Register a SA Forums Account here!
JOINING THE SA FORUMS WILL REMOVE THIS BIG AD, THE ANNOYING UNDERLINED ADS, AND STUPID INTERSTITIAL ADS!!!

You can: log in, read the tech support FAQ, or request your lost password. This dumb message (and those ads) will appear on every screen until you register! Get rid of this crap by registering your own SA Forums Account and joining roughly 150,000 Goons, for the one-time price of $9.95! We charge money because it costs us money per month for bills, and since we don't believe in showing ads to our users, we try to make the money back through forum registrations.
 
  • Post
  • Reply
r u ready to WALK
Sep 29, 2001



does it have built-in power reporting?

the dell poweredges i dumpster dived has a really neat power analysis tool that lets you know exactly how much energy you are wasting on pointless hobbies



a bare server typically idles at 100-150w but if you stuff it full of drives and memory it quickly adds up
my total continuous usage is around 500w now for the server, an extra disk shelf, network switch and an online UPS
if electricity gets more expensive here i should probably shut it down and get a simple 4-drive nas as a more reasonable setup

Adbot
ADBOT LOVES YOU

Adolf Glitter
Aug 13, 2016





r u ready to WALK posted:

does it have built-in power reporting?



Mine says it does, but It's been too broken to actually have enough graph time. I'm annoyed anyway that in brexitland it's stupidly expensive, and more so with the EU gas wars or whatever the gently caress is going on, but I just got a 140 quid credit from the power company/government, so that's cool. I'll get the power meter thing off my Dad and see how much it costs for a month and decide how important 24/4 torrent/plex/etc is to me

eschaton
Mar 7, 2007

Don't you just hate when you wind up in a store with people who are in a socioeconomic class that is pretty obviously about two levels lower than your own?


my ProLiant DL360p Gen8 generally draws around 150W and thats after filling half the RAM slots (12x 4GB modules) and all the 2.5in drive bays (1x SSD, 7x HD)

maybe if I add a bunch of higher density RAM itll start drawing more

Adolf Glitter
Aug 13, 2016





The lack of graphing was actually down to a loose PMBus wire. I've soldered that back on and it is reporting properly now.
80W currently. That's both CPUs and all the RAM back in too, and 4 old drives. I'm just messing about with TrueNAS so nothing taxing.
-e-There's still 5 fans and NIC to go back in too, so that'll bump it a little bit
I've put the info into a calculator assuming an average of 120W since it will mostly be relatively idle. Comes out to 4ish per week. Now I'm pretty financially embarrassed at the moment but I can manage that for a while I reckon.

Now, the need for some new large drives is a different story. I guess I'll be asking santa for some hard drives this year

Adolf Glitter fucked around with this message at 12:34 on Nov 4, 2021

spankmeister
Jun 15, 2008








Adolf Glitter posted:

The lack of graphing was actually down to a loose PMBus wire. I've soldered that back on and it is reporting properly now.
80W currently. That's both CPUs and all the RAM back in too, and 4 old drives. I'm just messing about with TrueNAS so nothing taxing.
-e-There's still 5 fans and NIC to go back in too, so that'll bump it a little bit
I've put the info into a calculator assuming an average of 120W since it will mostly be relatively idle. Comes out to 4ish per week. Now I'm pretty financially embarrassed at the moment but I can manage that for a while I reckon.

Now, the need for some new large drives is a different story. I guess I'll be asking santa for some hard drives this year

Buy some WD Elements external USB drives and shuck them.

CRIP EATIN BREAD
Jun 24, 2002

Hey stop worrying bout my acting bitch, and worry about your WACK ass music. In the mean time... Eat a hot bowl of Dicks! Ice T





Soiled Meat

aw shucks

Silver Alicorn
Mar 30, 2008

The Game is Never Over


I refuse to shuck a drive

Jonny 290
May 5, 2005

[ASK] me about OS/2 WARP




get the NSA drive alicorn

Silver Alicorn
Mar 30, 2008

The Game is Never Over


I already have a hard drive!

Midjack
Dec 24, 2007





oh hey, it works!





miraculously the sharp-branded aas that had been sitting in it since i presume 1994 hadn't corroded out so slapping some new cells in brought it back up. next up, see if the plotter still works and if the included pens haven't dried out.

FMguru
Sep 10, 2003

peed on;
sexually

Midjack posted:

next up, see if the plotter still works and if the included pens haven't dried out.
if they have, no worries - just call the 1-800 number on the sticker to order some new ones

axolotl farmer
May 17, 2007

Now I'm going to sing the Perry Mason theme


hey YOSmind, is there a thread or resource for doing OCR stuff?

I have some scanned old books about insects that I would like to turn into searchable PDFs. had some success with tesseractOCR and pdftk, but I just have a bunch of stupid questions and I'd like somewhere to ask.

this is my idiot spare time project, making old insect books available again.

SYSV Fanfic
Sep 9, 2003


axolotl farmer posted:

hey YOSmind, is there a thread or resource for doing OCR stuff?

I have some scanned old books about insects that I would like to turn into searchable PDFs. had some success with tesseractOCR and pdftk, but I just have a bunch of stupid questions and I'd like somewhere to ask.

this is my idiot spare time project, making old insect books available again.

https://archive.org/services/docs/api/ocr.html#code-repositories

PIZZA.BAT
Nov 12, 2016

:cheers:



so i've got the java part of my idiot spare time project up and working and am now looking at the python/ai side of things and man.... this poo poo is complicated lol

i'm reading through tensorflow documentation and while nothing is completely beyond my grasp it's just every single thing i read opens up like a hydra of even more things i need to read. i'm eventually going to hit a point where i go, 'oh... ok now i get it i know what to do now' but it's probably going to be a while

edit: the most frustrating part is even in their tutorials they do the, 'ok now draw the rest of the owl' a whole lot. i understand that this is a brand new and complicated field where the only real contributors are academics at this point but it'd be nice if their tutorial code used real loving variable names instead of things like _

PIZZA.BAT fucked around with this message at 14:10 on Nov 9, 2021

axolotl farmer
May 17, 2007

Now I'm going to sing the Perry Mason theme



thanks, but I'm not looking at doing this on an industrial scale. I'm mostly looking for things like how to proofread and change errors in tesseracts OCR and then get that into a searchable pdf.

I got so far as to make a searchable pdf from scanned pages, but don't know how to fix OCR errors.

barkbell
Apr 14, 2006

woof

PIZZA.BAT posted:

so i've got the java part of my idiot spare time project up and working and am now looking at the python/ai side of things and man.... this poo poo is complicated lol

i'm reading through tensorflow documentation and while nothing is completely beyond my grasp it's just every single thing i read opens up like a hydra of even more things i need to read. i'm eventually going to hit a point where i go, 'oh... ok now i get it i know what to do now' but it's probably going to be a while

edit: the most frustrating part is even in their tutorials they do the, 'ok now draw the rest of the owl' a whole lot. i understand that this is a brand new and complicated field where the only real contributors are academics at this point but it'd be nice if their tutorial code used real loving variable names instead of things like _

i joined a scala shop recently and its just _ and a and b and whatever meaningless name for every variable. must be an academic thing

pseudopresence
Mar 3, 2005

I want to get online...
I need a computer!



Is _ not just the 'don't care' variable for return values you don't need? Are they using it in other contexts?

animist
Aug 28, 2018




PIZZA.BAT posted:

so i've got the java part of my idiot spare time project up and working and am now looking at the python/ai side of things and man.... this poo poo is complicated lol

i'm reading through tensorflow documentation and while nothing is completely beyond my grasp it's just every single thing i read opens up like a hydra of even more things i need to read. i'm eventually going to hit a point where i go, 'oh... ok now i get it i know what to do now' but it's probably going to be a while

edit: the most frustrating part is even in their tutorials they do the, 'ok now draw the rest of the owl' a whole lot. i understand that this is a brand new and complicated field where the only real contributors are academics at this point but it'd be nice if their tutorial code used real loving variable names instead of things like _

strongly rec just cloning a project that does something close to what u want and then tweaking it, if possible. bringing up ML projects is hell, i always forget some tiny super important detail. then the model doesn't converge and it's undebuggable cause there's no way to debug why your gradients are wrong. if you use an existing codebase usually a lot of those critical learning rate tweaks or whatever are already baked in

barkbell
Apr 14, 2006

woof

_ is wildcard in scala, no idea about pizza.bats shop tho

PIZZA.BAT
Nov 12, 2016

:cheers:



pseudopresence posted:

Is _ not just the 'don't care' variable for return values you don't need? Are they using it in other contexts?

they were using it as an iterator that they pulled stuff out of and manipulated all the time

like.... please give me a meaningful name.... please

this is tutorial code for gently caress's sake!

PIZZA.BAT
Nov 12, 2016

:cheers:



animist posted:

strongly rec just cloning a project that does something close to what u want and then tweaking it, if possible. bringing up ML projects is hell, i always forget some tiny super important detail. then the model doesn't converge and it's undebuggable cause there's no way to debug why your gradients are wrong. if you use an existing codebase usually a lot of those critical learning rate tweaks or whatever are already baked in

the problem is no one has done* what i'm attempting so the only real way to go about it is learn what all these different components actually do and why, unfortunately. i'm attempting to teach a bot how to play poker. i spent a few months looking through the literature on other academic shops who all took swings at it and was always unhappy with their implementations because they'd miss fundamental aspects of high-level play. the next few months i built the simulator which was a bit of a pain in the rear end because, surprise, it turns out poker has a shitload of edge cases in it! after that i built a class that runs all the evaluation on the table/hand state which will create the massively simplified information i feed into the neural net so it doesn't have to spend who knows how many millions of generations to learn the difference between a low and high straight

the big problem is this is a mutli-actor game with not only hidden information but tons of scenarios where making the right play can punish you, wrong moves can reward you, and scenarios where a good situation can quickly turn bad as you see additional community cards. i'm pretty sure what i want is an actor-critic model, probably with a SAC agent because maximizing entropy is definitely something you want in a game like poker. the problem is everything i've read so far is making it seem like you have to explicitly define the expected reward function so the critic knows how to criticize but.... how am I supposed to do that if the whole point of this exercise is to let these models discover optimal choices on their own? every tutorial i've read just seems to gloss over this and it's making me think i'm in the totally wrong place

*there may be a handful of people who've pulled this off but i'm gonna guess they're raking in money in online casinos and keeping their mouths shut

Cybernetic Vermin
Apr 18, 2005



PIZZA.BAT posted:

*there may be a handful of people who've pulled this off but i'm gonna guess they're raking in money in online casinos and keeping their mouths shut

i assume you checked out rebel? https://github.com/facebookresearch/rebel

the bits needed not part of the open source release, but apparently did well on poker

Cybernetic Vermin fucked around with this message at 18:31 on Nov 9, 2021

PIZZA.BAT
Nov 12, 2016

:cheers:



yup. thats one of things that inspired me to make this in the first place. the problem i had with their implementation is it relies heavily on monte carlo which for a game like poker is insane. with a heavily simplified decision space they were able to run it on *only* 1.5 petabytes of disk

my main objective is to create a model that largely acts on information professional level players act on and also evaluates hands similarly. humans dont see the flop and go ok heres all the possible combinations of turn and river cards we can see along with all the actions the other players can take for each of those combinations so therefore my bet is X

they go, ok i have a flush draw, top pair with a strong kicker, and theres no risk of a straight. my opponent has a loose opening range and also tends to chase. my bet is X

Progressive JPEG
Feb 19, 2003



Progressive JPEG posted:

oh that's great thanks

so the fact that it takes the same 32K tag size across 128K-512K cache configs makes me think that the tag size isn't necessarily a fixed ratio, so i might be able to get away with two 32K chips in those slots if it doesn't work without them populated

the store owner claims to test their chips but yeah i figured i may as well just get eight of them even though only 4 or 6 are needed depending on whether those two undocumented mystery slots are indeed needed for tag memory

so i now have a functioning 128K of completely pointless L2 cache. it ended up requiring a couple 8K SRAM chips. luckily mouser has them. a bit pricy for a few kilobytes (particularly given all the "spares" i've bought along the way) but this is an ISTP after all


(picture: cache benchmark showing 16ns for 8K L1 cache and ~31ns 128K L2 cache)

so the working permutation was:
- 2x 8K SRAM in the two slots to the left of the CPU (bought from mouser)
- 4x 32K SRAM in the four slots below the CPU (bought from monotech, but mouser also has them), with the 28pin chips aligned against the flat/non-notched end of the 32pin slots

other permutations tried:
- if the two slots are left vacant, no cache is detected at all. left things like this while waiting for the 8K chips to arrive from mouser since it was easier than removing/reinserting the chips
- if i put 32K chips in the two slots, 128K cache is detected but the system becomes very unstable, acting as if the cache is busted. on the rare occasion that it passed POST, dos itself would fail to boot
- if i (accidentally) leave the jumpers in the 32K*9 position, the BIOS helpfully catches this and displays an error about invalid memory configuration
- if i put the 32K sram chips in the wrong position aligned against the notched end, the system fails to start at all (but didn't seem to emit any blue smoke - i quickly shut it off)

while waiting for the mouser order to arrive, i did some other fixes/updates:


(picture: SRAM populated/working, cpu heat fixes, and eeprom/bios card - removed large sound blaster card for photo)

- added an EEPROM ISA module with xtide. after some fiddling with bios offsets, i was able to get xtide working and now the CF card hard drive works great, with none of the write errors like i was seeing before with the stock bios. xtide also made it possible to switch from the crap 2GB CF card i was using to a much faster 32GB CF card that the stock bios would refuse to detect at all, and have ~8GB of it accessible via four 2GB partitions. after lots of failed attempts at trying to replace the IDE hard drive with something that isn't due to die a decade ago, xtide was ultimately the missing piece that finally made it possible to have the system fully usable on solid state storage. otherwise i'd still be having problems trying to install anything to the system to this day due to all the bios write errors

- the 486dx2-66 cpu was getting quite hot to the touch. i saw online that a heatsink (if not full HSF) was generally recommended for it, unlike the original 486sx-25 (pictured). i already had a whole spare bag of small heatsinks with sticky pads for use on raspberry pis, so i covered the 486dx2-66 with a tidy 3x3 grid of those. i also ziptied a small fan nearby to give it some airflow. now it's lukewarm to the touch.

- i replaced the original ~1993 AT power supply with a spare OEM-tier ATX power supply, using an adapter cable. happily the ATX psu standard seems to match the width/height of the prior PSU, so it was a perfect fit, just needed to dremel some new screw holes into the back of the case. as a bonus the ATX PSU fan is much quieter than the old PSU's fan was. after the swap i opened up the old PSU and it indeed looks due for replacement :stonk::

(picture: inside of vtech AT PSU from 1993)

at this point the system itself is pretty much "complete". short of switching to a new motherboard entirely, everything in the machine is maximum supported spec. the only remaining things to fix are:

- i want to try getting a more modern mouse hooked up. i currently have a very-90s microsoft serial ball mouse and i forgot how much it sucks for something like tie fighter. i'm going to try this rube goldberg solution with an unused rpi1 and see how it goes. just got a usb null modem adapter to try it with

- the new ATX power supply doesn't have it's own power switch, instead i'm supposed to add a switch to two leads coming off of the adapter. at the moment i'm powering the system by plugging in the IEC cable but should hook up a basic SPST switch that i just got to connect the leads and dremel a rectangular hole for the switch in the back of the case

Progressive JPEG
Feb 19, 2003



oh and another thing i still have left to try is reflashing the 32K EEPROM chips on the aforementioned isa card that i'm using for xtide

i don't know much about the underlying mechanics of flashing eeproms, but afaict the two chips i got with the card both seem to have distinct problems:
- one doesn't seem to be readable/writable at all - bios doesn't see it when its enabled, and trying to flash it with xtidecfg doesn't seem to do anything either
- the second one is what i'm using now. it has working xtide (flashed by the seller) but it came with a stock config and i don't seem to be able to write an updated configuration via reflash with xtidecfg. in the meantime this one is "aint broke/don't fix it"

to be clear when doing reflashing i am turning off shadowing and also enabling the respective 'write' jumper switch on the board, so in theory it should "just work" but lol. meanwhile the two chips seem to have different errors when i'm trying to overwrite them, iirc one is "failed to write" and the other is "read data didn't match written data". i tried swapping the chips between the two sockets and the behavior followed the chips, making it look pretty likely that the chips themselves have issues. since i was ordering a bunch of stuff from mouser anyway, i ended up just grabbing two new brand-name atmel 32K eeproms to see if they work any better

summary/question: is it common for 32K eeprom chips to go bad like i've been seeing with these (unable to write and/or unable to read)?

Progressive JPEG
Feb 19, 2003



so i tried popping one of the new mouser eeproms into one of the isa board sockets just now, and after a lot of experimentation the only way flashing it seemed to work properly was to disable the roms and boot onto a floppy, THEN switch on the 'enable rom' after booting had finished. feels like the stock bios is doing something sneaky like blocking writes to roms that have been loaded into memory, even when shadowing is disabled for their address ranges

so idk what exactly is going on, but i seem to have stumbled on a magic formula for being able to flash things using xtidecfg

in the "failure" state it'll just flash a progress bar for a fraction of a second then say "okay done! rebooting!" but then nothing will actually be written. meanwhile in the "success" state it actually takes a couple seconds to write and then show the success screen.

hbag
Feb 13, 2021



got like 3 hours to kill because i missed my lecture (northern rail ftw!!!!) so im going to work on the fyad flag tracker and also maybe yosbbs while i get lunch

Empty Pockets
Jun 11, 2008


Im beginning work on a character combat tracker for TTRPGs on my calculator, starting with testing this printer I just got in the mail.

GWBBQ
Jan 2, 2005




This guy batteries.

Hed
Mar 31, 2004



Fun Shoe

Empty Pockets posted:

Im beginning work on a character combat tracker for TTRPGs on my calculator, starting with testing this printer I just got in the mail.



I had to double take that calculator on the right neat!

eschaton
Mar 7, 2007

Don't you just hate when you wind up in a store with people who are in a socioeconomic class that is pretty obviously about two levels lower than your own?


the SwissMicros are really awesome, I need to use my DM42

I really wish they made a DM48, I still find 48SX-style RPL to be the pinnacle, to the point where Ive installed DOSBox so I can run HPs IDE

Empty Pockets
Jun 11, 2008


eschaton posted:

the SwissMicros are really awesome, I need to use my DM42

I really wish they made a DM48, I still find 48SX-style RPL to be the pinnacle, to the point where Ive installed DOSBox so I can run HPs IDE

Im trying to figure out a dev environment on my PC so I dont have to do everything from the tiny calculator screen + keys, any tips?

Also, not sure if you keep up with them but theres speculation since their May update video on youtube that the DM48 might be next. Nothing from SM themselves though, exactly.

eschaton
Mar 7, 2007

Don't you just hate when you wind up in a store with people who are in a socioeconomic class that is pretty obviously about two levels lower than your own?


no idea how to do HP41/42 dev on PC, I bet Usenet archives from back in the day have some stuff though

the HP-48SX IDE for DOS (Program Development Link, PDL.EXE) was pretty bomb

Sniep
Mar 28, 2004

All I needed was that fatty blunt...



King of Breakfast


eschaton posted:

the SwissMicros are really awesome, I need to use my DM42

I really wish they made a DM48, I still find 48SX-style RPL to be the pinnacle, to the point where Ive installed DOSBox so I can run HPs IDE

I would so buy one just to play with but then i remember, i don't do math

Empty Pockets
Jun 11, 2008


the DM41X apparently has a serial console mode, but the interaction is all through issuing byte codes for the keys on the calculator, plus some other functions. https://dm41.swissmicros.com/ I might just work with an emulator though, V41 looks pretty good so far. http://www.hp41.org/Emulation.cfm. There's also this insane excel spreadsheet that connects to an emulator through VBS macros, but I'm not sure I can bring myself to run it.

animist
Aug 28, 2018




excel in the wild. a magnificent beast

SYSV Fanfic
Sep 9, 2003


So I made a prototype video card with a pi pico for my rc2014. PIO is nice.





Due to my prototyping method it's very sensitive to jostling and when I play music via bluetooth.

pls excuse my gaming mouse pad. It was on sale for $5 and was jumbo.

QuantumPotato
Feb 2, 2005



Fallen Rib

SYSV Fanfic posted:

So I made a prototype video card with a pi pico for my rc2014. PIO is nice.





Due to my prototyping method it's very sensitive to jostling and when I play music via bluetooth.

pls excuse my gaming mouse pad. It was on sale for $5 and was jumbo.

nice.

did you use micropython?

SYSV Fanfic
Sep 9, 2003


QuantumPotato posted:

nice.

did you use micropython?

I initially prototyped my PIO code in micropython. I wound up going entirely with C. Unless I missed something somewhere: anything you want to do with DMA means writing a C extension for micropython, which means learning how to build micropython for that board. It wound up being easier to just do it in C.

https://github.com/keyvin/docnotes/tree/master/z80

Adbot
ADBOT LOVES YOU

hbag
Feb 13, 2021



my current free time project is getting loving zonked over the weekend and Posting Goodly thanks for listening love you

  • 1
  • 2
  • 3
  • 4
  • 5
  • Post
  • Reply