01:32 <chadmed> Osira: the pi is absolutely useless without its downstream kernel and a bunch of proprietary patches in the raspi os repos. genuinely one of the worst computer related purchases i have ever made in my entire life

01:33 <chadmed> javascript absolutely tanks the the crappy little broadcom soc regardless too, so youtube will never work well even if you surrender to raspi foundation proprietary garbage

01:36 <psykose> i use my rpi4 with the generic alpine lts kernel

01:37 <i509vcb[m]> even when you consider that the big javascript engines and the jvm are VERY well optimized, the pi still can be brought to it's knees

01:37 <i509vcb[m]> I guess if you need a toaster the pi is a great candidate

01:37 <psykose> funnily, the other small arm socs are generally even worse and even harder to boot from what i've seen, with some exceptions

02:00 <chadmed> oh yeah but the 2711 is just such a thankless and evil thing. as i said a couple of days ago, hopefully these Macs demonstrating that non-amd64 workstations can be viable encourages other companies to start investing in exotic machines that actually work even half as well as the Macs do. im sick of all these arm and riscv boards being nigh on unusable because of quirks the devs have no intention of fixing

02:08 <psykose> i'm sure if you pay out the nose for the systemready platforms they are 'easy' to use

02:08 <psykose> (it's just altra and is all 10k+ pretty sure, lol)

02:08 <psykose> what a joke

02:26 <chadmed> well yea but then youre getting into big server build outs, not really practical for workstation use. a better example would be the raptor ppc64 workstations, but again theyre simply way too expensive and are overengineered with features 99.999% of people do not need in their workstation

02:26 <chadmed> but that said, they _lack_ features that people _do_ want, like a sensible number of usb ports and headers, m.2 slots, etc

02:27 <krbtgt> because they're server designs being contorted into workstation duty

02:28 <krbtgt> raptor did as good of a job they could papering over that, but that's fundamentally what they are

02:28 <krbtgt> whereas the apple stuff is designed specifically for desktop users

02:28 <chadmed> yeah exactly the problem, ive always maintained that a ppc64 or arm chip specifically designed for the enthusiast/consumer markets would absolutely go gangbusters and the macs seem to confirm this

02:29 <krbtgt> the macs don't sell because they're aarch64, they sell because they're damn good systems

02:29 <chadmed> i consider it the obverse of what has happened in the laptop market. apple releases no-bs laptops with nice screens, nice trackpads, nice keyboards and nice build quality and all the amd64 oems go "ah crap well i guess theyve got us beat theres literally nothing we can do to combat this"

02:29 <krbtgt> because they spent little to nothing on engineering

02:30 <chadmed> likewise, IBM and arm vendors see consumer amd64 workstations and go "ah crap they have usb 3 ports and overclocking theres literally nothing we can do to combat this" and never even bothered trying

02:30 <krbtgt> as much as i have beefs with apple they're one of the few OEMs that are willing to spend to solve a problem

02:30 <krbtgt> microsoft wants to do similar - surface seemed born out of frustration of how unambitious windows OEMs were

02:31 <krbtgt> IBM has no internest tin the consumer/workstation space and hasn't for years

02:31 <krbtgt> i think you're confusing cause/effect and why people buy systems

02:31 <chadmed> yeah, its ironic that many online communities will chastise apple for being 99% marketing and 1% product when it seems that, at least for the last few years, this has been more true in the PC world than anywhere else

02:32 <chadmed> yeah i know they have no interest in the market, but my point is that if they _did_ and produced POWERx chips that werent ten quintillion square millimetres and worked with board makers to produce PC-like consumer grade motherboards, they most likely would have been successful simply by being competition in the market

02:32 <krbtgt> the fact "lol, aarch64" was weird and quirky, but i bought my MBA because it was the fastest and best battery life of anything on the market.t apple's SoC enabled that but if it was amd64 and the same otherwise it wouldn't have changed a thing

02:33 <chadmed> no one seems too optimistic about intel's gpus in terms of performance or driver quality, but they will sell screamingly fast simply by virtue of being priced well in a duopolistic market

02:34 <i509vcb[m]> there is also the other end of the stick with IA64, but that probably has other variables associated with it also

02:34 <krbtgt> the problem is what market exists there for a low end power CPU? without someone to invest in ecosystem (a linux-only workstation is a niche audience) and everything else required it's a significant investment for little gain. and the last time IBM crippled a server chip for workstations, the end result was the final straw for apple

02:35 <krbtgt> ironically the only market for a low end power CPU for IBM i users, many of which barely use a single-socket server (and i suspect move a lot of the low-end servers) but are forced to buy power9s anyways (and most don't upgrade as a result)

02:35 <chadmed> IA64 was a rubbish fire mistake because it was a very tansparent attempt by intel to re-monopolise the CPU market with a closed off ISA that AMD couldnt get their hands on. at least with ppc and arm, the ISA is readily (though in arms case not freely) available for licensing by anyone who wants to make a chip

02:35 <krbtgt> itanium was HP's baby, not intel's

02:35 <krbtgt> no one complaining about itanium acttually has the real story about it

02:36 <krbtgt> i.e. the fact everyone else realized their own designs were dead end cost sinks - server and workstation CPU designs got more expensive to develop, they were maintaining bad system Vs they didn't need to anymore, etc.

02:37 <krbtgt> or the fact HP basically spent a decade buying VLIW firms

02:37 <krbtgt> i.e. multiflow

02:40 <krbtgt> circling back anyways the problem is unless you can beat epyc/xeon at some important metrics (or meet a specific niche) an arm/power workstation is a solution looking for a problem. apple's beat intel/amd on price/perf/power usage, but rcs didn't have that. they instead played to other niches (specific security reqs, I/O lanes until epyc beat them)

02:41 <krbtgt> unless you're a nerd going for CPU aesthettics, but that's a market that doesn't exist, i bet

02:42 <chadmed> its a larger market than youd think but youre right, probably not one any company could profitably sustan a platform's development for

02:42 <krbtgt> that's the thing, you have to play for basically ecosystem, and what you'd be lacking

02:43 <krbtgt> i.e. without windows a lot of fields won't buy, and you have to convince microsoft (sure, they did port windows to arm) and unambitious ISVs (which are other problem).

02:43 <chadmed> it would be nice to see just from a competition standpoint though. consumers seem to be kinda fed up with the windows/intel/amd ecosystem but because they are consumers they cant really articulate why. part of my hypothetical solution and pouring money into more open architectures/platforms so that chip vendors can compete with each other only on raw performance/efficiency and not rely upon guaranteed sales from being

02:43 <chadmed> the only one or two players in town

02:44 <chadmed> like, even if zhaoxin (the chinese-resurrected corpse of VIA) managed to produce a good amd64-compatible design, they would never _ever_ be sold into western markets

02:45 <krbtgt> the consilidation had market advantages - for years you could ship x86 binaries and have them just work. in the 90s workstation world you shipped multiple CDs with multiple binaries, and porting, supporting, and testing them all had a cost

02:46 <krbtgt> and of course, vendors don't want to spend jack shit on doing more than adapting reference designs

02:48 <chadmed> oh yeah i dont doubt this, it is very apparent that having multiple ISAs about the place is not sustainable, hence why MIPS and SPARC and Alpha all collapsed into irrelevance. my point is not this, rather that the industry standard ISA should not be something as closed off and inaccessible as x86/amd64. even ARM is not ideal, but at least anyone can buy an arm license and make a chip that _should_ be binary compatible

02:48 <chadmed> with any other ARM chip

02:53 <krbtgt> sure, i agree there

02:53 <krbtgt> i just don't think it'll be too likely because lack of OEM ambition

02:54 <krbtgt> though it does mean intel/amd will sit on the hoard until it turns to trash when the mandate of heaven is lost for them

03:24 <chadmed> pretty sure amd have an arm architecture license so they would probably be fine, and intel have been working with sifive on high performance riscv socs that hopefully will see the light of day in the next few years

06:34 <kode54> microsoft doesn't support windows on anything other than the vendor they've locked themselves into a relationship with, qualcomm

06:34 <kode54> at least in arm land

06:34 <kode54> their whole site even says it, not "arm" chips, "qualcomm chips"

06:35 <chadmed> if they ever realease windows arm publicly there will be third party drivers very, very quickly

10:35 <marcan> windows for arm will never be portable to M1 even if available without MS cooperation

10:36 <marcan> it assumes the system has a GIC and such, it needs core changes to the kernel to work

10:44 <j_ey> marcan: did you make any progress on the.. progress report?

10:51 <maz> marcan: even in a guest (which sees a GICv3), it dies very early on.

10:52 <j_ey> maz: I was going to ask about that, anything in particular to get a crash? I booted a vm with kvmtool and it booted fine

10:53 <maz> if I really cared, I'd have a look, but life is too short to care about windows.

10:53 <j_ey> oh hm, maybe I need to specify gicv3 w/ kvmtool?

10:54 <maz> j_ey: kvmtool should pick the right stuff by default, and in general KVM works just fine on M1, with the expected limitations (no PMU, stupidly small IPA space).

10:55 <j_ey> maz: specifically for https://lore.kernel.org/linux-arm-kernel/20210924082542.2766170-1-maz@kernel.org/T/#m672a275645075791c21ff6d9a2d838d729fb6f83 I meant

10:55 <maz> j_ey: you'll need that if you want to run EFI as shipped by Debian, for example.

10:55 <j_ey> or was that the bad thing that EDK2 did

10:55 <j_ey> yeah ok

10:56 <maz> if you don't have this patch, you'll get a nice SError and a panic.

10:56 <j_ey> cool, i will test that later.

12:06 <marcan> j_ey: I made *some* progress, but the past few days have been... interesting.

12:07 <marcan> (I had an MRI taken today. it came back clean, I'm okay.)

12:07 <j_ey> marcan: sounds scary

12:10 <phire> MRI is the one that makes loud banging sounds around you?

12:12 <chadmed> yeah the cool one

12:12 <chadmed> MRI is such an awesome modality, i love it

12:12 <chadmed> sorry i will contain my med sperging as it is off topic

12:41 <alyssa> phire: 🙉

12:50 <marcan> yeah, the sounds were crazy

12:59 <chadmed> surely they gave you earmuffs or some other form of hearing protection

13:23 <marcan> it's not *that* loud, and my ears were pressed into the head support anyway

13:23 <marcan> but it was a lot of cool sounds

13:26 <chadmed> interesting, its standard procedure here that we give patients ear protection when in the tube. the noises are really cool though its basically just a giant electromagnet switching at a frequency to get the water in different tissues to resonate with it and output an RF signal. therell be a bog standard high gain antenna somewhere in the room that picks this up and turns it into images

13:39 <alyssa> chadmed: can't tell if terrifying or awesome

13:39 <alyssa> maybe awesome in the etymologically faithful meaning

13:42 <chadmed> awesome if youre interested in it, terrifying if you have any ferromagnetic metals in or on your body :P

13:43 <chadmed> the field inside the tube can be multiple Teslas strong. most are ~1-1.5T

13:50 <alyssa> 🤖

15:35 <dottedmag> Actual voight-kampf test

17:57 <eta> loosely relevant question — is it possible to enter 1TR on M1 macs without knowing any administrator passwords? (the intended aim being to wipe the system clean)

17:57 <eta> (note: Activation Lock isn't enabled)

17:58 <alyssa> where does it ask for an admin pw?

18:01 <eta> alyssa, upon entering 1TR it seems

18:01 <eta> (note: I'm mainly reading documentation, don't have the device in front of me)

19:14 <j_ey> sent out v2 of pinctrl https://lore.kernel.org/linux-gpio/20211001191209.29988-1-joey.gouly@arm.com/T/#t

19:35 <alyssa> woop!

19:41 <povik> i look forward to trying it out

19:43 <ar> /40/40

23:12 <Method> stupid power outages keep screwing with znc