tmsr - Search: amd

diana_coman: and apparently the bits and pieces known/sorted about this are not that easy to find or something, hmmm; esp the paths issue was discussed and sorted in the logs + documented at http://ossasepia.com/2019/07/14/a-working-cuntoo-install-on-amd-fx-8350-with-script/

BingoBoingo: And the four cores in ye olde Qntra/DNS box are still AMD. Just... now I got more drives to throw in it.

diana_coman: not sure how the "won't pay for shipping those all the way from uruguay" directly translates into no value on amd processors; not like I don't have around me mainly amd processors for that matter.

BingoBoingo: I see value in AMD, but with portability at a premium... I'm only interested in lugging so many cores around.

mircea_popescu: apparently i'm the only other party seeing value in amd processors, what can i say.

mircea_popescu: in which case, should get a 8+ core amd of some kind, 32+gb and a coupla disks, raidless ok, and then review for upgrade in what, a few months ?

ossabot: Logged on 2018-02-05 20:57:25 mircea_popescu: BingoBoingo phf trinque diana_coman so i'm thinking of basically getting a dozen amd fx-8350s (because just 125w compared to say fx-9590s 220) racked in u2 mbs (say maybe ASRock 970M PRO3 ? i thinkl that fits in 2u neh ?) with 64 gb ram each and a coupla ssds ? any complaints/suggestions ?

diana_coman: http://logs.ossasepia.com/log/trilema/2019-11-06#1949915 - preference is for AMD arch as easiest & more familiar to deploy on & work with; that being said, if no AMD available, will go with Intel; for concrete specs, the current server is rather undemanding and it ran fine on fx-8350 eg something like this should work absolutely fine.

diana_coman: to fully clarify: I can boot both/either servers on dulap-gentoo on either amd or intel hardware

diana_coman: mircea_popescu: that is AMD with dulap-gentoo aka proto-cuntoo aka NOT fully statically linked

ossabot: Logged on 2019-10-14 14:18:25 diana_coman: http://logs.ossasepia.com/log/trilema/2019-10-13#1945106 - got around to it and can confirm I got it working on Intel too (with minimal changes of kernel config ie selecting Intel rather than AMD flags)

feedbot: http://ossasepia.com/2019/07/14/a-working-cuntoo-install-on-amd-fx-8350-with-script/ << Ossa Sepia -- A Working Cuntoo Install on AMD FX-8350 (with script)

asciilifeform: fanless, 2-core 'amd g-series' box , vendor publishes 100% electric schematic . runs 100% blobless.

diana_coman: http://logs.ossasepia.com/log/trilema/2019-10-13#1945106 - got around to it and can confirm I got it working on Intel too (with minimal changes of kernel config ie selecting Intel rather than AMD flags)

ossabot: (ossasepia) 2019-10-11 diana_coman: wtf can't quite find the thing; the motherboard is Asus M5A78L-M PLUS/USB3 HDMI Motherboard; the cpu is AMD FX-8350 Eight Core to 4.2GHz

asciilifeform: the item strictly-as-given is expected to function on any 64bit amd ( asciilifeform's test bed was a 'e350', circa 2010 . )

asciilifeform: diana_coman: should need adjustment only in kernel flags : no amd-exclusive cpu instrs were used in building the binaries (given the vintage of the cpu targeted, none exist such that would fall into this trap)

asciilifeform: ^ mainly for diana_coman . physically tested on a 'e350' amd, excavated for the occasion. req'd 0 changes to kernel config !!

diana_coman: mp_en_viaje: I'll run it, as promised on all amd around, at the very least; will at least have *something* to say after that, I guess.

diana_coman: now ofc how the fuck to "test to exhaustion" here, not like I have the full range of amd and I'm not living in a dc either, ugh.

diana_coman: asciilifeform: if I got this straight, you say the "specific" here is simply "any amd" , right?

ericbot: Logged on 2019-10-10 14:34:11 mp_en_viaje: UY1: » 128 GB ECC RAM » 2x 16-core AMD Opteron 6376 (2.3 GHz) (32 Cores) » Reserved Space, Just for you on a 1.7 TB RAID 10 Array composed of Samsung Solid State Drives » Apache, MySQL, PHP 5.6 » Located in safe and politically stable Montevideo, Uruguay outside of NATO or other belligerent states

mp_en_viaje: UY1: » 128 GB ECC RAM » 2x 16-core AMD Opteron 6376 (2.3 GHz) (32 Cores) » Reserved Space, Just for you on a 1.7 TB RAID 10 Array composed of Samsung Solid State Drives » Apache, MySQL, PHP 5.6 » Located in safe and politically stable Montevideo, Uruguay outside of NATO or other belligerent states

BingoBoingo: asciilifeform: All in it came under 500 USD. One of the AMD mobile processors on ITX board in iirc 10-ish inch deep rackmount case

asciilifeform: ( since was asked for inventory : there are 3 'apu1' amd machines, w/out ears. an' a set of ears. )

diana_coman: asciilifeform: from bucharest with amd-love: amd opteron x2150 apu 4 cores 1.5GHz 8 GB ddr3, centos7

ossabot: Logged on 2019-10-08 06:41:12 diana_coman: o.O there's at least one isp in bucharest actually offering an AMD Opteron dedicated ; take that asciilifeform !

diana_coman: o.O there's at least one isp in bucharest actually offering an AMD Opteron dedicated ; take that asciilifeform !

asciilifeform: the 1 exception being 'apu1' , but is sewn from 2013 'g-series' that amd is constantly threatening to 'end of life'

a111: Logged on 2019-07-28 18:41 asciilifeform: in other noose, 'sse' is... a scam! the 'simd' variant of tlb is actually ~slower~ (at least on my amd 2393se) !!

asciilifeform: in other noose, 'sse' is... a scam! the 'simd' variant of tlb is actually ~slower~ (at least on my amd 2393se) !! ☟︎

feedbot: http://ossasepia.com/2019/07/14/a-working-cuntoo-install-on-amd-fx-8350-with-script/ << Ossa Sepia -- A Working Cuntoo Install on AMD FX-8350 (with script)

a111: Logged on 2019-06-05 16:47 asciilifeform: cuz yer 'amd f' spends 95% of its cycles waiting for ram.

asciilifeform: intel & amd enthusiastically put in 9000 worthless piece o'shit instructions, but how about a 256-bit multiplier ? nope. evidently Only A Terrorist (tm)(r) would ever ask for such thing.

asciilifeform: incidentally when asming on amd (possibly intel also, but i threw out my intel datashits..) can hand-crank the cache when desired.

a111: Logged on 2019-06-04 13:29 asciilifeform: http://btcbase.org/log/2019-06-04#1917019 << i'ma describe, for the l0gz, the hypothetical. suppose mips emu in asm ( the heathen item in fact compiles to a 30kB elf , not esp. complex, but would want clean rewrite really ). this'd be proggy that takes disk image and it tx/rx raw frames to/from nic, so can have own ip etc. then 1 large amd box can host coupla 100 instances of mips linux from known image. <1s teardown/bringup.

asciilifeform: http://btcbase.org/log/2019-06-04#1917019 << i'ma describe, for the l0gz, the hypothetical. suppose mips emu in asm ( the heathen item in fact compiles to a 30kB elf , not esp. complex, but would want clean rewrite really ). this'd be proggy that takes disk image and it tx/rx raw frames to/from nic, so can have own ip etc. then 1 large amd box can host coupla 100 instances of mips linux from known image. <1s teardown/bringup. ☝︎☟︎

mircea_popescu: in other sads : amd fx proc has one helluva hea tsink. it even includes a cool light rim around the outside

asciilifeform: meanwhile, in archaeology, apparently possible to enable eccism-with-amd-fx on certain board which route the pins but have retarded bios.

asciilifeform: there's the famous 'KFN4-D16' asus, afaik the very last to support the very last amd-sans-fritzchip. but that was 'opteron' , rather than 'fx' (and quite rare.)

asciilifeform: btw afaik ~all~ amd cpu since k7 support eccism. issue only ever was chipset/mobo.

a111: Logged on 2019-05-14 08:25 spyked: re. ticket machines: whole ticket infrastructure in buc (and as far as I could tell timis too) is contracted to UTI, who installed embedded shit that for years wouldn't go past the power-on BIOS screen (they used AMD CPUs with some Linux distro on top). ticket machines are a few decades away, although there are maybe three of those in buc.

spyked: re. ticket machines: whole ticket infrastructure in buc (and as far as I could tell timis too) is contracted to UTI, who installed embedded shit that for years wouldn't go past the power-on BIOS screen (they used AMD CPUs with some Linux distro on top). ticket machines are a few decades away, although there are maybe three of those in buc. ☟︎☟︎

asciilifeform: in other noose, amd 'family 22' cpu (i.e. where they introduced fritz core in 2013) has obvious breakage in 'usg.random' instruction (also introduced then, in footsteps of intel) : https://archive.is/BMJwJ

mp_en_viaje: and re the brains of toad : various dorks explained to me why "tjhey don't recommend amd".

asciilifeform: http://btcbase.org/log/2019-05-10#1912285 << heh, no one ever said 'i wish i hadn't bought these vintage amd' ! ☝︎

asciilifeform: BingoBoingo: currently i'm thinking, 2 x dulap-style, 2 x 1u w/ ea. 4 x apu1 'amd g' where cuntoo in write-protected rom . ( i have a rom booter for the latter, built on ancient 'coreboot' )

asciilifeform: BingoBoingo: diana_coman provoked me to think : until arm64 gnat is fixed ( and we dun know when this will be! cuz ave1 is MIA ! ) it does not make sense to expand the rk pilot plant . do we want instead to fill the next crate entirely with amd ? let's make a decision on this in the near term, so that we can put down a definite date re next buildout

asciilifeform: diana_coman: i would proclaim moar loudly that it is 'standard' , but i have no way to guarantee the continued availability of this (or other) particular iron, given as i do not manufacture it , and there is no plan to ever obtain it 'new from factory' , 'clean' amd is no longer produced.

diana_coman: re run: does this mean that currently this is the de-facto "pizarro-iron amd" preferred spec ?

asciilifeform: mp_en_viaje: lacking even photograph, i cannot pontificate in further detail re nvidia. but also do not automatically conclude on acct of 'faster than amd's' that it were hand-sewn.

a111: Logged on 2019-03-23 10:36 mircea_popescu: i mean i get it, they're 20 years old by now. but give me a break, i can find chicks that were spawned 20 years ago just fine, for one thing. and for the other, what the fuck, 2001 era amd cpus STILL to this day the state of the art, they've not yet made better.

a111: Logged on 2019-03-23 10:33 mircea_popescu: sooo, my amd socket 3 mobo that was powering the chatstation crapped out. just like that, turned off for the last time, wouldn't come back.

mircea_popescu: i mean i get it, they're 20 years old by now. but give me a break, i can find chicks that were spawned 20 years ago just fine, for one thing. and for the other, what the fuck, 2001 era amd cpus STILL to this day the state of the art, they've not yet made better. ☟︎

mircea_popescu: and, of course, NO SHOP IN THIS COUNTRY (literally, tried all shops) even has socket 3 amd mobo.

mircea_popescu: sooo, my amd socket 3 mobo that was powering the chatstation crapped out. just like that, turned off for the last time, wouldn't come back. ☟︎

asciilifeform: and meanwhile2, we have answr to above quandary, 'In 2017, BMI2 was further incorporated in AMD's Zen-architecture...'

asciilifeform: http://www.loper-os.org/pub/amd64_mulx.jpg << amd seems to support ~currently~, but dun say from what date of chip specifically

asciilifeform: in lulz inspired by bvt's article, asciilifeform went and dug re 'modern' cpu arithm instructions, and found https://lemire.me/blog/2018/04/19/by-how-much-does-avx-512-slow-down-your-cpu-a-first-experiment/ << intel's crud apparently ~drops frequency~ if you use'em , ultimately nuking all gains from doing so ( they want you to use, so as to shit out binaries that crash on amd, but really gains 0 )

trinque: any more than my amd zenwhatever thing with 30 firmware to load before GPU will work properly

trinque: this is precisely what I do, and I load all manner of "liquishit" hardware (like amd GPUs) without initramfs

shinohai: I wish ... would like to have a barn full of the AMD boards, not the most powerful in the world but run really great, even with own gentoo recipe.

BingoBoingo: <mircea_popescu> vaguely relatedly, is there such a thing as a 64 bit laptop that's acceptable as far as anyone knows ? << I don't hate the thinkpad x120e, AMD E350 Bobcat chip

asciilifeform: ( the 64x64 iron multer in amd/intel, possibly surprisingly, is in fact constant time, in all boxes i've tested to date )

asciilifeform: ( and before anyone asks -- neither intel nor amd publish 'tick tables' the way they did in 1990s, and haven't for many yrs )

phf: i had a 300mhz amd that i brought from russia when i moved here in the 90s, but it finally gave out couple of years ago. i suspect booting templeos was one of the last things i did with it..

asciilifeform: i suppose they were right next to the worx of shakespeare and the amd launch coadez!11

asciilifeform: mircea_popescu: needs diddled bios + the crown jewels of intel/amd, to diddle microcode (intel's is rsa'd, amd's simply obscure/undoc'd) , and if yer diddling bios can make much simpler trap. but yes, would work

asciilifeform: eh |\n it's a duck : apparent source is https://ecc2017.coreboot.org/uploads/talk/presentation/38/Microcode.pdf talk , and demands a pre-diddled, per the recipe, old amd k8/k10

asciilifeform: funnily enuff, as recently as '16 we briefly had a finn here, helped me ferret out the amd key thing, but went back into the allconsuming swamp, never heard from again

mircea_popescu: yes, but... "here's one for amd with raid so and so and fg no external ssd" "here's one for the vibrating bulled you gotta wear per mp orders" "here's one for..." we'll catch our ears.

asciilifeform: mircea_popescu: note that this kernel is tailored to dulap-style box (i.e. amd, and with that particular raid card)