#
✕ nick:phire
#asahi · 2022-02-01 23:33
phirecurrently I'm running linux in a parallels VM on my macbook. Because it's very hard to find good ARM systems to test on
#asahi · 2022-02-01 23:31
phirelinux only
#asahi · 2022-02-01 23:31
phirehttps://github.com/FEX-Emu/FEX
#asahi · 2022-02-01 23:30
phirethat we have been working on
#asahi · 2022-02-01 23:30
phireits' a userspace x86_64 emulator
#asahi · 2022-02-01 23:30
phirewell... steam is working under FEX
#asahi · 2022-02-01 23:28
phireI haven't used that distro for over a decade
#asahi · 2022-02-01 23:28
phireI didn't know there was an ARM version of slackware
#asahi · 2022-02-01 23:22
phirewhat does the backlight get set to? some default, or whatever macos last set it to?
#asahi · 2022-02-01 09:15
phirethat's not a blocker :P That's just a big time investment
#asahi · 2022-02-01 09:14
phireare their any major blockers that must be done before work on an AGX kernel driver can start, or could I theortically start hacking one together if I have free time?
#asahi · 2022-02-01 09:13
phireUsing virgil in parallels for 3d accleration is a bit frustrating, so few games work
#asahi · 2021-11-01 23:06
phire"<marcan> oh wow, the antialiased edges are done in hardware" I'm curious if that's done in the LCD controller, or if it's a physical property of the panel itself.
#asahi · 2021-10-22 06:40
phireApparnetly the two decoders can even operate out of order, whatever that means
#asahi · 2021-10-22 06:39
phireand persumably two complex instructions per cycle?
#asahi · 2021-10-22 06:39
phireFor 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
#asahi · 2021-10-22 06:38
phirethen they cache Instruction-length-decode infomation in the L1i cache, and the two decoders can actually operate in an interleaved mode
#asahi · 2021-10-22 06:37
phireGracemont (the little core) is more intersting. Instead of trying to do a big massive 6-way decode, it has two 3-way decoders.
#asahi · 2021-10-22 06:33
phireit's actually 6-way
#asahi · 2021-10-22 05:40
phireThe P5, P6 and netburst were all developed in parallel
#asahi · 2021-10-22 05:39
phireI'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
#asahi · 2021-10-22 05:30
phireWas still a better design than Itanic that was Itanium
#asahi · 2021-10-22 05:23
phireIt encouraged Intel and AMD to go for Out-of-order designs very early compared to most of their RISC competitors
#asahi · 2021-10-22 05:22
phireI think you can argue that the legacy restrictions actually helped x86 core designs in the long term.
#asahi · 2021-10-22 05:17
phireAnd over time, the size of that extra hardware realtive to the rest of the core has shrunk
#asahi · 2021-10-22 05:17
phireThe cpu designs with legacy ISAs just got suck with extra hardware to make it work
#asahi · 2021-10-22 05:16
phireIt'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
#asahi · 2021-10-22 03:09
phirediffrent scheduler
#asahi · 2021-10-22 03:07
phireit turns out that hardware scheduling is just unreasonably effective at producing good preformance
#asahi · 2021-10-22 03:06
phirethey aren't normally public ISAs, beacasue they only really work when paired with a good scheduling compiler
#asahi · 2021-10-22 03:05
phirethe "real" VLIW ISAs tend to not get documented so well
#asahi · 2021-10-22 03:04
phireum, ARM32 is a good example of that
#asahi · 2021-10-22 03:02
phireAt the narrower end, it's basically fixed fields again, except you have multiple instruction formats to select between
#asahi · 2021-10-22 03:01
phireand felids for controlling the two ALUs
#asahi · 2021-10-22 03:01
phirethere 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
#asahi · 2021-10-22 03:00
phireThe 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.
#asahi · 2021-10-22 02:56
phireIt'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
#asahi · 2021-10-22 02:50
phirewith the ablitly to pack a conditional, a barrel shift and an arithmetic/logic operation into a single instruction
#asahi · 2021-10-22 02:48
phirepersonally, I think the original ARMv1 itself has a dubous claim to being RISC. I think the ISA design is closer to VLIW
#asahi · 2021-10-22 02:47
phiresure, 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
#asahi · 2021-10-22 02:46
phireIt doesn't help that the term RISC got closely assocated with modern cpu ideas like "Pipelined" and "Superscalar"
#asahi · 2021-10-22 02:45
phirebecause RISC was the trendy buzzword and all the CPU designers went "ofcourse my ISA is RISC"
#asahi · 2021-10-22 02:44
phireand it's very simple really. Any ISA designed before 1985 is CISC and every ISA designed after 1985 claims to be RISC
#asahi · 2021-10-22 02:43
phiremuch of the RISC vs CISC stuff is a marketing debate that happened in the 90s
#asahi · 2021-10-22 02:42
phireno, I think the terms RISC/CISC should mostly stay in the 90s where it belongs.
#asahi · 2021-10-22 02:41
phirethey are all very wide, out-of-order designs
#asahi · 2021-10-22 02:41
phireignoreing the ISAs, modern x86 cores and modern aarch64 cores look very simular to each-other
#asahi · 2021-10-22 02:40
phireI 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
#asahi · 2021-10-22 02:36
phirethat happens to have both a load-store archtecture and fixed-width instructions
#asahi · 2021-10-22 02:35
phireI personally don't think ARMv8 counts as RISC. It's just a well-designed modern ISA

plus ›