Gregory2025-12-30 13:16:19 +03:00
f1d42f37b9До тотального отнюха бебры осталось 11 часов. Я сохраняю модель в дурацком формате и читаю их потом из файлаAndreew Gregory2025-12-30 00:16:52 +03:00
cf237d1c0bFixed some bugs in MargaretImgAllocator and MargaretBufAllocator. Refactored r0, now it is not r0, I added Alice game engince namespace. I added mesh handler type, all stupid SceneTemplate crap is gone. There is no limit on model count and isntance count. I added haskell bindings for running Alice. Unfortunately, calculus exam is after-tomorrow, I don't have much time to organize my codeAndreew Gregory2025-12-27 21:35:23 +03:00
6a0681b42eRemoved more unrelated stuff that I don't want to touch anymoreAndreew Gregory2025-12-23 23:02:09 +03:00
1328d194beRemoved unrelated old stuff that is completely unrelated nowAndreew Gregory2025-12-23 22:51:36 +03:00
19f92d9207I wrote Lucy. It works. I can render text. Example is written in r0. Too bad I am running out of time. Exam is the next day after tomorrow :(Andreew Gregory2025-12-23 22:47:15 +03:00
5615594762Added this stupid normal vector inference bullshit from learnopeng.com. I checked, it's garbage. Will probably change it. But now I have to write GenericModel normal vector inference. And it's, like, 200x times harder. And normal texture generation is the type of hell you don't just drop on somebody. I dropped it on myself anyway...Andreew Gregory2025-12-19 01:21:01 +03:00
8e3a306459I FINALLY REWROTE r0 TO USE NORMAL STUFF!! YEEEES. AFTER 100 years. But now session is back on my ass, againAndreew Gregory2025-12-14 04:42:35 +03:00
aadc346f43What I have been doing for two months could actually be have been done in one day effortlesslyAndreew Gregory2025-12-11 02:12:21 +03:00
6f418827dcI just realized I wasted 100 days of my life, of the most important period of my life on a useless shit that served no purpose. I just deleted MargaretMemAllocatorAndreew Gregory2025-12-10 21:32:16 +03:00
202e11ab56Saving progress. I just realized that my current MaragertMemAllocator is still very very bloated. I am about to change that. Yep, another rewriteAndreew Gregory2025-12-08 23:31:05 +03:00
f80dc0ded0Removed useless crap that I sepnt a lot of time on, but din't actually planeed to use. src/l_15/anne/marie/clipping.h says: Please, forget that I had ever livedAndreew Gregory2025-12-08 19:59:54 +03:00
cadde8714bFixed r0.c problems. Finally, now r0.c runs on MargaretMemAllocator!!!!! And everything is wroking!!!Andreew Gregory2025-12-02 20:09:29 +03:00
fac2fde22bFinished rewriting r0 to MargaretMemAllocator. Changed MMA interface during refactoring. It compiles. Finally. It took only 5 weeks to write compiling version. Unfortunately, it crashed before even starting. Today will be the long dayAndreew Gregory2025-12-02 04:52:06 +03:00
dc67475e7aSaving progress. I am getting insane just by looking at this crap. November is over. Engine is still not done. I can't take this anymoreAndreew Gregory2025-12-01 01:48:46 +03:00
438015b842Saving progress. Rewrote r0_scene.h, r0_assets.h, vk_ctx struct according to new design. Но потом я отвлёкся от прогресса из-за обязанностей няньки. Anyway. I need to test somrthing really quick. Going back to masterAndreew Gregory2025-11-30 03:54:02 +03:00
8ec7bff490I think I finally wrote MargaretMemAllocator. I hadn't even tested it. It is 3:44. Were you aware of 'Tractat about Reptiles'? Me neither. MargaretMemAllocator is the most multifunctional allocator possible, you will still need separate allocator for buffersAndreew Gregory2025-11-28 03:46:06 +03:00
d5854dd5a3Improved rbtree, bufrbtree, wrote shitty liked list. Progress update on MMA: almost complete, but I have to blow some dust off my knowledge of pipeline barriersAndreew Gregory2025-11-27 18:12:44 +03:00
2ea49d61d7Was in the proces of finishing my allocator, when I realized that NO THIS WHOLE API IS TOTALLY DUMB, why would I take a pointer to a place to place a pointer to a node and then update that pointer to a node through a pointer to a pointer, which is somehow stored in the node? That does not make any sense at all. I am so bad at programming, oh my god AAAAAAAA, I am going insane. I am actually gonna go insane. I am rewatching Parkour Civilization third time already instead of writing double-linked list. Yes. I am gonna store block in a double-linked list. And ou yeah, I am gonna learn how to transfer allocated RB nodes between RB trees, yes.Andreew Gregory2025-11-25 19:45:10 +03:00
98af159dbcУдалил кучу хлама. Сохраняю прогресс. Иду читать самый длинный щитпост в миреAndreew Gregory2025-11-24 14:04:30 +03:00
cd6777a3b6I just realized that subbuffering is entirely useless waste of time. I wasterd more then 3 weeks pn that nonsense. It made MMA super overcomplicated. I am deleting ALLL of this in the next commitAndreew Gregory2025-11-22 00:41:42 +03:00
8a2e8b48adCollapsed a lot of copypasted code. Hope I didn't broke anythingAndreew Gregory2025-11-21 01:50:11 +03:00
4cd6e65ea9Almost wrote RBTree, almost refactored RBTree. Minor methods missing. All tests passed. Method _equal_ is no longer requires for set/map instantiationAndreew Gregory2025-11-21 00:43:59 +03:00
704930efccWrote RBTree. The real pointer-based rbtree. The fake one (with buffers) requires purging. But I have yet to debug the stupid delete operationAndreew Gregory2025-11-20 18:00:23 +03:00
310b56e372УУф... Только я наконец закончил придумывать этот алгоритм и почти написал его, как вдруг я обнаружил ошибку и начал всё переосмыслять. Только я пошел по прямой, как наступила ночь, а завтра в институт ((((Andreew Gregory2025-11-17 03:32:35 +03:00
1a45c22952saving progress. Wrote __get_rid_of_memory_occupant methodAndreew Gregory2025-11-13 18:24:44 +03:00
6ba756fe1aI rethought everything again (how surprising) + fixed some bugs at map templateAndreew Gregory2025-11-12 16:01:24 +03:00
e68a16d8fcSaving progress. Rewrote rb_tree_set_map_template_inst.h ENTIRELY just to have Map<K, V> support. Nothing works right nowAndreew Gregory2025-11-10 18:45:40 +03:00
49ee178eb6Saving progress on MargaretMemAllocator. The exam is near, can't do this anymoreAndreew Gregory2025-11-04 12:24:08 +03:00
497808ddebFixed hella lot of bugs in rb tree. Wrote tests for BuffRBTreeBy***_SetU64Segment (they work)Andreew Gregory2025-11-01 15:26:27 +03:00
941a6da1d7I would absolutely love to write my GPU memory allocation algorithm, but unfortunately calculus colloquium is a thing. Yes, holydays are fucking over, I spent them watching Hermitcraft 9, now 228 deadlines/second are back on my assAndreew Gregory2025-10-29 02:24:42 +03:00
e09238a8e7Merged with adding-extures branch. Removed _int_primitives.h template amalgamation. Yet another refactoring is doneAndreew Gregory2025-10-08 17:04:44 +03:00
122f3dba24int_ptimitives template instantiation and vec,cvec templ instantiation split into many separarate filesAndreew Gregory2025-10-08 16:42:10 +03:00
Andreew Gregory