#asahi-gpu · 2021-01-12 13:33
bloomaustriancoder: right, the problem with that is -- afaiu -- Vulkan requires a deep understanding of the hardware, whereas I'm at the point I could throw together a gallium driver passing most of deqp-gles2 by just bitbanging magic numbers#asahi-gpu · 2021-01-12 03:00
bloomGL-on-VK is a thing but meh#asahi-gpu · 2021-01-12 03:00
bloomWith 2x the work..#asahi-gpu · 2021-01-12 02:57
bloomI am still undecided about GL or Vulkan first.#asahi-gpu · 2021-01-12 02:57
bloom^^ that#asahi-gpu · 2021-01-12 01:27
bloommarcan: Turns out the framebuffer needs to be 128 byte aligned. I was previously aligning to 64 bytes only, and that's where the missing 64 bytes came from.#asahi-gpu · 2021-01-12 01:26
bloomOk, got the detiling wired in.#asahi-gpu · 2021-01-12 01:10
bloomBut you might be right :)#asahi-gpu · 2021-01-12 01:09
bloomicecream95: That's cheating :<#asahi-gpu · 2021-01-12 01:09
bloomjn__: Contrary to what you might think based on my code, I am not good at using computers. ^^#asahi-gpu · 2021-01-12 01:04
bloomjn__: ?#asahi-gpu · 2021-01-12 01:04
bloomDetiling should write linearly but read scattered, just as scalar code.#asahi-gpu · 2021-01-12 01:02
bloomicecream95: Yeah, probably :p#asahi-gpu · 2021-01-12 01:01
bloomI guess wiring in the dwetile#asahi-gpu · 2021-01-12 00:53
bloom(nothing is if I just read random github issues fo an hour..)#asahi-gpu · 2021-01-12 00:48
bloomAlright, let's see what I can get done in an hour :)#asahi-gpu · 2021-01-11 17:51
bloomdavidrysk[m]: Ah, thank you :)#asahi-gpu · 2021-01-11 17:30
bloom(there is some code I need to push to help with that, but my local setup is still a giant mess of shell scripts split acorross two machines)#asahi-gpu · 2021-01-11 17:29
bloomand wrap should be clang'able by just *.c -framework IOkit -dylib or something like that#asahi-gpu · 2021-01-11 17:29
bloombut it shows up in mem_0.bin if you dump with the wrap library there#asahi-gpu · 2021-01-11 17:29
bloom*hardcoding#asahi-gpu · 2021-01-11 17:29
bloomI was hardcoded offsets#asahi-gpu · 2021-01-11 17:29
bloomhttps://github.com/AsahiLinux/gpu/blob/main/wrap/wrap.c#L140#asahi-gpu · 2021-01-11 17:29
bloomdavidrysk[m]: 'documentation'? no, fraid not#asahi-gpu · 2021-01-11 04:09
bloom(the rockchip, mac is off)#asahi-gpu · 2021-01-11 04:08
bloomWhy is u64 load/stores more than 2x slower than u32 load/stores on a 64-bit machine?#asahi-gpu · 2021-01-11 04:08
bloomWat?#asahi-gpu · 2021-01-11 02:38
bloomOverloaded term :(#asahi-gpu · 2021-01-11 02:38
bloomtiling != tiling..#asahi-gpu · 2021-01-11 02:37
bloomMary_-: Sorry for forgetting to respond, but neat stuff! :) re: meallib#asahi-gpu · 2021-01-11 02:25
bloomkrbtgt: The specific tiling patterns vary, but everyone uses some form of tiling.#asahi-gpu · 2021-01-11 02:22
bloomkrbtgt: Everybody uses it as much as they can.#asahi-gpu · 2021-01-11 01:01
bloomHow could I forget?#asahi-gpu · 2021-01-11 00:36
bloomIf this is about your NEON routine, yes, you win, I'm over it, we both know you're better at writing tiling routines than me :P#asahi-gpu · 2021-01-11 00:35
bloomicecream95: Oh, hello there.#asahi-gpu · 2021-01-11 00:32
bloomI don't think that should come as a surprise to anyone, memory performance is critical to both CPU and GPU performance, and if you want to kick ass at both for an SoC, optimizing the hell out of the memory access is a 2 for 1 deal..#asahi-gpu · 2021-01-11 00:32
bloomThe optimization reducing ALU while keeping memory access constant. So it seems not only is the ALU on the Mac all around faster, the memory is _so_ much better.#asahi-gpu · 2021-01-11 00:31
bloomWhat's intriguing there is that the optimization helps only 4% on the Rockchip but 23% on the Apple.#asahi-gpu · 2021-01-11 00:30
bloomOh, and by properly splitting out unaligned/aligned portions, the M1 gets the job done in 0.33s while the RK3399 is at ~1.58s#asahi-gpu · 2021-01-10 23:59
bloomdavidrysk[m]: Every GPU does, that's not special :)#asahi-gpu · 2021-01-10 23:57
bloomNot sure how that compares to expected perf#asahi-gpu · 2021-01-10 23:57
bloomdavidrysk[m]: 🎉#asahi-gpu · 2021-01-10 23:56
bloomso just under 4x faster#asahi-gpu · 2021-01-10 23:56
bloomSingle-thread CPU/memory bound#asahi-gpu · 2021-01-10 23:56
bloomAlso, curiousity got the better of me: my texture tiling benchmark runs in 1.649s on Rockchip RK3399 (an arm64 processor from ~5 years ago) and 0.43s on the Apple M1#asahi-gpu · 2021-01-10 23:44
bloom^ The single best resource about texture tiling out there :)#asahi-gpu · 2021-01-10 23:44
bloomhttps://fgiesen.wordpress.com/2011/01/17/texture-tiling-and-swizzling/#asahi-gpu · 2021-01-10 23:44
bloomttps://fgiesen.wordpress.com/2011/01/17/texture-tiling-and-swizzling/#asahi-gpu · 2021-01-10 23:39
bloom(Kudos to marcan for figuring out the swizzle pattern.)#asahi-gpu · 2021-01-10 23:39
bloomBut it applies beautifully to M1!