#tech-chat-1
1 messages Β· Page 1433 of 1
id have spent that extra
OK thats less than the tax it doesnt matter
932.32 vs 1087.70 to be exact
only reason im not still using my 48C1 is because i moved, and its big
if they're ikea ones with a melamine face or sound hollow like a fire door, they'll want extra support tbh
ill take a look when they get here (about 2 weeks im told), i may add an extra leg in the middle for support
D a y u m . . .
Mood...
if you can, get one of the extendable legs
yeah, was thinking about the adjustable leg from ikea
similar ones on amazon also, https://www.amazon.com/dp/B08H4RWFR2/
yeah, there's some where you can screw in the lower legs for locking them off at their adjustment point if you have an issue with the friction style this and the ikea ones have
https://www.amazon.com/dp/B09YCFR6MM <- $40 for 2
also the footpad is unscrewable for adjusting
its ikea time
this looks like a pretty cool option
friction locking on this one right?
i.e. twist one way to loosen, and the other way to tighten it
appears to be, yeah
nice to have the options available more so nowadays
i'll take a look into all 3 options and see which one matches my table the most to be honest, it just needs to have the same height as my other table legs and i wont touch it again
That would be be a 5090
You are homosexual
Its more of a should I stay with my porsche or buy a track spec porsche
I indeed am not and i am sure about that
I will buy the 5090 or 6090 if they fall under 3000 euro
But my 9070XT nitro is well good for the moment
yeah, which is why i'm thinking of the one you can lock the adjustment on tbh, set and forget with it
As the 5080 doesnt seem to have that big of a boost in performance
missing your caffine fix?
βi was gonna buy a bugatti but i saw gas prices so im buying the cheapest ev out thereβ
worse
knowing that my refund review is only gonna be out on monday
Is your purpose to present the lack of intellect or what?
The 5090 is currently like 3500
The astral is like 4000
then you suddenly see electric prices go up
As soon as the astral falls under 3500 or the basic 5090ts fall under 3000 i will buy one
The 9070XT nitro is also a 1k GPU, so I do not see the comparisons you are trying to make
buying a 9070xt nitro for presumably 1k USD has gotta be one of the biggest fumbles oat
Switzerland, what can I say
A good 5080 is like 1800 here
I mean I bought it for 800 euros, so close to 1k US
whats ur cheapest current 9070xt u can find anywhere
post tax
669.99 euro
wait a 9070xt is $650 usd here
771us
771 USD
I found one for 650 euro
Thats like 750ish USD
The nitro + is the most expensive you can get rn here
i mean at least thats still understandable but that 5080 is not inline with anyt lol
Wdym inline with anything?
2k usd for a 5080 assuming its a shitty model
ROG astral
I will spend more for looks then just shitty cards
Thats why I bought thr nitro+
It looks good and was more available then the 5070ti or 5080 at that time
About close to 1 year ago
and this is how nzxt fools people with shitty products because led's and looking good
I didnt buy my things because it was basic stuff to max performance. I bought a lot of those things to look good aswell. I bought an 150 euro AIO to cool a 9800X3D even tho a normal air cooler would be enough (Lightloop 360)
I bought veganange RGB DDR5 and not the stealth versions
I bought an ROG board even tho a normal tomahawk would be enough
I only own a H6 Flow RGB
also buying a nitro+ means performance as well as looks, when it comes to amd cards
Ofc i took that into account, as a nice extra
But like a quicksilver is also close for a lot less
Or mercury
I mean also true
An AIO is also a nice to have anyways
Like if AMD decides to bring out a Intel type beat CPU that pulls 240W, im strapped for it
thats not uncommon
wsg rauko
Hi chat so rn I have hex os installed on my old pc and I have zipline installed on it, I have a cloud flare tunnel setup and pointed at the right ip and port with my domain but cloud flare can not reach zipline, I can reach zipline from my local pc but going to my domain errors 502
i can't justify 5090 price for work, meanwhile people are buying it for gaming π
contemplating on how much rum i should knock back
as your personal doctor
i say all of it
worst GPU I ever used fr
idk why u thought it'd be good
could be worse, could be xfx instead
the preformance is great for what I used it for, MSI just had a small brain moment and put a GTX 960/1060 OC cooler on an RX 580
since im poor the only thing I could replace it with was basically the same card just with an insane arctic cooler
zipline is a file sharing service?
Is a webserver
yeah to share files?
@native path highk after the refund do I just grab a 27β tandem WOLED 280hz glossy to bide time till 4K 32β tandem
Yeah and?
im just wondering
anyways are you able to access zipline locally
I use zipline
Ofc, you have to host it locally to be able to host it remotely
well he cant get it to work
I also use cloudflare for zipline
you arent running into issues tho cheez is
But I just have a port open locally and then forward it on cloudflare tunnel
tbf when the a6000 pro exists, the 5090 is pointless
clippy, the only assistant that did truely help in windows
I need to somehow justify getting a pair of arc b70βs for myself
5.7k euros ? for home use, even the 3k euros for 5090 is hard pill to swallow
and anything between 32 and 96GB VRAM is in a zone of 'not really much to do with local LLM there'
Its only 150 with the right stuff 
better than a 5090 for both gaming and work use, also has a fuck ton more vram,
A6000 pro is older than the 5090, and not better
never understood why the less than 96gb rtx pro cards existed
like 90% sure rauko meant the rtx 6000 pro blackwell, not the a6000
βBudgetβ
I know 
yup, my bad
in that budget a couple 5090s makes far more sense
a single 64gb rtx pro is more than 2x the price of a single 5090, so just buy two 5090s and get more compute
Arc b70 would be even better for workstations since it actually supports multi gpu
Even has vram pooling
Iirc
for LLM it is not really usable. ~32GB is for models like Qwen3.5-35B-A3B or Gemma4-26B-A4B. Then next level are models like Qwen3.5-122B-A10B, which would require ~90GB VRAM (both with reasonable context length)
so the 48GB cards sits in middle of nowhere
B70 has the issue of "intel support is even worse than AMD for LLMs"
@brave brook something even better for llmβs would be an apple sillycon
Comfyui dynamic ram is pretty nice
prompt processing speed is horrible from the benchmarks i seen
actually no dude
even the M5 Max are hitting only ~2k tokens/second for prompt processing, which is horrible compared to 500t/s of my dual 1080 Ti setup
How much ram though
wouldnt the whole selling point of the blockchain integration be to keep validation after the store is closed?
nvidia's new nvlink backplane, that's what you want for llm's
how is that relevant ? if the model fits, i care about how long do i wait for that 30k context of Claude code to be interpreted π
you grabbed the source code from the leak?
and for that 60-120k or more context of debate with LLM
you didn't need to, setting up Claude code to talk with any LLM is just environment variable settings
as long as the LLM talks Anthropic (which LM Studio, Ollama and llama.cpp all can do), you just set environment variables and it talks to your model and servers
M series can have 256gb of ram on just the gpu (on paper, in reality its closer to like 250-253gb)
but if you did, you could take out a couple background parts that make it monitor
If you have one of the older studios, you can have 510gb of vram
there is always the python and rust rewrites if you want that
but we are again back to prompt processing, that is important to me. Huge VRAM is nice for bigger models, but i am fine with the lame 35B Qwen3.5 model for now. Sure, would be nice to have the 122B model, but not at cost of needing to buy 4x B70 Pro instead of 1x B70 Pro/2xRX9060/2xRX9070/2x5060Ti
What model did you use for your benchmark of llm on your 1080ti setup?
Was it qwen3.5 35b?
https://huggingface.co/unsloth/Qwen3.5-35B-A3B-GGUF - UD-IQ4_NL, 17.8 GB. with ~120k context length. On average on this setup it gets ~400-500t/s prompt processing, ~20t/s token generation (in claude). In chat the generation is higher
unfortunately i am not willing to pay thousands for this. ~1.5k euros is tops i would pay right now
@brave brook this?
more or less. search for unsloth
but the diff is not big
keep in mind the prompt processing is the issue for me, so throw it a ~30-50kB text file to process π
opinions on what the quietest aio could be?
Can you dm me one?
i don't really have one, that is the size of initial claude code prompt usually
running that fucxker without cooling
well the fan is directed at it but no heatsink
oki
Not really that bothered but i got 32gb ddr5 today but in my task manager it says i have 31.1 π
kinda crazy how you can just use a cpu without cooling or perform a repaste while its running
anyone know why?
windows
usually just take a random long webpage/PDF, copy-paste the content and then ask it to sumarize it. and then on developer tab count how long it takes for it to leave prompt processing π
hardware reserved
rest is hardware reserved and reserved to igpu
oh
hah i have more ram than you
get rekt
its a safety thing
been scammed for 900mb (10 million dollars)
its allg ig
ram prices fell of a cliff and im poor again
damnn
im only scammed for 800mb
do you have an IGPU ? Reserved memory for that, it doesn't have it's own RAM
i legit just upgraded to 32gb ddr5 and a 9800x3d today
get real lil bro
yea i have a 7800x3d
i was having alot of issues at first but i managed to fix it
ongg
i need a new motherboard to get to this level
bro i bent the pins on my usb 3.0 header
i cant get it in anymore
oh my god its installing an update thats why the fan began screaming
laptop cooling ahh
i just got a motherboard
im back on desktop
what do u actually use that much ram for though
even better, im on linux :3
also i never even touched 32 on windows
is linux any better for windows?
uhm
arch user try not to mention theyre on linux challenge impossible
its going to explode
do i loose out on performance if im running pcie gen 4 on a 5070ti π ?
hey arch users would be very offended about that
im not a arch purist
absolutely not, the difference between 4.0 and 5.0 is so minimal youre fine without 5.0
Like this chart - 2000t/s prompt processing - would be worth investing. 1000t/s - barely double of my current free HW. And these are M5 Max and M3 Max (src https://www.reddit.com/r/LocalLLaMA/comments/1s5np41/m5_max_vs_m3_max_inference_benchmarks_qwen35_omlx/)
alright thanks cos my motherboards pcie 5 but i use a gen 4 riser cable so
well no pcie 5 is 2x as fast
but for gaming the difference is next to nothing
fuck you mean no bro people told me otherwise here months ago
my pc literally wasnt working with my new motherboard before because of the riser cable
it was constantly crashing and lagging
yes u are
wat
every pcie version has been 2x faster than the last
but it does not mean its better for gaming
when does it actually matter?
looking at debate from link above, Qwen 3.5 seems to be ~2-3x higher numbers than llama 7b numbers (https://github.com/ggml-org/llama.cpp/discussions/4167), but that would still put the base m4 mac mini with 24-32GB RAM only to same-ish speed as my current setup, and it would cost me more than buying the dual GPU options
for other equipment than gpus
like networking
oh right
Even the 4090 runs at full performance at pcie 2.0x16
another post https://www.reddit.com/r/LocalLLaMA/comments/1s4bggo/benchmarked_qwen35_35b_moe_27b_dense_122b_moe/. I mean i can test out the m4 mini with 24GB RAM next week, so i can see how will it perform - but i have my doubts if it is worth it
this server has people who sometimes say completely opposite things
who to trust π
with PCIe connectivity of the GPU, the bandwidth only matters when transfers are occuring. Which do not happen as often as people would think. Which is why often the losses from going slower generation or less lanes are smaller than people expect
going from 5.0 x16 to 1.1 x16 you lose 26% of performance
3.0 x16 is 4% loss... with 5090, at 1080p
at 4k, the differences are even less
so all vram important things do benefit a newer pcie version where regular usage doesnt
I had a stroke reading this
read better
no, pcie speed doesn't hurt other gpus either. 5060 Ti 16GB going from 5.0 x8 to 3.0 x8 is also just 4% performance loss. The game engines simply do not transfer enough data back and forth for the PCIe bandwidth to be the limiting factor.
Type a proper sentence and use punctuation
usually only smaller VRAM situation is where it hurts, like 8GB cards
Check out the UGREEN 100W USB-C Magsafe Charger: https://amzn.to/3TjjwII
Support us on Patreon: https://www.patreon.com/hardwareunboxed
YT Membership: https://www.youtube.com/channel/UCI8iQa1hv7oV_Z8D35vVuSg/join
Buy relevant products from Amazon, Newegg and others below:
Radeon RX 9070 XT - https://geni.us/CH0CxfR
Radeon RX 9070 - https://gen...
dgaf, not a native english speaker
There you go
i absolutely get it
Which ai
yeah, token generation is fine, it's when it needs to process long input is the issue i fear, because claude code usage starts with claude sending it a ~25-30k prompt telling it what tools can it use etc. And that takes good ~50 seconds on my dual 1080 Ti setup, because it processes prompt only at ~400-500t/s. And from what i seen, a base M4 mac Mini with 24-32GB RAM should give me similar numbers, which would make the investment not worth it.
going from 50 to 20 seconds (~2k t/s prompt processing) with something like dual 9060 XT would push it to a very usable territory on other side. I mean the current setup is not horrible, i fire a prompt to it, it takes some time for it to do the first prompt and due caching next prompts are relatively OK. But when it decides to compress prompt at reaching ~100k prompt length, well, that is a good 3-4 minute break π
Qwen3.5-35B-A3B in Q4
Have you tried the new ollama release? Its pretty cool, its the MiniMax-M2.7 and it has self improvement built in
running it with latest llama.cpp. Ollama and LM Studio have a bit worse performance and less configuration π
not sure what you talk about tho π
if this https://x.com/JulianGoldieSEO/status/2037262406104313879 then that's just an openclaw logic thing, is not reallyt tied to ollama or minimax - it is just processing information to text files based on certain logic
This free AI stack is way more powerful than most paid tools.
And it can run on your computer 24/7.
MiniMax M2.7 + Ollama + OpenClaw gives you:
β a self-improving model
β a free local/cloud runtime
β a real AI agent layer
β API switching inside OpenClaw
β skills that get
i might try running a local model on my workstation
probably will run like crap
gpu only has 12gb of vram, plus 128gb of shared
you can always run something smaller - depending on what you want from it. For example Qwen3.5-9B can still do good enough text summarization, or vision tasks.
@brave brook just tried to get it to make a program that detects movement, and it hit 34 tokens/s
i have a 25gb gpu
Didnt mean to ping
you can either use opencode and configure it against your LM studio (via OpenAI compatible provider), or do stuff like https://lmstudio.ai/docs/integrations/claude-code if you want to use claude code. I also set few others (CLAUDE_CODE_ATTRIBUTION_HEADER=0, API_TIMEOUT_MS=3600000, DISABLE_TELEMETRY=1, DISABLE_ERROR_REPORTING=1, CLAUDE_CODE_DISABLE_FEEDBACK_SURVEY=1, CLAUDE_CODE_DISABLE_1M_CONTEXT=1, CLAUDE_CODE_AUTO_COMPACT_WINDOW=115000 - this one to whatever context length you have in lm studio) and ANTHROPIC_DEFAULT_SONNET_MODEL/ANTHROPIC_DEFAULT_HAIKU_MODEL/ANTHROPIC_DEFAULT_OPUS_MODEL to the model name in the LM Studio/Ollama/llama.cpp/vllm. And boom, it talks to local LLM instead of the paid sub π
is overclocking a 9800x3d worth it? will i see a decent increase?
nope, it's like the difference between pcie gen 3 and gen 4, it works out to 5fps difference total
pbo +200 curve -20 and forget
@frosty gyro this but also run a stability test
aight
LMAOOO
QRT: Snowl_OW
Thank you @PulsarGears for including the normal mouse and some extra goodies with the April fools limited mouse!!! Not expected but very fun and enjoyable :D
g502 users in 2026
Nothing wrong with the G502
the wireless one is like 130β¬
and theyre heavy
i mean for the people using them sure
though some people just dont care ig
but yeah its a pretty overpriced mouse for wat u get, battery good tho logitech batt management
though i can never go back
i had to go from a superlight to a g502 while waiting for a new mouse
Some people like the weight
terrible
i have 2 g502s just sitting due to circumstances
I love the G502
Using a g502 X plus at the moment after someone stole my g304
scyrox v8 and saber v2
I love my G502 X Plus
the scyrox is great but it doesnt like my mousepad
Though I have always been a use to heavy mice
i now use fingertip grip (probably cause of my mouse) so i doubt the g502 is very usable to me anymore
a used viper v3 pro for like 60β¬ seems nice
but im probably not hardstuck because of my mouse
Mx Master series was the only mouse I have used in the past. That rubber material did not last.
Well till I put something on the Mx Master 3s, seems to be helping in the meantime until that one becomes gross.
is that why my b650 still has a ps2 port
been a long time since I last used a heavy mouse
My RC-310G certainly is
gameurrs who want to blame 100000 things before their skill issue will riot when their muh lesser response time port is removed
sounds like the same crowd who still swear by CRTs
or do they want me to pay extra to rid my eyes of this dinosaur port
lesser response time yes but
wtf do i plug in to it
high performance ps2 keyboard / mouse
nvidia pulsar sadly snatched crt's chain
for latency fans atleast
we already got sub-1ms on wireless mice/keyboards, latency has been a solved problem for a while already
in terms of peripherals at least
the black levels are fantastic and they're pretty satisfying
well the superstrike seems cool
wayyy too expensive tho
waiting for the chinese brands to copy it and buy it in like 5 years used
or if my aerox 3 wireless dies
3 payments of 59.99β¬ π₯
do any of you guys know where i could find the geforce led visualizer for controlling the geforce gtx brightness on older cards?
maybe i understand it wrong but isnt it just like a hall effect mouse
wooting was expensive af and now u can get hall effect keyboards for like 50β¬ new
pretty much
don't remember the actual tech used, but it's adjustable actuation point similar to HE
so there will probably be others on the market soon
haptic feedback and the travel being much much shorter than for any keyboards probably makes it somewhat different technologically than the keyboard stuff
yeah the haptic thing is weird
this was already in laptop touchpads so i wonder why nobody did this before
wasn't necessary with a mechanical switch
It's probably for getting into bios in stages where USB might not be active
Because PS/2 has higher latency than modern USB so that's not even it for gamers
why is it a lower-midrange feature tho
Idk
put like empty space instead of it
they got spare room on IO panel on lower range mobos
i dont like that its there cause idk
so might as well slap on stuff that costs 30 cents like PS/2
its for liquid nitrogen
generally when the board is super cold the usb controllers drop out
ps2 does not have that issue
if its on a shitter board its just leftover from making a higher end board with the same pcb design
LN overclocking my b650m s2h
except who does LN2 XOC on a midrange board 
people who have ln2
.
ln on tap water
i really dont think ln2 cooling is a major reason boards still have ps2
I notice an extreme lack of IO
yeah but still freakin ps2
spam usb c
and separate keyboard and mouse wow
sometimes the USB controller doesnt work while cold, and if the PCB design already has it why not keep it
id have to think its more related to legacy/businesses still needing ps2, ln2 overclocking just cant be that big of a market share to be the main reason
with ddr5 can 2 slots be the standard
can someone email gigabyte
Z690 aorus tachyon
no IO shield 
im sure there are other uses for them putting it there but the only explained reason ive ever seen is for the oc boards, and its the only one that really makes that much sense to me
it just being left over from typical xoc designs makes sense
can you explain the lack of integrated io shield mr wise guy
thats because gigabyte is regarded
i dont ask ai i ask floppa
only ever had one mobo without integrated IO shield but that's because it was a server board
i mean even with oc boards how many people are really doing subzero cooling? im a bit outside of the OC scene, but idk it just doesnt seem like it is a large enough market of people to be a big contributor
id really just guess it doesn't cost anything to add and they still have function outside of usb
gigabyte invented side io with that usb a port
it is required for anyone to do sub zero cooling
6x sata for your overclocked nas
it doesnt matter if its 8 people. those 8 people are sponsored by asus and asus wants records lol
same for the other vendors
oc for vendors is like the racing world to consumer car companies
big marketing
yea fair enough
"easy stable ram oc"
it is the best DDR5 board on z890 so
this is the type of oc my "ai" z87 board would recommend me
i mean look at the apex with insanely niche oc focused features, you think they wouldnt put ps2 on the boards for ocers as well?
there is no consumer in the world using that slow mode button
rename it to turbo
true
this looks so cool wtf
Interesting
I think I might have an old PS/2 mouse from a gateway PC with windows ME somewhere
yur
vga for extreme overclocking
and also i think technically ps/2 keyboards are lower latency technically technically
err at least if you had a modern ps/2 keyboard
if such a thing exists
"this sub-1ms thing has lower latency than that sub-1ms thing" - I don't think these arguments make sense anymore
usb doesnt have true N key rollover. boom.
peripherals combined adding like 2ms to the total 50ms end-to-end latency and gamers being like "must get lower latency peripherals, SURELY I'd be able to tell if it was 49ms instead of 50ms"
and yet x3d is raved about
....what?
X3D helps the CPU access things faster because it doesnt have to keep them in slower memory instead, slower meaning on the scale of nano seconds
X3D is raved about because it gives double digit performance gain in a lot of titles, not because "lower latency"
the performance is from the lower latency bro

π do you understand how the levels of cache in a computer work?
floppa im back on desktop
decided laptop wasnt for you?
7600 with 96GB ddr5
or just have two now
No I didnβt pay attention in class
latency perceived by humans directly for individual actions is a completely different topic from latency that affects the roundtrip time of something that is done 100k per second internally in the PC
nah more like i wanted to build something maybe a friend was interested in but he poor and i got a 7600 for 100$ and a b650 for 100$
and my ddr5 and 7900 XTX has just been sitting doing nothing for months
yep but i was being an asshat but it turns out you didnt get it right away so meh
eh not bad
fair enough
game of semantics sed
sadly im back at 200w vs 100 on lappy
tc1 my goat
im buying ryzen here soon
problem with being silly is that there's always people who'd say that exact same thing and actually mean it
yeah im not that braindead its okay my bad though π
not for daily but im buying it
i needed something zen4 so i can use znver4 packages
nah my /s detector is definitely slightly off today
getting an MPOWER
Do i get a bulova, tissot, or citizen? 
which i will probably not give to my mom lowkey
and a 7800x3d which is going to my mom
Hai floppa
my mom has a 9800x3d here so she's winning
i lied for clicks
yeah and i have a 13900KS sitting doing nothing so whos really winning
on a semi-related topic, I noticed my dad always used crap mice, and I just casually gave him a maya x acting like it's just a random cheap mouse lol
lmao
well my mom has a macbook air with 24gigs of ram
I stealed
basically the same as my i5 6500 and heaps of mining board celerons
i will bin 9850X3D's at work and have a goated am5 combo
quantity > quality
cant believe she was deadset on getting 24gigs
I have a pentium g4400 collecting dust
going to kyoko
im going b4b with my 660 ti
I give my unused stuff away usually
the only thing she does is use google chrome with multiple tabs open, her tech team at work originally got her a m4 mackbook pro with 32gigs of memory because the pro has "fans" and "works well for multitasking"
i give anything good away
DAMN.
i keep things i have uses for or is garbage
like nobody wants an RX 570 4GB thats kinda fawked
I do
go dumpster diving then
oh hey i have one
Yummers
how many broken psus do u own π₯
had 4, gave one away
so we're twinning
quangzhi 2000w 95+ emerald btc psu counts?
as one
what the fuck is a 95+ emerlad
my remaining 3 1kw+ are all AX1600i
i have 3 computers within 2 foot of me
but 1 doesnt work, 1 shuts off randomly, 1 works but the cpu 8 pin is missing (idk how)
HOW
i got it included with some stuff
how do you know it works without the 8 pin
i have 0 clue how they managed to do that
thats only useful for like, some wires
it was with a cpu mobo ram combo, it supposedly worked but they somehow broke it while removing it
eeeegghh i see
0 clue how
goated
did they remove it with scissors idk
i have 1300 G2
the superflower 1200W
and da msi 1kw
i have a bunch of pc parts but theyre all dogshit π₯
same bro same
ok larper
other than a handful of shiz
u dont know the trenches
Got a new mobo for my server a gigabyte b550 eagle wifi6 (atx) gonna replace my current mobo (a random asus b550 micro atx) gonna love the 4-5 x16 slots (not all of them are wired for x16) instead of just the 1 x16 and 2 x1 that I have currently
how many amd semphron combos do u have
mine was that semphron combo
idk the model but solitaire and freecell was fire
i still have 0 clue how to play freecell
my first computer was an intel pentium III 450mhz and... I forget the GPU
the hdd still has the og windows xp insttall with pictures n stuff
yeah im not old
but something was failing last time i tried it

toaster
i mean technically my first computer was like a pentium 3 or 4 but i didnt know how to use it
windows ME thing i got for free at like age 9
and before my gaming optiplex i had a way shitter dell optiplpex 780 with a core 2 duo but i dont count that one
ive only been pc gaming for like 6 years total idk
6 or 7
i see am3 and i assume toaster
am3+ smh smh smh
two sides of the same turd
Finally gonna be able to use more than one dgpu
what's your use case for multi gpu 
hey my h81 pro btc has mobo molex too
shit and fuck, probably
pci e 6 pin on motherboards before it was cool
Many (non headless) vms and/or a small ai model
well, AI can easily use up any gpu resources, so I guess there's never too much gpu for that purpose
My fatal1ty mobo also has a input molex connector
i have no idea if its actually useful
even for multi gpu
id imagine sure if you run more than 4 gpus on 1 board it might be useful but like
molex to sata lose all your data
mole to gpu...
why is it on my extreme4
If I remember correctly itβs to supply the watts for the power connector of the pcie slot before 24 pin was a thing
well you remember incorrectly lol
24 be a thingy
Damn, not surprising
yuo see somethin on that board that looks kinda like a 24 pin?
How do we feel about using this PSU for the little PC?
maybe not cause the res, but you get it lol
unethical, disingenuous, discombobulate
Hey, donβt need to be mean, I just tried to help ><
castles made of sand
if you think that is mean big dawg respectfully ignore me before you cry or something like the last guy who said that
just a lil sarcasm, nothing against you
I low-key lost the big 72 pin psu cable thinging so im gonna eyeball it and pray I get the pin out correct
It was condescending tbh
Silly little pfp :3
floppa how much 800mhz ddr4 do u have tho
Thanksies >w<
0
i will keep milking it
besides my server i have very little DDR4 in general
its so bad im proud of it
The phrasing can be seen as malicious / rude / condescending to some
Even if you didn't mean it
i have a fair bit of ddr4, but itβs all in use
i donβt have a single unused ddr4 dimm
i have 64 in various quantities and speeds
32 will be going in a build rest i will sell
oh 80 actually i remember nwo
I have a single unused 4gb stick
ram prices are seemingly going down ?
64 in my server, 16GB bdie in the am4 oc rig, like 5 other mismatched dimms of various quantities
what prices are you looking at
What language is your display username in? And what's it mean?
I have a single 32gb ddr5 rdimm stick that I bought as a temporary replacement for when I had to RMA my main kit
French, it means door handle
those used to be like 15-20usd a dimm
Had a feeling it's french
Didn't at all think it'd mean door handle
What's the story behind it, madam?
not really that good homie
maybe im desensitized
π
kinda regret it yeah
yes thats what, like 110 US?
It was my summer camp name I was an animator
"hmm why didnt anyone else bid on this"
really not a ll that good especially if you werent IC sniping so they oc well
in my defense prices kinda suck here
That's..
Huh, really interesting
nah they probably suck
Pretty silly x3
I almost bought 128gb for NAS before the prices went up
auctions are unhealthy idk
so on one hand regretting not buying because the prices are much higher now, but on the other hand... the NAS really does not need that much
@azure topaz I almost bought a T14 G2 with a 5650U for $320 earlier this morning
Backed out last second because the money on that is enough for a 30% downpayment on an LOQ
ts peak
And I'd rather not buy two laptops
Looks very interesting
What is the purpose
overclock ddr3
atleast im not the only one ?
biggest auction regret is buying 4 thermalright daisy chain fans
To what speeds
2666 so far
yeah i ended up just putting 5 arctic p12 maxs i found for 15β¬ in my pc
these cost 30β¬... 2 were reverse 2 were normal, only 1 daisy chain connector
Whoa that's pretty fast for DDR3
not particularly
higher than average but its DDR3 lol
you can do like 3000 sometimes although probably not on this platform
& i have no ddr3 experience
What's the maximum you could realistically get on it, even if unstable
these were a cool find for 5β¬
idk what to do with them so i gave them to my brother cause he only had 8gb of ram
somehow he managed to put a ddr3 dimm in backwards
and then there was magic smoke
2666 so far, idk i have no experience on ddr3
the dimm still smells funky
1066 above what I believe is JEDEC maximum for DDR3
That's good progress
theres 1866 jedec
wasnt DDR4 originally rated maximum for 2400 or so
2133, but ram is weird and will generally double in speed before the generation ends
WTF 250 Mbps for $90 a month
there was kits of d4 rated for 5000
That is outrageous
AMDs A series APUs had horrible DDR4 support, they forced RAM to be 1866 which means money wasted on 2133
well if you had to use one i donβt think you had money for 2133 anyways but
lol
First gen Ryzen and x370 mobos had horrible ram support
I used a 9420e for a bit, it is abominal
1866 is a jedec for ddr4 tbh, also i has one of those still, bulldozer cpu but for am4
2008 mobile core 2 duo runs better
idk bulldozer clocked ram ok unless it was just an arbitrary limit
i got my FX 4300 to boot the max ratio the fx board i showed earlier even supports, which is 2400
think i still have the patriot 2400mhz kit around somewhere
its a normal sized computer wym
580 8gb powercolor thing, idk its garbage
displays a picture tho
its not a 4 slot 5090
hey dont hate on the 580
really hate my board, wont ever default to iGPU even without a dGPU
as a gtx 1650 user i got offended by that
consider that you just dont have an igpu
i5-7400:
7400
wack then, you got some gross ass prebuilt?
i dont know if the build would be the issue, they chose a decent board
MSI Bazooka B250M
Even if you reset the bios?
yep
m yes decent
idk my msi z270 board was fine with IGPU
nice job rpcs3
Weird
i had a z270 tomahawk opt boost
also im 90% sure iv had that board before
had the amd version of it myself
rest in peace my granny she got hit by a basooka
LMAO
I might be able to hook up my cooked RX 580 for a moment to see what settings I can change
i bet it works
works amazing
look at what i managed to get for cheaper than a a620
also great bios
i try not to uby anything that cant oc ram/cpu
so i generally always buy Z series or at least B series AMD
well eh more like on par with a a620
aint ram OC possible on all A boards while pbo is what can be limited?
and it's a fuckton better than an a620
well yeah but im not buying garbage
it got all the bells and whistles
fair nuf
you bought your pc so obviously u do buy garbage
true
laugh at this user
3 nvme slots, some nice metal slots here and there and humungus amongus VRM cooling, and i use that for a 7600 that has a wraith prism :3
Nvme prices are so bad the second hand market looking TEMPTING
if you want to see a somewhat weird one, look for one of the asus a320's
they have one that's atx
i dont want to buy an a620 board for a computer that im "not going to overclock" then in a year when that pc is useless for its original purpose go to oc it and get very disappointed
yea i should sell mine before google makes some storage saving thing and crashing the market there too
go to a junk jard and steal old kingston ssds
Need it for my ps5
yuh, i just do not wanna buy it due to most of em looking like shit and well they are quite limited
i once found a pc with in the trash with an athlon 2 gigs of ram and some ahh gpu but it had a kingston ssd, guess what happened to it
yeah i dont like boards that arent full cover or at least integrated io shield
i try to tell people thew same thing
i stole and salvaged it
buy the cheapest board that has an integrated io shield as a minimum level of quality
steal 2 then
most people wont ever need any better
Is it an nvme or a standard sata?
standard sata
ngl idc about that but every board iv bought for myself come with one so meh
also one thing i just thought of
a shite ssd in the process you took
not just for the io shield
every single personal pc iv built myself now has had a asus board :3
sata 2.5
generally boards that have an io shield built in are not garbage
or at least generally very usable
not lacking features yknow
runs all this fine so i aint got nothing to complain about and i still have 65 gigs left
more than 1 case fan header, etc
yuh
the a400 is known for dying out quickly
whenever it dies ill replace it
i used to have 2x 480gb ones of the a400
i put it in like a one and a half years ago
after how long did they go kaboom
up to you
had them in for 2-3 years, swapped them out at 65% for a couple mx500 2tb's
someone else in here using a shit ssd nice im on a wd blue sn550 for my boot drive
these used to be good if you have the v1 and not v2
mine is still going strong at 97%
it came from a skytech prebuilt and i forgot to check lol, i assume its the v2 considering the timeframe i got it vs when the v2 released
that would be the one with the controller swap then
i still have an sn500 that needs retiring
hmmmm
floppa
do not worry about it
Why is it so hot
Hmm dream machine
Think of how hot your room would get π
I want this

But it looks cool, surely itβll cool my room down
eh unifi stuff aint too bad
I got POE injectors π
Iβm frying fries
Panada room faces the sun in California
Lock screen
O_O
you guys think people will buy a motherboard cpu and ram combo or should i sell it individually
People are interested in those
yes
Mobo & cpu even
You would get more separate
Iβd think you get more with a mobo & CPU combo then selling ram seperate no?
i sold my x670e and 7800X3D individually and the person asked to make it a combo, sold it for like 600$ then bought a b650 and a 7600 for 200$
planning on selling it for like 500 gbp its a 5700x3d, 32gb 3600mhz and b550
5700x3ds are pretty expensive in the uk
In the mid afternoon and evening yea
Why I always keep my window blinds closed so itβs always dark in my room
i still need to get to selling my parts
You would for sure have a better time selling it separately, 5700x3d is the second best option for a same platform upgrade for gaming , so people with older Ryzen cpus would be looking for that
ye i bought it for Β£250 a while ago but im not on am4 anymore so im just tryna get rid of it
i just dont have the boxes for any of it sadly
which is why it would be easier for me to sell it as a whole.
https://www.printables.com/model/499374-amd-ryzen-am4-cpu-clamshell-case own or have access to a 3d printer?
Or $5-8 for regular
ill have a look thanks
eeeuuggghhhhhh
I chose CachyOS because that was reccomended to me as the best for gaming
interesting monitor layout
Yeah its 1080P monitor, a 1440P monitor, and then my Telly being used as a monitor
and I think the tiny thing might be it trying to use my index as a monitor lol
this is a silly
This is a trial run of linux I'm not comitted but um I already like quite a few features lol
when i was on bazzite it was the same
lots of features i liked, but i switched back to W11 because i needed the gaming performance
One that I'm excited to fuck around with, my Focusrite Solo, Windows just sees it as 1 input
Congrats man welcome to the open source family
CachyOS sees the mic input and the instrument input as 2 seperate audio sources
you should be able to either use direct audio or "Pro audio" which uses JACK which are both technically lossless
I'm gonna cream my tar heels pj pants lol
I thought lossless was reserved for the macOS fiends lol
Question how is the latency if anybody has ever tried recording music on linux?
i think when using JACK your latency is almost absent
but it uses a good chunk of processing power
Direct just lets the interface handle it
i could also be making this up. i dont know a lot about the different audio processors.
jack
hehehe
oh hi
sup
I 3D printed a case that says up to 4 drive bays and up to 5 120mm fans. But I can't get enough clearance on the bottom for the two HDDs I'm trying to get in the bottom of the case, if I GWR the drives the sata cables don't have room to stick out.
It says can fit the hard drives though so idk am I missing something?
lol you gotta look at this
using the best distro too, totally not biased
and we do not even market it as a gaming distro
btw go to the cachyos site and look athe the user of the console window and say their name
Oh Heck its the youuu
:3
Well just wanted to say I really like your OS so far, if I could somehow end up getting VR to work properly on here for like VRChat I'd be sold for forever and omg the way the audio sources work in your OS as a musician I say thank you
well i do not do any dev work im just a admin and helper
this might be nice to know about
I have an Nvidia GPU π
should work on nvidia
Follow up then because I have an index and not a quest lol
but i myself have not had any luck with VR yet but now rebuilding my desktop pc i will try once more
what happens if you just try steam vr?
Um idk haven't tried it yet lol
But I know that Cachy was trying to use my index as a display lol
but I should try it and report back
yea
Yeah
wayland kde should work just fine
well as a minimum
omg i think wayland is the issue
or not, i dono
seems like people are raving on about https://gitlab.com/gabmus/envision
Ooo okay Ill try it
always fun when something just works on their machine
if its still broken you can try and make a post in here https://discord.com/channels/862292009423470592/1400865136373403708 high chance someone else has faced this issue and fixed it there.
Like its promising that it recognizes it kinda lol
it did for me too even one game started i believe but then it never worked again
but i will try once more soon.tm
Is there a way to make it not need a password every time I want to install something?
Envision says this so maybe thats the issue or something
you using the aur package?
Getting started#
Upon launching Envision, you will need to select a profile on the bottom of the left side bar.
Profiles that are available by default:
Lighthouse driver: Proprietary SteamVR Lighthouse driver with top tracking quality. Recommended for Lighthouse HMDs.
it really should just work
https://wiki.vronlinux.org/docs/fossvr/envision/
Envision# Envision GitLab repository Envision is a graphical app that acts as an orchestrator to get a full Monado setup up and running with a few clicks.
Envision constructs a working runtime with both a native OpenXR and an OpenVR implementation (provided by xrizer or OpenComposite), for games or other client applications. Please note these Op...
I think I might have it also funny that Discord is giving me a quest for MS Azure after I left windows lol
Thank you Sim XD I'm stupid turns out the stuff that was issues was in preferences in envison lol
So much for a mac being reliable, this is the second time in 4 months ive had to bring it in to get fixed (ive had it since august) December had to get my logic board replaced due to the battery ic dying and today it seems like either my backlight or gpu dying (signs are pointing towards gpu)
No I tried to start it and it says I need to install Monado
im looking to buy a standalone microphone as i keep encountering issues with all, and i mean all my pc audio playing through my mic, even when muted (the mute button on the mic -- this was with a steel series arctive nova 1p)
whats a good standalone mic i could get with a budget of like Β£50
yea i read a bit more and saw it might have been needed
there are so many tools
always fun when stuff is split
Yeah I'm learning that lol
try
paru -S monado
thigns are doing things so mayhaps thats good news
I'm so sorry I'm a linux virgin so i have 0 idea what I'm doing with some of this stuff lol
Cough, I might not be a massive amount better
I mostly just needed to use Linux to update my soldering iron but never went back to windows for the most part
I just learnt what I needed, and did not care much for anything else
But I was active and nice enough to be promoted to admin on cachy and from there I have been helping with certain thing
Its certainly tryinglol but for some reason everytime I hit start now it just stops itself
you might also wanna save this
paru -S xrizer lib32-xrizer
no wait
vr is a pain in de ass
tried this page?
https://wiki.vronlinux.org/docs/resonite/crashing/
Crashing & Not Loading# Resoniteβs renderer is known to randomly crash suddenly - this is currently tracked at Issue #5325.
There is also a known memory leak when running the renderer on Linux - however, it is uncertain currently whether this is an issue with Proton or there is something else at play - see Issue #5246.
During VR gameplay, the ...
Thats wher eI am right now actually lol
xd
i just gave up myself last time i tried VR and just bit the bullet and booted back into windows for the first time in months
That migth be how things end up or I do have an extra like m.2 drive I just don't have the m.2 screws, so I might just install windows on this other drive and just use it for VR
@azure topaz what is a βfaster game roadβ i was looking at aliexpress and saw this
a game road duh
Thats my favorite mario kart track
Ah, thanks
Leave it to a furry to have a possible tutorial and they area lso using CachyOS
lol
although the most helpful videos are usually the ones by a random channel with 29 subscribers and a very strong accent
@tidal geyser Their envision looks different then mine
no clue
I'm gonna take a break but if I figrue it out I'll let you know lol
Shout out to the middle east for helping me bypass pixel login
couldnt do it w/o them lol
Arabic keyboard at the time still had the web function button
Thus allowing you into the web to login and become the last login user but that was ages ago
arrived a bit early (Dell RTX 2080-Ti OEM)
rear IO shield was slightly bent, but it was very easy to fix
Yee old plyers, solid gpu. Hopefully last you many years
I love my old turing bby
i'll probably have it until i can afford a 50 or 60 series (when it releases)
Thats fair should hold ya over
definitely a great card
In other news.
I've got a network card PCIE or PCI whatever the short slot it (not full GPU length) it works without this being plugged in so what is it for and where do I plug it in?
But y'all need to remember how nvidia treated gamers
As second fiddle to datacenters.
I love nvidia
If theres actually a bubble about to burst, nvidia is up the creek without a paddle.
Might be usb I canβt recall
Yes thats a usb header.
read the manual, but it looks like its keyed for USB internally. Probably for using the card if it's free-floating in the case and not in a slot
Usb 2
yes ty captain obv
I totally thought it was a thunderbolt header tysm for your ingenuity
Its probably for Bluetooth if you have it? If youre not using all of the headers, no harm in plugging it in.
Does the network card have Bluetooth by chance? What is the exact make / model of card?
Most PCIe network cards with Bluetooth require a USB header be plugged in for Bluetooth to function.
I've got money on it being for using the card outside of a pcie slot
I have never seen this tbqh
Iβve bought a WiFi card before and it did this, itβs pretty neat ngl
I doubt the headers alone could supply enough power for the wifi and Bluetooth modems
that's.. dumb
Power and data i should say.
just pass the bluetooth thru the pcie slot like..
Especially usb 2.0
one internal usb header is 2 ports, so that's about 1.5a@5v
I wonder if itβs to mimic the BT devices as USB
Usb 2.0 is 5v at .5a
I hate this thought
but if its TWO PORTS ON ONE HEADER
as internal fucking usb headers are wired for
Thatβs my first thought
and nowadays the 5v rail is kinda beefed so power can be higher on single ports nowadays
that.. that makes no sense
onboard wifi+bt devices on motherboards dont use usb
and thats essentially what third party wifi cards are
Well unless you have bifurcation on the card and the motherboard supports it, then you don't need it
But if both devices dont have whats necessary to have that. Then...yeah probably
why would it need bifurcation if it's one dev id and 2 devices behind it
its a 1x slot, there's no bifurcating
do yall use tailscale or good ol' wireguard? i use both, tailscale on my offsite OMV, wireguard to just vpn into my network. prob just just use one lol
wifi+BT is integrated into one SOC with two different radios. it's one hwid to the system
You need to separate one slot into multiple with the signals to address more than one device per slot
Like two nvme drives in a card
that.. no
Tailscale
Need to split the signals up
its ONE CHIP. TWO RADIOS.
your board either needs PCIe Bifurcation to address a dual NVMe Carrier card, or the card needs a built in PCIe switch
not TWO CHIPS
Thats if they designed it like that
How would it free float it has a PCIE connector or are you talking if I took it off its bracket?
im always eyeing to rid myself reliance of corp middleman services, and tailscale is somehow both one and not at the same time lol
Lol you do not need bifurcation for a wifi+BT card
Every card in existence is a single chip lol
Also I no idea the brand so how I get manual?
its ONE single SOC driving the radios and acting as the pcie device. there is no fucking bifurcation
Again, thats if they wired the card that way, along with using one chip for the bt and wifi modem
if it was installed in a slot in the case, but not in the motherboard, you could probably use that usb header to run the card
Gpu yiss
thats how they're ALL WIRED YOU NEMATODE
That is how every Wifi card in existence is wired my friend
Tailscale has been good for me so far
EVERY fucking m.2 wifi+bt module is wired like that
The fact you just called it a "modem" lets me know you have absolutely no idea what you're talking about
yeah, bifurcation within an M.2 slot isn't even a thing
Probably does I think I saw Linus on a wan show once say something about he is unaware of a WiFi module that doesn't also have BT but I don't know the make or model of card. I can maybe just mess around and plug into a USB2 header? I have no front IO to be taking up a slot from.
Even on multi lane M.2 slots, there is no bifurcation that can occur, let alone single lane slots designed for an wifi card
@grizzled frost Remember negative braincells?
I'm reminded daily, i work in customer service
Maybe still not sure why it would ever be done like that besides maybe a shitty OEM PC but it was sold as a stand alone card so I can't see why anyone ever would.
mmm
Feels, I dont miss doing phone tech support
because if its an older pc with no pcie slots available?..
most older pcs have an internal usb header, like
Oh so you designed every one of them on the market so you know how they all work then?
No I have near 20 years of experience in this field bro, step off lmao
Sorry I doubted you master engineer of all electronics.
My brother in christ, you have absolutely no idea what you're talking about and it shows
That all companies go to
They are just saying, no biggie
And you act like you know it all
theres these things called technical standards that people abide by, that makes it easy to know how 99% of most products work my dude
Cool
Find me a single M.2 Wifi card with separate Wifi and BT chips, a single one

I know a lot more than you that's for tiddly-winkin' sure
Not all they just know this thing, and a few other things ofc
And a Chinese adapter is definitely going to follow them lol
i'm so stealing the phrase "tiddly-winkin'"