|
Shinku ABOOKEN posted:adding dependencies to rust projects is a one-line affair. why on earth would you want batteries-included stdlib? Shinku ABOOKEN posted:adding dependencies
|
#
?
Jul 19, 2019 22:36
|
|
|
|
| # ? May 31, 2024 17:38 |
|
|
As NPM as shown, having every piece of software depend on 5000 random packages by unknown people is a great idea.
|
|
#
?
Jul 19, 2019 22:39
|
|
|
guess what stdlib is a dependency
|
|
#
?
Jul 19, 2019 23:19
|
|
|
unrelated to stdlib chat, hey jawnv6 is there any particular reason you're aware of why ppc chips are and always have been so incredibly bad
|
|
#
?
Jul 19, 2019 23:48
|
|
|
the architecture is bad and ibm sucks a lot
|
|
#
?
Jul 19, 2019 23:51
|
|
|
can an architecture truly be bad if it doesn't have branch delay slots?
|
|
#
?
Jul 19, 2019 23:58
|
|
|
what's bad about the arch itself, it seems fairly unremarkable? the bank of condition registers?
|
|
#
?
Jul 20, 2019 00:00
|
|
|
PPC found its niche in this awkward "too energy-using for mobile", "not fast enough for PCs" market where it fit in snugly with "wall-power embedded" devices like set-top boxes and consoles, but x86 has come down a long way in price and ARM is somewhat efficient now to run a Switch. There are things I like about the ISA and things I don't. From a reverse engineering perspective, I love it. I'm sure as a compiler engineer it's quite terrible. And I haven't heard great things about the actual processors shipped using it.
|
|
#
?
Jul 20, 2019 00:41
|
|
|
don’t forget Motorola, they sucked a whole lot too and of course the third leg of the PowerPC triumvirate was early 1990s Apple, not exactly a well run company consider PowerPC’s origin story. it was very loosely Apple going “hey Motorola your 88000 RISC is kinda ok and we’ve even built prototype macs complete with working 88k MacOS, but we don’t trust you at all after your shameful failure to keep 68k at least even with x86 so we need a second partner. how’s about you get in bed with IBM and us and we’ll take almost all of the ISA from IBM POWER, not 88k, because we’re real tight with ibm right now. we’ll throw you the bone of PowerPC using the 88000 bus, since that’ll save us from having to redesign our RISC Mac chipsets from scratch. yeah you have to eat the loss of designing 88k, is that ok???” yeah that was doomed tbh ive never seen reason to believe PowerPC the ISA was fatally flawed. there were issues but you could’ve made good implementations of it. it’s just that the actual implementations got off to a rocky start due to the bad companies and politics involved, and they never recovered from getting dumpstered by 1990s Intel, which shocked the world by proving you could build a superscalar pipelined x86 that was pretty awesome despite the ugliness of the ISA (all hail Bob Colwell and his team)
|
|
#
?
Jul 20, 2019 01:02
|
|
|
whew, pressure's off since knowledgeable folks already chimed in i've never worked with ppc in depth. i know it has the best instruction mnemonic ever, predictable given that it Enforces In-order Execution of I/O EIEIO this is a good book on a particular implementation of Power: The Race for a New Game Machine the author's a self-aggrandizing rear end, but if you're willingly reading my posts it's probably not an issue. the gist is that Sony had IBM developing the Power7 for the PS3, MS comes sniffing around for a Xbox360 core, some desperate IBM exec leaks details about the sony chip and MS bites hard. the IBM team has to build the chips for both consoles in parallel and try really hard to not let on to Sony what's happening. the (bitter expat) gives a pretty bleak picture of upper IBM mgmt all the chip design stuff rang pretty true to my experience. especially the lead architect's opinion of the validation manager.
|
|
#
?
Jul 20, 2019 01:38
|
|
|
Progressive JPEG posted:sounds complicated, ill just paste this function i found on stack overflow odds are part of that stack overflow answer is gonna include "add the rust crate to your cargo.toml" or similar? ive only been playing around with rust off and on for a couple weeks so i dont know if their package management is trash but christ after all those years of managing c and c++ dependencies im easily impressed
|
|
#
?
Jul 20, 2019 01:59
|
|
|
the package management is actually quite good in rust i'm being a bit ridiculous if it was go or python then yeah you'd better hope that poo poo is in the stdlib
|
|
#
?
Jul 20, 2019 02:23
|
|
|
Suspicious Dish posted:PPC found its niche in this awkward "too energy-using for mobile", "not fast enough for PCs" market where it fit in snugly with "wall-power embedded" devices like set-top boxes and consoles, but x86 has come down a long way in price and ARM is somewhat efficient now to run a Switch. ppc is currently the fastest architecture available on the planet. out of the last 25 years, ppc has been the performance champion for at least 15 of them what the gently caress are you talking about with "not fast enough for PCs" (are you sure you are not confusing power, the architecture, with motorola, the vendor, who failed to deliver fuckall?)
|
|
#
?
Jul 20, 2019 04:13
|
|
|
right now, today, a ppc chip is roughly 1.5x to 2x as fast as the best intel x86 chips, depending on workload it will also suck down 4x the wattage per core, and cost you 5x to 20x as much money it's fuckloads faster but with a billion provisos ppc assembly sucks poo poo, but plainly ibm knows how to design some fuckin silicon
|
|
#
?
Jul 20, 2019 04:14
|
|
|
how much did big blue pay you for those posts i hope for their sake it wasn't much
|
|
#
?
Jul 20, 2019 04:19
|
|
|
JawnV6 posted:the author's a self-aggrandizing rear end, but if you're willingly reading my posts it's probably not an issue. the gist is that Sony had IBM developing the Power7 for the PS3, MS comes sniffing around for a Xbox360 core, some desperate IBM exec leaks details about the sony chip and MS bites hard. the IBM team has to build the chips for both consoles in parallel and try really hard to not let on to Sony what's happening. the (bitter expat) gives a pretty bleak picture of upper IBM mgmt either your summary is horrendously wrong or the book is horrendously stupid
|
|
#
?
Jul 20, 2019 04:22
|
|
|
Kazinsal posted:how much did big blue pay you for those posts big blue can eat poo poo the ibm sales process rivals oracle for horror their price list is a shitshow. you would have to be a moron to pay for linux on power. you would have to be blindfolded with a gun to your head to pay for aix on power. the only good thing i have to say about the whole loving thing is that power chips are very, very, very fast. (don't buy them)
|
|
#
?
Jul 20, 2019 04:23
|
|
|
as far as i can tell from spec.org power9 isn't very impressive tbqh it looks very good in terms of specintrate 2017 per "core", but that's because of ibm's extremely aggressive 8 threads per core thing. they don't seem to have submitted any specint results, so it's a fair guess that they'd look bad on that metric. so it's one of those machines where you're trying to make up for lackluster ST performance by running lots of threads, which is not great. and even then, intel is considerably ahead of power9 on 2-socket specintrate i don't doubt there are workloads where power9 is better (specintrate isn't the best thing in the world, ibm is also exotic in the sheer amount of cache they've got and that will help some things), but it is not the clear leader you're making it out to be
|
|
#
?
Jul 20, 2019 04:55
|
|
|
PRNG is stored in the platform
|
|
#
?
Jul 20, 2019 05:44
|
|
|
entropy is stored in the balls
|
|
#
?
Jul 20, 2019 05:49
|
|
|
BobHoward posted:but we don’t trust you at all after your shameful failure to keep 68k at least even with x86 this is not in any way true though, up to the 68040 they were even it was in the 486 and then seriously the Pentium where Intel really started to race ahead and even then, 88000 and PowerPC were pretty even at the same clock and then G3, G4, and G5 managed to be pretty substantial advances at their release, but IBM couldn’t reduce G5 to mobile and Motorola wouldn’t (because they were angry at IBM about G5 and getting progressively more focused on embedded anyway) so that catches you up to 2005 and Apple’s Intel transition quote:tbh ive never seen reason to believe PowerPC the ISA was fatally flawed that’s because it’s not, POWER is a perfectly fine ISA, though better implemented using the well-defined POWER runtime than the “let’s do everything PC-relative on a CPU that doesn’t have PC-relative addressing” Mach-O runtime now ROMP (IBM PC RT), that architecture is a shitshow it’s a second system RISC (successor to 801, predecessor to POWER) except it doesn’t have uniform instruction size, allows unaligned access, and defines things like I/O channels in the CPU architecture (I’m writing an emulator right now, like I literally have it open in Swift Playgrounds on my iPad Pro at the moment, and while it’s interesting and nobody else seems to have done it, I’m not pretending I’m super eager to run AIX or AOS…)
|
|
#
?
Jul 20, 2019 05:54
|
|
|
Notorious b.s.d. posted:ppc is currently the fastest architecture available on the planet lmao
|
|
#
?
Jul 20, 2019 06:12
|
|
|
assuming you're talking about IBM POWER processors (which, in an extremely-IBM fit of naming, do not run "POWER" nor "PowerPC" but "Power". Yes, IBM thinks of those as three separate ISAs. But the processors themselves are "POWER", not "Power") uh, "fast". Sure bud.
|
|
#
?
Jul 20, 2019 06:16
|
|
|
Volte posted:this is the reason to expand on this, it's not that adding dependencies on its own is bad. it's when you've got an ecosystem and a culture that always relies on it to the extent where a bunch of modules are really just stitching together a couple dependencies, and those dependencies are actually just pretty simple module stitching together a couple of dependencies with some new semantics, and before you know it you've gone from "oh i'll add a little module to interact with github's api so i don't have to remember the auth details" to ok, well that added 5 deps on its own transitively, then i wanted to interact with this other thing, that was another 5... and you've got hundreds or thousands of random deps many of which you may be completely unaware of because they're dozens of levels down the dependency tree and half of them are doing insanely basic things like typechecks or basic numeric operations. a shitload of that can be nipped in the bud with even a basic standard library, and a bunch more with a batteries-included standard library. now, things like python and the c++ stl have taught us that tying standard library versions to language versions (as in, shipping the one with the other) probably isn't smart, but as much as people are decrying it having an "extras" or "batteries" or "stuff" crate that's developed by the language ecosystem maintainers would legitimately be a great complement to the language
|
|
#
?
Jul 20, 2019 06:30
|
|
|
eschaton posted:this is not in any way true though, up to the 68040 they were even i might be misremembering the sequence of events since i was a teen at the time, but i remember the 486 being widely available well before the 040 agreed that the lack of an immediate answer to the pentium was the nail in the coffin. motorola did eventually ship the 060, but it wasn't great
|
|
#
?
Jul 20, 2019 06:35
|
|
|
BobHoward posted:as far as i can tell from spec.org power9 isn't very impressive tbqh specint blows and lol @ ibm's idiot idea about MOAR THREADS the reality is on actually-existing java and c++ workloads in my actual job, POWER9 demonstrates significantly faster per-thread performance and i give zero fucks because lol @ the wattage and cost holy poo poo how crazy would you have to be to spend literally 20x as much money to get 1.5x the single thread performance
|
|
#
?
Jul 20, 2019 06:49
|
|
|
BobHoward posted:i might be misremembering the sequence of events since i was a teen at the time, but i remember the 486 being widely available well before the 040 your experience is crap. the 80486 was made available in 1989 what you are remembering is that the 80386 and 80486 became hella cheap in the 1990s in the 1980s, the 80386 and 80486 were crazy expensive chips that only showed up in weird systems BobHoward posted:agreed that the lack of an immediate answer to the pentium was the nail in the coffin. motorola did eventually ship the 060, but it wasn't great the pentium competed with high-end risc designs by that time the 68k series was long dead, even by motorola standards. the only remaining uses were embedded. by the time the pentium shipped, the motorola 88k had already shipped, failed to find a market, and been discontinued
|
|
#
?
Jul 20, 2019 06:50
|
|
|
Suspicious Dish posted:assuming you're talking about IBM POWER processors (which, in an extremely-IBM fit of naming, do not run "POWER" nor "PowerPC" but "Power". Yes, IBM thinks of those as three separate ISAs. But the processors themselves are "POWER", not "Power") ibm has shipped three or even five separate "ISAs" that are all closely related, with three or more processor families, for the last thirty years for the last 25 years, things branded "RS64" or "POWER" have been speed champs for 15. not contiguously ibm is very good at designing processors. whether you want to actually pay ibm's prices for them chips, that's a separate question i would not encourage anyone to buy software or hardware from ibm but it's fuckin stupid to pretend ibm isn't faster than x86
|
|
#
?
Jul 20, 2019 06:55
|
|
|
Notorious b.s.d. posted:holy poo poo how crazy would you have to be to spend literally 20x as much money to get 1.5x the single thread performance not very crazy at all depending on what you're up to. single-thread performance is one of those actually durable advantages, as there is no substitute whatsoever for it in some tasks.
|
|
#
?
Jul 20, 2019 07:58
|
|
|
Cybernetic Vermin posted:not very crazy at all depending on what you're up to. single-thread performance is one of those actually durable advantages, as there is no substitute whatsoever for it in some tasks. i know in abstract those workloads exist but they scare me and I’m glad it doesn’t fall on my rear end to justify budget for POWER chips
|
|
#
?
Jul 20, 2019 09:16
|
|
|
Notorious b.s.d. posted:
this is something bankers and investment bankers are into so the scum of the earth basically
|
|
#
?
Jul 20, 2019 09:47
|
|
|
Suspicious Dish posted:the only thing that made mersenne twister get adopted was the cool name. it's otherwise an extraordinarily mediocre prng, and was at the time. It also has amazingly, mind bogglingly long repetition period, which nobody actually needs.
|
|
#
?
Jul 20, 2019 16:00
|
|
|
Suspicious Dish posted:There are things I like about the ISA and things I don't. From a reverse engineering perspective, I love it. I'm sure as a compiler engineer it's quite terrible. And I haven't heard great things about the actual processors shipped using it. Can confirm that Power CPUs are Actually Fast. They're no SPARC or Itanium, they've actually kept up. I don't think they're particularly annoying to target with a compiler either, try Thumb for that.
|
|
#
?
Jul 20, 2019 16:34
|
|
|
Boiled Water posted:this is something bankers and investment bankers are into ah, professional-grade tools for picking up nickels in front of a bulldozer, huh?
|
|
#
?
Jul 20, 2019 17:35
|
|
|
well, the operation of the bulldozer is also a case (e.g. generating all implieds in derivatives markets is usually np-complete, but the best possible effort should be put forward as failing to generate some of the volumes, or failing to do it quickly, leaves pennies around for the vultures to pick up)
|
|
#
?
Jul 20, 2019 17:57
|
|
|
rjmccall posted:ibm sucks a lot shocking if true
|
|
#
?
Jul 20, 2019 17:59
|
|
|
i'd imagine you could get quite a lot of single-thread performance if you make a die the size of a coaster, fill it with cache, then decide you don't give a poo poo about TDP whatsoever but yeah your yields will be dogshit and so will your price intel at least recognizes that most server farms care a lot more about performance per watt than they do about performance
|
|
#
?
Jul 20, 2019 18:13
|
|
|
is ppc libre like speech?
|
|
#
?
Jul 20, 2019 18:57
|
|
|
is spirv real
|
|
#
?
Jul 20, 2019 19:48
|
|
|
|
| # ? May 31, 2024 17:38 |
|
|
also cargo is great but compiling takes so drat long >:?
|
|
#
?
Jul 20, 2019 19:48
|
|