2021-10-22
00:07FireFox317 has joined #asahi
00:46yuyichao has joined #asahi
01:30riker77_ has joined #asahi
01:35riker77 has quit [Ping timeout: 480 seconds]
01:35riker77_ is now known as riker77
01:49rohin2 has joined #asahi
01:51bgb_ has quit [Ping timeout: 480 seconds]
02:00bgb_ has joined #asahi
02:09bgb_ has quit [Ping timeout: 480 seconds]
02:15PhilippvK has joined #asahi
02:18bgb_ has joined #asahi
02:18phiologe has quit [Ping timeout: 480 seconds]
02:19chadmedcitizen1[m]: its not really quite as simple as that. the graphics portion of the M1 Max is like ~30-40 billion transistors, which is significaantly more than the GA102 die used in the RTX3090 even. Apples advantage with these chips isnt really perf/mm^2 or even raw perf in general, but specifically in perf/W
02:21chadmedconsider that this enormous graphics SIP pulls down something like 50W at full tilt, vs a GA102 card (RTX 3090, 3080Ti) which can suck down upwards of 300W. thats the impressive part, not so much the performance by itself. the M1 Max chip is enormous, almost from the nvidia and intel school of thought where you just throw transistors at a problem until it goes away
02:24chadmedadditionally, RISC is not magic. AArch64 starting with ARMv8 is just a very very well designed series of RISC ISAs, and at this point theres easily been trillions of dollars of combined R&D effort over the last ~25 years into making ARM the most efficient architecture in the industry.
02:26bgb_ has quit [Ping timeout: 480 seconds]
02:26chadmedas it currently stands, the major advantage these well designed ARMv8 chips have over, say, an AMD64 chip, is that the fixed instruction length makes it very easy to fetch and decode multiple instructions at once. even with modern AMD64 implementations (which are RISC-like at the hardware level) the software-facing ISA makes doing this extremely difficult
02:28maennich_ has joined #asahi
02:29maennich_ is now known as maennich1
02:29steev_ has joined #asahi
02:30steev has quit [Ping timeout: 480 seconds]
02:30steev_ is now known as steev
02:31maennich has quit [Ping timeout: 480 seconds]
02:31maennich1 is now known as maennich
02:32eichin_ has joined #asahi
02:32eichin has quit [Ping timeout: 480 seconds]
02:35bgb_ has joined #asahi
02:38chadmedphire: well yeah i think the meaning of "RISC" in 2021 has kinda shifted to just "load/store distant descendant of berkley risc" rather than be any indication of how many instructions the set has
02:38chadmedthe latest version of the power isa manual is like 3000 pages lmao, they still call it a risc afaik
02:40chadmedi guess if we consider the traditional meaning of the term, the last "real" RISC before RISC-V was probably SPARCv9, for which i have the manual
02:40phireI disagree. I just think people have kept using the terms RISC/CISC out of habbit and applying them to processors that might have started off as RISC/CISC but arent' anymore
02:41phireignoreing the ISAs, modern x86 cores and modern aarch64 cores look very simular to each-other
02:41chadmedwell in that case given the convergence of all designs on load/store and the increasing complexity of all instruction sets, do the terms really mean anything in everyday parlance anymore?
02:42chadmedlike im not even sure if ibm z/architecture is a "real" CISC anymore or if they went the uop route like amd and intel
02:42i509vcb[m]No one can agree what number at which an architecture moves from risc to cisc
02:43i509vcb[m]Is the original 8086 ISA with 81 ish instructions CISC while modern armv8 can get to 1300 instructions in a fulled fitted out chips being RISC?
02:43bgb_ has quit [Ping timeout: 480 seconds]
02:43i509vcb[m]s/fulled/fully/, s/fitted/kitted/
02:44chadmedwell yeah im actually with phire on this one, the terms are utterly meaningless in 2021 and people only really use them because we havent invented new ones to distinguish designs (because there really isnt much to distinguish them anymore)
02:44phireand it's very simple really. Any ISA designed before 1985 is CISC and every ISA designed after 1985 claims to be RISC
02:45phirebecause RISC was the trendy buzzword and all the CPU designers went "ofcourse my ISA is RISC"
02:45chadmedi think people just use CISC as a pejorative to express their dislike of amd64, but very few people can actually exlpain *why* amd64 is crap
02:45chadmedand to be fair to them, its not an easy thing to explain. the problems with it are multifaceted and extremely complex and so its easier for people on the left hand side of the dunning-kruger graph to just say "oh yeah its cisc so its bad lol"
02:46phireIt doesn't help that the term RISC got closely assocated with modern cpu ideas like "Pipelined" and "Superscalar"
02:47phiresure, most of the early pipelined and superscaler cpus were RISC. and that was the point. But mostly because simpling the ISA and getting the instruction count down saved transistors and made pipelined/superscaler designed viable a few years before they would be viable for older ISAs
02:48phirepersonally, I think the original ARMv1 itself has a dubous claim to being RISC. I think the ISA design is closer to VLIW
02:50phirewith the ablitly to pack a conditional, a barrel shift and an arithmetic/logic operation into a single instruction
02:50bgb_ has joined #asahi
02:52chadmedive tried multiple times to study VLIW and every time it kinda just makes my brain shut off
02:56phireIt's like out-of-order, with multiple functional units running each cycle. Excpet instead of the CPU's scheduler working out the dependencies dynamically, the ISA has some way of packing multiple sub-instructions into a single combined instruciton
02:57chadmedyeah thats about as far as my understanding goes lmao, i cant wrap my head around _how_ it does that effectively
03:00phireThe designs vary. At the simpler end you have design's like the raspberry pi's QPUs (shader coers). 64 bit wide instrucions (hence Very Long Instruction Word) and there are fixed feilds in the instruction.
03:01phirethere is a 4 bit flag field (this instruction triggers a immedate load, or this instruction ends), there is are felds for selecting two registers from the big reigster files
03:02phireAt the narrower end, it's basically fixed fields again, except you have multiple instruction formats to select between
03:02riker77 has quit [Quit: Quitting IRC - gone for good...]
03:02riker77 has joined #asahi
03:06phirethey aren't normally public ISAs, beacasue they only really work when paired with a good scheduling compiler
03:06chadmedyeah the last good one was terascale i think, and all radeon cards have hardware scheduling
03:07phireit turns out that hardware scheduling is just unreasonably effective at producing good preformance
03:09chadmedcrazy how windows has only just recently received support for addressing hardware schedulers in GPUs, and only through dx12u
03:17marvin24 has joined #asahi
03:20marvin24_ has quit [Ping timeout: 480 seconds]
03:25bgb has joined #asahi
03:28Gue___________________________ has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
03:31bgb_ has quit [Ping timeout: 480 seconds]
03:34kov has quit [Quit: Coyote finally caught me]
03:53eichin_ has quit []
03:53eichin has joined #asahi
03:55Gue___________________________ has joined #asahi
04:16rohin2 has quit [Ping timeout: 480 seconds]
04:18sorear"no, I think the terms RISC/CISC should mostly stay in the 90s where it belongs." very this
04:19dlss^ has joined #asahi
04:19chadmedi think i may have already mentioned this here a while ago but the progress of hardware design shares many similarities with convergent evolution in biology
04:21chadmedsuch that there is no longer any meaning to terms like risc and cisc because the canonical way to design a cpu core has converged so much that architectures that we once called risc and cisc now look virtually identical at the hardware level
04:52rohin has quit [Quit: Konversation terminated!]
05:16phireIt's not like cpu designs ever diverged that much (along the RISC/CISC split). It was more about the ISA you would design from scratch to go with that ISA
05:17phireAnd over time, the size of that extra hardware realtive to the rest of the core has shrunk
05:22phireI think you can argue that the legacy restrictions actually helped x86 core designs in the long term.
05:23phireIt encouraged Intel and AMD to go for Out-of-order designs very early compared to most of their RISC competitors
05:25jeffmiw has quit [Ping timeout: 480 seconds]
05:33chadmedif itanium came out when it was meant to maybe it would have been competitive. dont get me wrong, im thankful it didnt but a part of me wonders what wouldve happened to the PC had it succeeded as intel and hp planned. would people have held themselves captive to intel and hp and continued to use the platform?
05:33chadmedthey hold themselves captive to intel and amd now, but thats only because until recently There [Was] No Alternative(tm)
05:38maor26 has joined #asahi
05:39phireI'm really not sure how serious the plan for itanium on the desktop ever was. It's not like Intel paused or slowed down development on x86
05:41nobodynada has quit [Quit: leaving]
05:42Hotswap has quit [Ping timeout: 480 seconds]
05:55bgb_ has joined #asahi
05:55marcansure, you can µop cache your way around that for tight loops, but it doesn't work so well with general code
06:00bgb has quit [Ping timeout: 480 seconds]
06:02bgb has joined #asahi
06:08bgb_ has quit [Ping timeout: 480 seconds]
06:23chadmedyeah we were saying this before. dunno what sort of deal with satan intel did to get to 5-way decode for alder lake
06:34Hotswap has joined #asahi
06:37phireGracemont (the little core) is more intersting. Instead of trying to do a big massive 6-way decode, it has two 3-way decoders.
06:38phirethen they cache Instruction-length-decode infomation in the L1i cache, and the two decoders can actually operate in an interleaved mode
06:39phireFor fresh code after a cache-miss, you only get upto 3 instructions per cycle, but the second time around you get upto 6 instructions per cycle
06:54aleasto has joined #asahi
06:56aleasto has quit [Remote host closed the connection]
06:56aleasto has joined #asahi
06:57aleasto has quit [Remote host closed the connection]
06:59aleasto has joined #asahi
07:34dlss^ has quit [Ping timeout: 480 seconds]
08:25Gue___________________________ has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
08:37bps has joined #asahi
08:47off^ has joined #asahi
10:10riker77 has quit [Quit: Quitting IRC - gone for good...]
10:10riker77 has joined #asahi
10:39X-Scale` has joined #asahi
10:45X-Scale has quit [Ping timeout: 480 seconds]
10:55aleasto has quit [Remote host closed the connection]
10:57aleasto has joined #asahi
11:08hspak8 has joined #asahi
11:08hspak has quit [Read error: Connection reset by peer]
11:08hspak8 is now known as hspak
11:08m42uko_ is now known as m42uko
11:09X-Scale has joined #asahi
11:13gabuscus_ has quit [Read error: Connection reset by peer]
11:13X-Scale` has quit [Ping timeout: 480 seconds]
12:01gabuscus has joined #asahi
12:59jbowen has joined #asahi
13:08darkapex1 has quit [Ping timeout: 480 seconds]
13:16nsklaus has joined #asahi
13:19nsklaus_ has joined #asahi
13:22nskl has quit [Ping timeout: 480 seconds]
13:22aleasto has quit [Remote host closed the connection]
13:24nsklaus has quit [Ping timeout: 480 seconds]
13:27chadmed has quit [Quit: Konversation terminated!]
13:28chadmed has joined #asahi
13:30aleasto has joined #asahi
13:49aleasto has quit [Remote host closed the connection]
13:51aleasto has joined #asahi
14:00yuyichao has quit [Ping timeout: 480 seconds]
14:27yuyichao has joined #asahi
14:49kenjigashu has joined #asahi
15:31kenjigashu has quit []
16:05kenjigashu has joined #asahi
16:05kenjigashu has quit [Remote host closed the connection]
16:12bps has quit [Remote host closed the connection]
16:12bps has joined #asahi
16:13loop0 has joined #asahi
16:17bps has quit [Remote host closed the connection]
16:22aleasto has quit [Quit: Konversation terminated!]
16:22aleasto has joined #asahi
16:23aleasto has quit [Remote host closed the connection]
16:24aleasto has joined #asahi
16:28jbowen_ has joined #asahi
16:28jbowen has quit [Read error: Connection reset by peer]
16:36Gues__________________________ has joined #asahi
16:57___nick___ has joined #asahi
16:57___nick___ has quit []
16:58___nick___ has joined #asahi
16:59loop0 has quit [Quit: Leaving.]
17:02dff has quit [Quit: WeeChat 3.1]
17:03citizen1[m]<chadmed> "additionally, RISC is not magic...." <- whatever it is, it's a flat single chip that is doing both cpu+gpu work silently with major power saves. Whatever apple is doing everyone else should do unless there is a point where x86 intel chips can or perform them . i imagine what happens if you have a dedicated gpu running m1 like processors
17:11jmr2 has joined #asahi
17:13jmr2Question... I'm trying to add j_ey's version of Corellium's keyboard driver to alyssa's tree. I believe that I have the code right but the DT wrong.
17:13jmr2Driver is loaded, visible in /sys/bus/spi/drivers/applespi, device is visible at /sys/bus/platform/devices/23510c000.spi. No symlinks between those.
17:13jmr2The compatible strings match between the dtsi and spi-apple-mc.c - but that driver's probe function isn't called.
17:13jmr2Only reference to SPI in dmesg is "SPI driver applespi has no spi_device_id for input,applespi-kbd-v1"
17:16jmr2That's where I started from, but alyssa's tree has conflicting definitions for clocks and pinctrl/gpio, so I have to merge.
17:17jbowen_ has quit []
17:20loop0 has joined #asahi
17:24loop0 has quit []
17:47jmr2So this one makes more sense as my current best merge of your 8b44149d: http://paste.debian.net/plainh/bcfb9c39
17:48jmr2But I now see that you've moved the clocks into pmgr0 (clock-controller), while in alyssa's branch, 23b700000 is still power-management.
17:48jmr2Here's what I'm merging into: https://github.com/mu-one/linux/blob/20211015/arch/arm64/boot/dts/apple/t8103.dtsi
17:49rohin has joined #asahi
17:50kenjigashu has joined #asahi
17:50kenjigashu has quit [Remote host closed the connection]
17:55Fanfwe has quit [Quit: ZNC 1.8.2+deb2+b1 - https://znc.in]
17:57Fanfwe has joined #asahi
18:02kenjigashu has joined #asahi
18:10Fanfwe has quit [Quit: ZNC 1.8.2+deb2+b1 - https://znc.in]
18:10kenjigashu has quit []
18:12Fanfwe has joined #asahi
18:26kenjigashu has joined #asahi
18:27Raito_Bezarius has quit [Ping timeout: 480 seconds]
18:29bps has joined #asahi
18:31jmr2OK. Do you have a link for the full version of the dtsi? Tip or your main (or m1-keyboard) branch?
18:31Gues__________________________ has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
18:44jmr2 has quit [Remote host closed the connection]
18:50kenjigashu has quit []
18:57aleasto has quit [Remote host closed the connection]
19:08Gues__________________________ has joined #asahi
19:18alexstore06 has joined #asahi
19:29kenjigashu has joined #asahi
19:29kenjigashu has quit [Remote host closed the connection]
19:32jeffmiw has joined #asahi
19:34kenjigashu has joined #asahi
19:49Raito_Bezarius has joined #asahi
19:49Raito_Bezarius has quit [Remote host closed the connection]
19:50Raito_Bezarius has joined #asahi
19:52rohin has quit [Quit: Konversation terminated!]
19:54alexstore06 has quit [Ping timeout: 480 seconds]
20:02c10l has quit [Quit: Ping timeout (120 seconds)]
20:03c10l has joined #asahi
20:03Gues__________________________ has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
20:08___nick___ has quit [Ping timeout: 480 seconds]
20:08c10l has quit [Quit: Ping timeout (120 seconds)]
20:10c10l has joined #asahi
20:14Gues__________________________ has joined #asahi
20:25stzsch has quit [Read error: Connection reset by peer]
20:26stzsch has joined #asahi
20:31alexsto__ has joined #asahi
20:33alexsto__ has quit [Quit: Leaving...]
20:33alexstore06 has joined #asahi
20:51Gues__________________________ has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
20:55off^ has quit [Ping timeout: 480 seconds]
20:56kenjigashu has quit []
21:04kenjigashu has joined #asahi
21:05kenjigashu has quit [Remote host closed the connection]
21:25Gues__________________________ has joined #asahi
21:41Gues__________________________ has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
21:51Gues__________________________ has joined #asahi
22:20rohin2 has joined #asahi
22:23yuyichao has quit [Ping timeout: 480 seconds]
22:55tertu has quit [Quit: so long...]
22:55tertu has joined #asahi
22:56X-Scale` has joined #asahi
23:00X-Scale has quit [Ping timeout: 480 seconds]
23:15Gues__________________________ has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
23:19rohin2 has quit [Ping timeout: 480 seconds]
23:38maor26 has quit [Ping timeout: 480 seconds]