#tech-chat-2
1 messages · Page 514 of 1
Hello based department
https://fixupx.com/rpcs3/status/2053632969357902050
Our guidelines for submitting AI-generated code are now up in our repository!
︀︀
︀︀As for all the AI bros seething on our socials, we're simply blocking you.
︀︀
︀︀Learn how to debug, code, and leave behind something useful to humanity when you're gone, instead of peddling slop.
Quoting RPCS3 (@rpcs3)
︀
Please stop submitting AI slop code pull requests to RPCS3. We will start banning those who do without disclosing.
︀︀
︀︀There are plenty of resources online to learn how to debug and code instead of generating slop that you don't understand and that doesn't work.
RPCS3 has been based for forever
this is insanely based holy shit
shouldnt be tested on x3d bcs its all margin of error
i don't agree, you can see in the dude's post that these same types of results are repeatable
in the other tests he performed as well
i wonder why he’s at CL24
also we need to know the game
I dont think its worth running if you already have a 8000 samples but I dont so I run it
https://www.reddit.com/r/overclocking/comments/1t3jpgu/ddr5_4400cl24_111_9800x3d_game_benchmark_results/ this is the only american post about it the others are all chinese afaik
"Higher" quality pictures: https://imgur.com/a/oLkvDjL Previous post about the 4400CL24 setup. Here are the requested in-game…
Based
also uses less voltage across the board
also my shitty micron kit doesn't dream of running 8k so
"we're simply blocking you"
because tcl below 24 doesn’t exist on the imc
@pure garnet how am i supposed to know what trc is supposed to be with tras at 127 lmao
AI Use
Use of AI tools for research and reverse engineering purposes is permitted. However, contributors are expected to fully own and understand all code they submit. Any communication with the team — including code, code comments, and GitHub comments — must come from the human contributor, not an AI agent acting autonomously.We have unfortunately seen a rise in untested and unverified AI-generated slop being submitted to this project. This wastes maintainer time and, in worse cases, such changes get merged and break functionality for all users. Repeated violations will result in a ban from the repository. Please be respectful of everyone's time.
Pull requests opened by AI agents or automated tools must include a disclosure in the PR description stating the scope of AI involvement — which parts were AI-generated and what human testing or review was performed prior to submission. PRs that omit this disclosure may be closed without review.
If you are unsure about your work, open a discussion issue to talk it through with the team, or reach out to a maintainer on Discord.
It's not even that strict and still the AI bros are malding
you kinda just wing it
intel down up to 10% bois all in
if only cpu prices perfectly aligned with stock prices
https://videocardz.com/newz/this-geforce-rtx-3070-survived-a-lightning-strike-and-fire clearly this 3070 was meant to die
i just heard someone drive by with cool for the summer hardstyle remix playing in their car
made me giggle
living large

fate did a sloppy job on this one
6090 going to be the king of gpus
my apex is a state away so possibly tomorrow
what about 7090
i think i set my trfc too low but i dont have time to test rn
it passes some tests but crashes in cpu stuff so i gotta do more testing
@topaz veldt playtime...
https://github.com/ggml-org/llama.cpp/pull/22673 this branch + https://huggingface.co/unsloth/Qwen3.6-27B-MTP-GGUF or https://huggingface.co/unsloth/Qwen3.6-35B-A3B-MTP-GGUF
How hards the setup para?
And how hard is it to configure for diffrent task? What kinda use cases can one do with Llama and qwen? I hear good things
this specific setup (with MTP) or general ? What platform ?
Windows 11 and in general im not familiar with the programs at the moment
well, what is your hardware (for starters ?)
well, for 5090 you probably don't need the many changes llamacpp does, LM Studio is llama.cpp in nicer packaging, sure, it sometimes trails behind by a bit, but probably enough for start.
So, it's relatively simple:
- Install LM Studio
- enable developer mode and enable server - it will be available at port 1234
- under model search, search for unsloth/qwen3.6-27B and pick at least UD-IQ3-XXS quant or higher (but probably 3-bit or 4-bit quants are reasonable), download model
- on developer page click the blue load model button, in the list at bottom turn on the "Manually choose model load parameters" and then click on the Qwen3.6-27B in the list. At bottom enable Show advanced settings.
Rename API identifier to whatever you like, pull the context length to maximum, at bottom of the options check K and V Cache Quantization Type and change it to Q4_0. Then press Load Model.
Now you got an OpenAI and Anthropic compatible LLM server running at http://localhost:1234
after that initial setup, you just need to configure LLM harness of your choice to use that URL. For example for Claude Code you can either set the environment variables of system, in batch file or in .claude/settings.json (or custom settings.json file you pass through CLI parameter to claude code) with settings like:
"env": {
"ANTHROPIC_AUTH_TOKEN": "does_not_matter",
"ANTHROPIC_BASE_URL": "http://localhost:1234",
"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": "200000",
"CLAUDE_AUTOCOMPACT_PCT_OVERRIDE": "95",
"CLAUDE_CODE_DISABLE_EXPERIMENTAL_BETAS": "1"
}
i need to run actualy stability testing but it doesn't crash in basic benchmarks
Thanks para
with llama.cpp you go to https://github.com/ggml-org/llama.cpp/releases, download both "Windows x64 (CUDA 13)" and corresponding "CUDA 13.1 DLLs" for latest release, unpack them in same folder.
Then you run something like:
llama-server.exe --ctx-size 200000 --fit on --no-mmproj --jinja --flash-attn on --cache-type-k q4_0 --cache-type-v q4_0 -hf unsloth/Qwen3.6-27B-GGUF:UD-Q4_K_XL --host 0.0.0.0 --temp 0.6 --top-p 0.95 --top-k 20 --min-p 0.0 --presence-penalty 0 --repeat-penalty 1.0 -ngl 999
That will start a server at http://localhost:8080 (well also at your public IP's, because --host 0.0.0.0) with that model - it will download it and start it.
You can also try to have k cache at q8_0, i think i got up to like 27GB VRAM used with q8_0 k cache and q4_0 v cache.
The difference at client side is just the port.
this is pretty dookie but not the worst result on hwbot by a couple of pages so for 4400mts ill take it
For other CLI harnesses you need to find out how to configure a local LLM or an OpenAI compatible LLM. For example with Pi it is just configuring JSON file, with Hermes Agent it is literally just picking Custom OpenAI server in the configuration screen and typing the URL
@merry stirrup is there such a huge spread in hwbot yc1b times because of people subbing g1 and g2 results?
it seems like results for people running 8000 are far faster even with looser timings
I need jaykihn to tweet more NVL information
Creative studio install itself to anyone else's phone?
If it has no limits going to be fun
And model choice is up to you. Qwen3.6-35B-A3B is a bit dumber, but still good, but it will be a lot faster. Qwen3.6-27B is going to be better, but slower. Both will fit in your VRAM with 200k context with q8_0 k cache and q4_0 v cache
Ill have to play with it inbetween free time looks neat
@low sorrel
Need help with 4 extra noctuah fans im getting replacing my u12a fans with new ones. I have 0 fans on the top or below the gpu. C700m case has 3 140mm front fans that the front panel loves to choke off and one rear 140mm
dust
Yah im waiting till thursdsy till clean since thats when the fans come
Also power viewer 2 came in and corsair powerprotect
I’ve measured some very bright TVs over the years… but never anything like this. 👀
For the first time ever, I recorded a peak brightness measurement of 10,000 nits on an actual TV, namely the 85-inch TCL X11L SQD-Mini LED TV.
To put that into perspective:
• Most OLED TVs peak below 3,000 nits
• Even flagship MiniLED TVs typically st...
starting attempt 400 of 8000 on the 7800x3d
shit turned into a setsuna chip in the box 
oh interesting, Phillips 951 uses the same panel as LG G6 but will have Dolby Vision 2 support
ez pz 4400 tune asap
buns
it was an impulse buy in the era of unobtainable 9800x3d's
You are a gentleman and a scholar. Now to figure out how to get it working on Linux
literally same thing. LM Studio deb/AppImage for easy way, llama.cpp precompiled download for everything but CUDA. If you want CUDA for NVIDIA, then you need to compile it yourself.
LMStudio doesn’t listen on a port though does it?
one more trick - if you know you don't have a good enough GPU (<24GB VRAM), then you could go with Qwen3.6-35B-A3B and then use llama-fit-params to find out the best command line to squeeze out as much as you can from the hardware, to offload as little as you can, and offload only parts which do not hurt the performance as much
of course it does
I have a 5090, I’m also considering adding in a 3090 I have sitting around for the fun of it.
and in logs (not sure why it gives me the "DHCP failed" IP address there, but you get the point:
to be honest, 56GB of VRAM will allow you to run Qwen3.6-27B at higher quant, or at higher K/V cache quantization, which could mean better precision, but personally i haven't had much issues with UD-Q4_K_XL model quant + Q8_0 K cache + Q4_0 V cache. Had to downgrade to Q4_0 K cache with MTP support, but i have yet to decide if i want to keep it or not (i only got 2x 5070 Ti, so slower than your setup and only 32GB VRAM)
That’s the question: 27B at a higher quant or a 70B model at a lower one.
all "bigger" models are older tho, and Qwen3.6-27B beats a bunch of bigger older MoE models, and even Qwen3.6-35B-A3B sometimes does beat the big older brothers
what's the point of running an LLM locally?
privacy, no usage limitations, lower price (if you got the HW already)
but isn't the output way worse tho?
Yeah no token limits, nobody spying on every query, and it’s respectably intelligent for something running on a PC.
Yes, but they also come uncensored.
i don't think i have ever run into a situation where i would need the LLM to not spy and be uncensored
Before it was common knowledge: Microsoft Copilot used to refuse to answer questions about how many times Bill Gates rode a certain airplane or went to a very special island where people did horrible things.
i kinda struggle to imagine why you would need to discuss sensitive topics with an LLM
Qwen3.6-27B is on par with Sonnet 4.5. Yes, we got Sonnet 4.6 since, and of course there is Opus 4.6 or 4.7; but a free small local model being comparable to half a year old mid-range commercial model is good enough for majority of tasks
usually what you lose with smaller models is world knowledge, but for example as coding assistant that is not as important.
You don’t, but I was giving an example of LLM censorship. If it refuses to answer that, what else do they have censored?
How about how many times he asked Jeffery for medication?
and for example https://x.com/leftcurvedev_/status/2048959358743072808 (and othe videos from this account) show how "good enough" the small local models became
and in case of some visual tasks, the older Sonnet literally failed https://x.com/leftcurvedev_/status/2047331780496109701
Qwen3.6 35B A3B vs Claude Sonnet 4.5
Made them fight on the same prompt.
🌸 "Cherry blossom" challenge
Both models took exactly 41s to deliver, no retries
Oh boy 🤯
Crazier part is that I used @UnslothAI's UD-IQ3_XXS gguf to make it fit in 16GB VRAM...
is that like a css animation or?
lol exactly. An uncensored local LLM can swear and say some really funny and offensive shit.
usually HTML+JS canvas, yes
can open weight models be used just as well for learning?
like do they have just as much info from books?
what do you mean "for learning" ? As i said, the smaller local models usually have less world knowledge, as you need to save model size somewhere. If you want a local model with lot of world knowledge, you use something like Kimi 2.6... if you got like a 8x B200 system 😄
https://unsloth.ai/docs/models/kimi-k2.6 (595 GB model size for Q8 quant)
(plus add the K/V cache on top of the model size)
The local models also aren’t agentic. They can’t google stuff and analyze the web results. You have to do that and feed it in as part of the context window.
like learning programming and stuff
uhm... that is job of the harness 🙂
I’ve never used one, now I want to try it
it can program. You give it a task, it will do the task. If you can learn from that, then you can call it learning.
i mean does it have enough knowledge to explain things?
you connect Claude Code/OpenCode/Hermes Agent to it. or Pi coding agent, if you add the necessary addons.
For example here are the web skills of Hermes Agent: https://hermes-agent.nousresearch.com/docs/user-guide/features/web-search
Also they can always just call curl or they can use puppeteer/playwright to interract with browser.
for programming, usually yes, you can for example give it a stacktrace of a failure and ask it to explain it. Or point it at the project or file and tell it to explain it.
in Claude Code, it usually just calls curl and grabs the page and then reads it. Without a subscription, i only needed to replace the search functionality with a duckduckgo MCP server (which gives the agent a capability to do web search via DuckDuckGo https://github.com/nickclyde/duckduckgo-mcp-server)
in the end, the classic "chat" usage of these models is the absolute basic functionality, and agentic use (via CLI or desktop coding client) allows it to create/edit files, run commands etc.
could be a plethora of things but yeah on the cold side
nobody has any reason to run 1:1 on cold
Ok i forgot what a setsuna bin is
People are crazy
That seems like normal pricing these days no
$1600 MSRP 2022 card though
3090 is over $1.2k if i remember correctly
gpu market worse than 2020
Turns fucking 6 years old in September launches at $1500 😂
US govt bans China from importing 4090 & 5090 GPUs by law, so now they have people buying cards and smuggling them to China that way. Not even joking.
aren't they made in china? how would a ban like that even work?
Taiwan
the banned products are most likely not made in China (which is only 4090, 5090, without the D addons)
Many are assembled in China though like mating the cooler to the PCB
I was about to reply with that, that’s not sane.
This is true, my dad's cousins friend in China does that
Only first party sellers are being watched by anyone. eBay is the Wild West. Nobody watches. Nobody regulates it. The cards go to resellers for export.
this a sign
it must be done
i got the apex otw
i guess that will have to do
don't touch it
Why
not cool to the devs
can you try sending them $30 in cash?
that might work
What would even be the postage rate from Vancouver to San Fran
if it's just a tiny letter couldn't be that much
3 bucks
probs what you pay in taxes for the game anyways
Yeah but those won't be good
Should I even get a amp for my hd560s
if you can already drive it don't bother
How do you define driving it
can listen at reasonable volumes
Yeah I have it at 40 most of the time off my motherboards port
Mm I'll check tonight
yeah that's fine but if you really want one, I think this is a solid amp https://a.co/d/0dScKd4p
The TOPPING E1X2 OTG is a high-resolution USB audio interface with 192kHz/24bit support, ideal for live streaming and recording. It connects directly to iPhone/Android via OTG, supports dual mobile streaming, and offers real-time mixing with 12-channel control. Features include ultra-low-noise mi...
I run my headphones at 4% with my overpriced sennheiser thingy
I have not
https://jdslabs.com/product/element-iv/ @low sorrel you should get this fr fr
We’ve reimagined our flagship amp+DAC to create a listening experience you will notice and appreciate. Element IV operates on our third-generation XMOS hardware and firmware, blissfully integrated with our latest discrete optical encoder. You’ll hear perfect channel balance down to the black noise floor. Automatic gain helps you focus on lis...
Perfect to pair the the hd560s I bought for <100 USD?
oh yeah big time
bare minimum tbh
it comes in silver too now
yall ever heard of the rtx pro 500 blackwell?
it's an rtx 5050 die kneecapped to 2060 level performance
slowest 50 series card that i know of
big dial
fuck i love that so much
like 90% of the reason i bought it
it's so nice controlling volume with it
Supreme
$170
after taxes, iirc
for that price it's decent
any mroe than that and i feel like it's too much
actually taht reminds me, i remember asking linus if they had something at the sale if they had something
🤔
can't remember what it was though
they were also selling ram for alright prices but i wasn't too interested at the time
warehouse sale
LEGO Icons Lord of the Rings Minas Tirith coming June 4 for $649.99. Has 8,278 pieces, making this one of the largest LEGO sets ever. https://t.co/AQvUxjBMfn #ad
LEGO LotR Orc Grond GWP with the set: https://t.co/m2qNGEOyrH
@edgy laurel
ONE LEGO SET IS 650 dollars
It’s almost 8300 pieces
so expensive
7 cents per piece
don’t like lego since they started putting different colored parts inside builds
it's very expensive to setup unique molds for individual lego pieces
if they dont plan on keeping the set in production for a long time then it has to be expensive to recoup cost
Lego racism
sus
simply only make mass produced sets
7¢ a piece is still pretty on brand for pricing too
i have the daylite as my purse
they just make the sets less and less reusable
yeah that seems like peyton things to use a backpack as a purse
Throw the redesigned Daylite Sling over your shoulder and go. This ambidextrous carry now features an easy access cell phone sleeve on the shoulder strap, dual side zippers and new daisy chain details. Breathable, soft spacermesh material offers next-to-skin comfort. It's convenience, style and f...
my fam was too poor for themed sets when i was a kid
i just have a bucket of random basic pieces of various colors
real
because instead of filling a set that’s grey, white or black with same parts as outside, they decided to put like orange, purple, green bricks inside
where you don’t see
i was just too poor for legos
I didn't even know lego came in sets until i was like 8
but someone even where you do see
Any set that met an unfortunate death went to the bin
like when you have rooms
A lot of sets
i had more fun playing like that anyway tbh
My uncle is a crazy collector though, buys two of every set he gets, one to preserve and the other to build
ngl i prefer building my own stuff
think of something, make it, play with it, destroy it, and put it back in the bucket
Playing is more fun with creativity, but I like having sets now as showpieces
kid lego vs adult lego
can you even still buy basic lego anymore
yes but barely
I must use the kragle on all my sets
the hell
Where’s the FIFA peace prize LEGO
how much would lego cost to 3d print
$$$$
$$
$$$$$
Not my collector sets
3d printed lego kinda sucks
Lego has suprisingly really tight tolerances
That's why offbrand lego always kinda sucks
They’ve had a while to perfect the process
well yeah
im utilitarian as fuck
The tolerances are required for a good experience
Agreed
didn't they actually have to loosen the tolerances at one point
They added in looser tolerances to the design, they didn't reduce the tolerances
IE making the hole a little bigger, not making the tolerance on the hole wider
yet you still can’t seperate two flat bricks when they stick together
Yeah you can...
good luck breaking your nails
almost like there’s a tool for it specifically
wym weak 😭 my nails literally get destroyed and the bricks had dents in them
That explains a lot
probably ate a bunch
fuck off
3060
actually never did
Grew up on legos lol.
my mom has a better gpu than you that’s all i’m saying
oh and hot wheels
I grew up on legos and snap circuits
for now
i’ve also had these
Still have it in the closet in a box so all creative freedom if I am willing to sort the pieces out.
for the foreseeable future
what gpu does she have
a 3070
at times those were even more hell to separate
@frank cypress
if you stuck a red knob onto a metal rod, there was a 30% chance it would never come off
like literally
two pliers, i asked my mom to help and it refused to separate
@turbid locust hot wheels > lego
tried hammer as well lol it moved a little and that’s it
no idea how it could get stuck that badly even
Agree
I had more fun with lego when i was a kid over hot wheels but that was bc i just had the cars, not anything else
why even buy hotwheels atp
Bc my family was poor back then
We got what we could
we even had a brick separator but some of the flat bricks just held too strongly to get a grip
actually this shit was peak
Gokarts is peak
i think it happened with older bricks as the plastic ages
and then it broke
and the battery started smoking
Super glue the bricks
and then i was sad
they already were
by themselves
oh you meant that yeah
Swap the battery or mod the car, twin engine and twin battery
man i looked into gokarting recently
and damn that shit is expensive
Its fun af
i used needles
Yeah, you would love it tho, its fun af, thats how i got into minibikes
then it was easy with nails
at the time we didn’t have anything else other than needles
it's like 200 bucks an hr
now i can use stuff like a slitted screwdriver
Better have a helmet on mr
Oh no, i mean having your own, i own and built mine
oh well maybe i could’ve used cans but i was affaid of cutting myself
Yup 😎
yeah but wouldn't i then need a track
Nature
Depends where u live, mine is for offroading and i live in the country
I do not infact live in a rural area
there's a whole go karting league by my college and none of my friends wanted to do it
You can do parks or sidewalks
Sounds expensive
it was like $180 for 8 races or smth? Not terrible
and it's races without kids or smth
Oh I thought it was a build your own then race in tournaments
Its an awesome feeling to own your own gokart or minibike and being able to fly at like 50mph
ah nah
it's a pre determined track with karts given
24 hrs of lemons would also be fine though
https://www.thedrive.com/accelerator/28150/downhill-power-wheels-racing-for-grown-ups-is-the-absolute-pinnacle-of-motorsports I wanna try this
Rent a storage unit 😆
Race mowers is also bad ass
TC2 24 hrs of lemons when

if only
What's the lemon budget
500 bucks + safety equipment
its just a lemon, what could it cost, $10?
Everyone cheats on the first part
yeah what about sales tax
Damn if it was $1k my daily driver would count
You can lie
I started on a bench lol
Bribery is allowed under the rules
sell me your car and then ill sell itb ack to u for $500
Give me your car and ill sell it back for 5k
"back in my day we didnt have autism" -> race lawnmower.
Does this meet the fia safety standards
Na, thats a gokart, race mowers is an altered tractor mower
Snail could fund it
Safety standards and gokarts dont go together lmao
Only safety is what you are wearing
You can build them to meet it, but most custom ones normally dont unless ur building it specifically for racing
just wear a helmet and leather clothes and youll be fine prorbably
LS swap it
The LS is probably bigger then the entire go kart lol
Ill install a knife that will point at my chest in the middle of the steering wheel so if i crash, i deserved it ☺️
make it a half-track and put a claw on it
Lol
No pain airbag! Brilliant
You can also sell parts to increase your budget
My hubcaps sell for $60 per
Sell them to me for $-2
lets ucking goooooo
You didnt download the free copy?
no wtf i bought it months ago
You pre ordered a game?!
well yeah its forza
140GB is fucking insane
hehehehehehe
True microsoft Forza team needs the extra funds upfront
A 512gb nvme is $88. Dcs world is taking $88 worth of SSD space 🤯
Fuck ai
i want to grab that moza clutch and turtlebeach hpattern so bad now
500 of 6000 would be 1/12 right
Yes
12 512gb sdds
I also bought a new keeb
What anime board is it
G515 LIGHTSPEED TKL gaming keyboard is the ultimate choice for competitive gaming. This low-profile keyboard offers a next-level experience with its sleek design and engineered performance. Tri-mode connectivity includes wireless keyboard mechanical options with LIGHTSPEED wireless, Bluetooth, an...
a non weeb board impossible
A Logitech low profile board without laser ablated ABS keycaps? I am shocked
The weeb model is cheaper 
this tbh
with the sole exception of buildable model kits
Oh not that weeby
lightspeed = dope tbh
159 for what 😭
yeah its very minimalistic
The keyboard Peyton linked above
I wanted a wireless solution so I can just grab the keeb and mouse and go to the tv den across the hall
lightspeed is actually insanely performant
I did some jankery for my tv wiring
Hmm
fiber hdmi, 2 usb extenders with retimers, ethernet
200 foot extension cord to your neighbor's outlet?
Don't you own the house so you could drill holes?
😭
Yes but I really don't want to deal with that
Phishing wire is infuriating
also considering I have no attic space
its ceiling drywall, framing, insulation, more framing, and then roofing
no gappage
Crawl space ?
yeah but I'm not armoring things just to put holes in the floors
I'll do that in my new house
Steal copper pipe and use that as free armoring

Been looking at land and researching home manufacturers
Lowes 2 storey shed
overhead today
Lookin' at 1000sqft on 2-4 acres
pavelow lets gooo

If anything I'll be going with them because they have a local branch and do custom stuff
bassbone would custom loop that whole shed
Because I could easily have my detatched garage and cabin
Wait nvm can't shop at Lowe's anymore they got them flock cameras now
No SM fiber to the shed no care
Rather drive 15 mins to home Depot vs the 5 mins flock lowes
No it'll be more of a cabin shell than a shed
and yes there will be fiber ran to the house. that's gonna be mandatory.
I've already budgeted 10 grand for that
What about that good old natural gas
shed pull
natural gas will also be another large consideration
because I'm gonna be considering more remote forested areas or the high desert in eastern oregon
so it'd be nice to have a backup generator that can run on NG or propane
also dual-fuel water heating and hvac
damn
Our natural gas saved us a few cold winters nights when our power or heater went out since we have gas fireplaces
they just installed a bunch in my home county
dude you're getting a dell
My local Lowe's had flock for like 2 years or more
straight out of tron
Huh. Looks like they do but not in my part
you see the news about the water main burst?
is it polluted with human excrement
nah, a 42" water main burst sunday night over here, and people lost water / were put on boil notice
jeez
In Auburn Hills, yeah?
yup, luckily I am only on boil warning, but that's crazy
downtown Auburn Hills was flooding
and having a 42" pipe burst is nuts
idk how that happens
yeah, they are fixed the pipe today, and are gradually increasing flow and letting us know about it
Do basically one of these things from where I am. Yeah those are almost non-existent in my part of SH

chump change
pocket change for a man of his status
Well the system I just implemented I spec'd like 200 of these lol
that's what, a casual $3mil?
How many are you taking home to test
Totally not all of them
meow
I love using my UPS as a heated foot rest
just gotta put padding on it
how noticable is purple blacks on a QD oled.
i like my room with a big of ambiet light
I see comparisons on youtube and it looks pretty bad but comments say its not as noticable irl than on camera.
depends on the panel gen, but I would say somewhat noticeable
I ran one for a bit and could tell it was there
i'd like a 4k 240hz but they dont make a new one with woled, so do i just wait for a new panel to come out.
I don't notice but I'm in the basement with the lights off
you would be
Coopa might just be gollum
yuh
think ill just wait then
yeah that's what I would do, unless you find a good deal on QDOLED
Price isn't really an issue.
I just want whats best for my use case.
These monitors all eventually go on sale anyways.
did asus only make like 5 z790 apex encores
like the name suggests yeah
“encore” they had no reason to make a bunch of them, the board was a refresh for specifically ocers
Was not expecting @frank cypress to look like that
i was expecting you to be a damn mouse
isn't the white apex even more rare
muehehehehehe
go fight
yes because they blew up. different reasons LOL
should have gotten a TC2 meetup selfie smh
define blew up this is news to me
iirc they put a cap or something the wrong way on the boards and they partially blew up
it was either z690 or z790
😭
fuck the thing im glad about not dealing with that PC anymore is that it's lowk kinda fucking heavy
yeah this was a whole production run issue too
it was way heavier when it had the second rad but fuck my back is too old for that shit
buying a z690 apex to struggle getting 7200mt on a 14900k
what did you sell him?
and then the board killing himself
😭
the whole dang pc aside from ram and storage
plus all of my watercooling stuff
2080 Ti?
4070
okay not whole, but whole revision type shit
oh holy shit
5950x/x570 godlike/4070/1600w + all the watercooling shit
might display 4K idk
pack watch for the 3060 bro
did he get something other than a 3060
4070
3060 ti
bro has made it
Lol it would be really funny if he got a 3060 ti
it only took me buying it from kyoko for him to buy it
4070 is what I consider to be the bare minimum for 4K
Yo where the fuck do I even grab this case from
good question
lift it with your mind
good luck 🫡
cross continental trip for this gpu bro
fuck ass case fr
the TC2 hardware sharing is so funny to see
outstanding move
salvation at last 🙏
i think it’s fun you guys also give da good deals
haven’t paid ebay pricing for hw in so long i’m spoiled af
We can't bully his 3060 now 
yeah good deals and it's cool to see how many people get the same piece of hardware
i can bully his shit 4070
meanwhile the mpower
@frank cypress did you sign the 4070 for jerry
no 😔
RIP
but kyoko didn't sign it either
How much was the 4070
6$
$450 but then ups bent me over and fucked me
floppa mac and i shipping the same crappy motherboard cross country multiple times
Cad?
so we don't talk about the end price
literally 3x or whatever
yeah
the only known signed GPU is Mac's 5070 Ti given to Floppa afaik
so far yeah
special edition frfr
so in TC2 land it's a collectors item
i want you guys to sign things when they go to me tho
well biggie signed the 6700xt? he sold to iono iirc
it is also surprisingly decent
I gotta find some HW to sell first
oc’s fine on mem and has a good core
i’ll pick up a tv from you like a year from now don’t worry
I gotta start practicing the akn signature
gotta mog mac’s
well biggie might've drawn a phallic object on the gpu instead of signing it, i don't remember
LOL
that sounds like biggie
@turbid locust see were special
it's also radeon so it's value is like -300$
true
fr
pay me to take it off you
last i heard iono has a 4070 from random or something
and to think i almost bought a 9070xt
I'll buy AMD products just to send videos of it taking a sledgehammer
i’ve figured out the amd method though so i won’t
support the gen really well until the next one comes out then forget about it and make the drivers 3x worse somehow
i almost did too
man i miss random posting shit like tossing GPUs out of a window or washing it in the sink
then kill the card 2 years later
(drivers)
RIP random (he's not dead just gone)
i’m still mad about r9 HBM lineup
never forget ✋😔
@frank cypress how the hell do I open the case
You guys did not own RDNA3 and it shows
take off the top panel
i had a 7800xt mac
then you can slide off the glass
that’s close enough
don't look in the back, it's an utter mess
Oh true I forgot my fault
Mr 7900XTX over here
power cable is in the box of custom loop shit
Not as iconic as 5nail breaking his ram modules
@low sorrel when are you getting a real 4k gpu with 16gb of vram
yeah it sucked im aware
that's a good one too
i owned rdna 2 but only play an mmo from 2013
so bad i swapped back to a 3080 from it
LOL
Wait what that's a first I am hearing about this
My 6700XT was nice, but only because I was at 1440p and didn’t use upscaling lmao
I had a 5700xt till I got my 9070xt 
To kiss?!
when he gets his concealed carry permit
https://www.youtube.com/watch?v=BciNjetLmLc @low sorrel
MetallicGear NEO QUBE case w/ dual system design - review
Even though that the MetallicGear NEO QUBE on the front side resembles and basically is identical like the Lian Li PC-O11 Dynamic, they manage to change it up a bit and just enough to place a miniITX motherboard with a capture card above the power supply. In other words, an ideal chassi...
he my goat i would /s
Don’t need a permit in my state 
btw the case is a utter piece of shit 👍
oh even better
If I meet mac, we'll be trying booze
Yep a video of him destroying his ram sticks with Eminem playing in the background

So you mailed trash?
i hated it the entire time i had it and wanted to upgrade to a proper o11 the entire time i had it
sure bro i’d get krunk with mac
idk bro whenever I’m in ur silly ahh state?
i mean tbf the case is probably the worst part of the build
or vice versa
well, spec wise
you, yoko, ennui, and akn i gotta see you guys once

we will see where I end up after college
my bedroom
Debt
YOOOO?????
that's obvious no need to state it
Massive debt with no jobs hiring?
Everything gives you debt
guys will my masters in creative writing pay off in the long run
Could help you get a job as a substitute teacher
my roommate of course
oh ofc ofc
i guess that'll work
in debt till my 60s
oregon intel job 🙏
Wonder when I will get into debt 
we’re all gonna bunk up with carter
whenever you let me max out your cards
studio apt 1 bed like the migos
I'll help him pay his mortgage if it means cheap rent
mow the lawn too
shii we’ll get that non existant basement
dig it out ourselves
boutta show up on mac’s door
step like a lost puppy
Floppa is there a world where my H24M does really dogshit tRFC like 180ns
if you get unlucky yeah - but it’s to my knowledge that trfc is based off electrical design characteristics of the IC so it shouldn’t vary that much
consider heat with anything RFC related perhaps
kentucky is only like 5 hours from me
When I tested at 177ns and max tREFI it failed mega fast
like fast enough that the sticks couldn’t even get to max temp
I should visit family back home at some point. ||Luckily the biggest hurdle has been taken care of as the previous regime, I would be jailed/disappeared if I were to land at the airport. Just a matter of when that day will come.||
yeah wow maybe you just have unlucky dimms
im going to break my monitor
i swar
the 7800x3d error'd again
😭
THANK YOU
i know some ball
I never tried to daily these dimms so this the first time they’ve ever seen Karhu. I set 160ns when benching and went LMAO
also this kid in high school in one of my classes had that as a stim and abused that shit
i have tried in order:
cleaning the dimms
reinstalling the dimms
changing the dimm order
vddio
vddp
vdd
loose memtryit timings
the chip just exploded in the box i guess?
like an alt with a 2 second cooldown he just spammed it
this makes sense yeah. probably just meh sticks. not a huge issue for daily although zen
It’s just mad weird that the difference between 8400 running vs erroring in 5 mins no matter what was swapping ICs
should def try and go for it in like a few years, maybe when oil goes a little lower lol
Shit would not run on H16A
NGL it feels weird that I might actually see my relatives again.... IDK it's just a surreal feeling as I thought that the day might never come but here we are
h24m is lighter on the imc this is known
yes but that much lighter?
i expect 9266+ on h24m daily stable on my system
plus one ratio? didn’t it do 8300?
yeah u right
i don't because intel becomes old and confused at high frequency
fucc it we slam 1.55v VDDIO when we get home and try for 8600
actually I need to see if H24M works at 8400 with less VDDP
Buildzoid told me it was fine but 1.2V VDDP makes me moderately uncomfortable
im gonna one up bro with 8800 fr
i haven’t had issues
This is all because 5nail got 8300 going
yeah i’d expect probably 1 ratio from h16a -> h24m
He’s on H16A running that so maybe he needs to swap too
Although he usually has better luck with H16A
iirc his h24m is wacky
His ARL ran shittier with H24M
isn’t he cudimm though? coulda been it
i dislike them heavily
iirc every score with cudimms is in bypass mode
With that being said my 270K liked H16A and my 265K liked H24M it made no sense lmao
I’m also cudimm
we*
mac did you see my ultra gamer 4400 tune
naw
it wouldn’t surprise me if the z890 apex was intended for h16a gskill just like the ocf is
true
because yeah, the ocf sets terms automatically for gskill dimms always
hnmmm jerry is awfully quiet, i hope he got it figured out
at least last like bios ver 2.20
i just copied some guys timings off reddit and set my own primaries and they worked
my apex is getting my gskil h24m udimm
are you using cinebench as a ram stress test
no
it was just to verify that overall system performance hadn’t suddenly taken a dump
lowkey didn’t know you had a 9800x3d
is that your high score?
the cinebench wasn’t really relevant it was just the screenshot i had
not exactly but my cooling isn’t that great so i don’t really get much higher than that unfortunately
i’ve only ever gotten around 23k on this chip

what’s on it? tower cooler?
old abused 280
i have a 3080 so upgrading my cpu cooling would be a complete waste of money rn
one would have to agree
a phantom spirit would probably do better, but if your aio isnt loud then it doesnt really matter
missing out on a few percent of multithread performance is irrelevant
yeah
that’s what i never understood
people will run cb then ask if their cpu is performing right for games
like dawg 😭
benchmarking cpu rendering performance lmaoooooooo
i ran a 5800x under this aio for like 2 years with broken airflow pattern and it ran at like 90+c constantly
so i’m assuming the condition of this cooler is not exactly optimal
the water hit over 50c regularly
i use it for that cause that's literally what i use my pc for sometimes but it's not actually relevant for most people out there
5800x is a very heat dense chip
i think 60 is where you have to be kinda worried of permeation
i’m sure*
shoutout before i learned how to stress test that i just ran my 5800x3d at like -50 or some shit
source - asetek lol
yeah but just saying it’s not operating at 100% and also is just an older cold plate design
yeah
didn't even check with r23 or anything i just opened my game
also i can’t verify it didn’t go over 60, very well could have
and cranked it till it crashed
like that one dude who ran -36 on his 9800x3d
the evil urge
my cpu hits like 30% usage and 60-70 watts in almost any game i play so it’s not gonna matter until i upgrade my gpu at some point
67
is this for the ceiling viewing
nah i just project it high up on the wall
computer
i can't do ceiling cause i have shitty asbestos popcorn ceiling
Can I eat it.
what is your question
ok idk if it actually has asbestos but it is popcorn
you can get a LG B5 OLED for $600 USD rn
Nothing I just wanted to flex.
That's definitely on the priority list. Sunday night we had a family outing discussing this actually. It will be years in the making but it's definitely something I have both been wanting to do and also something I never thought would happen. Now with the Assad Regime gone it might actually happen 
the most power i could get it to draw regardless of undervolt or whatever is 145-150 watts in cinebench, i dont know what non cooling limited chips hit but its fine for me
problem is my parents will probably kill me if i drill into the wall
which is why i've been going with projectors 😔
in a “room” of people with similar or better specs rn lol
super tall stand?
ya
a projector is one of the only types of display that i havent had in my house before
i mean i guess that could work, but i also kinda appreciate how big the picture from a projector is
ceiling projection seems awesome for lying down in bed but i also have that popcorn nightmare ceiling
Lets play league of legends.
yeah fair enough, B5 is 48" so not as big
same actually 😭
I've got a flat ceiling lucky, but I just do TV at the end of the bed method
are the B5s suitable for pc use like the c series?
yah I would say so
ohhhh 🤔
i have a flat ceiling
ouuuu shiiii 🫃
i'll add it tothe list
4K 120Hz 48" OLED
i’d like to add smart sensors to my home
@frank cypress where are the screws for mounting the radiator to the case
like door open closed
window open closed stuff like that
i put them in a ziplock bag for the ones i had left
two TVs in da room
i think those small stubby screws are for the rad
the only homewide upgrade im doing is faster wifi points at some point
Hey nice desk bro
and the long screws are the ones with the fan alr
matter enabled, specifically matter enabled are the peak setup
Low key I might have already lost 4 scres
even with a wired backhaul (which nukes my sisters speeds) we only get 500/500 over wifi vs the gigabit we have
oh thanks boss, I wonder who recommended it
i don’t live in an amazing area so it would just make me feel better at night
oh yeah cause we lied it down flat lol
noted noted
& sensors are pretty cheap
it'll be fine
matter is the Apple homekit stuff, so it's pretty well integrated with automations and Siri
this looks so fire
ooh okie
my mom does have android though does this matter
Ah yad your loop has an air bubble
hmm?
typically a matter device also supports google home, you'll just have to look for products that do both
There's an air bubble in the res lol
okie
thanks boss, could use some more decor but TVs are enough of a showpiece
nah i couldn't fill it more without leaking iirc
my favorite thing about my matter lights is that they turn off and on when I leave the house
or spilling it everywhere
ah cause it just uses bluetooth(?)
this is simple and smart
i assume it just connects to your phone every once and a while to see if you’re there
the pump is at the bottom anyways so it's not like that'll die cause of it either
I think so, I use an AppleTV as a smart hub too
oop dinner
uhh I assume my 850W PSU would implode if I added a 3060, right? (also dont worry, I know I wont get more FPS in games or anything like that)
you'll need a a Matter/Homekit hub, but you don't need an AppleTV specifically
oh wait nah i need one for the proprietary ness
special radio signals and whatnot
yeah okie sounds good
https://matterhubs.com/ ah this looks like a good resource
Buy all Matter Hubs, Matter Bridges, Matter Controllers, Matter Devices, & more. Add any Matter Devices to any ecosystem, like HomeKit, Alexa, & SmartThings
i legit just found this
okie okie i’ll bookmark that
yeah i’m gonna do storage on the box and then buy all that stuff
oh hmm I am also learning a lot
cause the ultimate goal is also surveillance but it can wait
yeah I really just wanted my lights + Siri integration
not having to go into some 3rd party app to turn off my lights when I could just use siri is peak
yeah there’s some freaky people here no lie but it’s alright
fire. incredible suggestion
just gotta outfreak them
idk man sometimes even i’m outmatched
damn that bad
😭yeah bro
The second rad is probably going to be just for show lmao
I cba to drain it right now
Raise a floppa ahh username😭
jerry are you delaying ddr5 for the foreseeable future?
brother akn im using the taxbook pro for something other than taxes right now, this is an unprecedented and momentous occasion
my name has been floppa for like the entirety of my time on this platform
most of my online aliases have something to do with floppa
Who’s this guy
i just like the cat not roblox
(im just using it to watch youtube and text in here)
LOL
youtube on the crackbook? why
don’t you have a million better displays in your home
holy shit
i was autonomously presented with an itch to update the taxbook, and then use it in bed
interesting
oh I updated my iPhone today
that is the upside of laptops for me i have no use for one otherwise
now apple maps has ads
bro i’m still on ios 18 💀😭
nah the macbook display is fantastic
damnnnnnnn bro, cavemen operating system
i did not like 26 and my phone is old so i figured it better not to update anwyays
yeah 26 is not great I just like staying up to date for some reason
if i didnt have an iphone 17 pro max then i would also be on 18.7.829e1weudshas or whatever the latest patch for that is
i have my 13 pro max on that, and i think it looks nicer
yeah that’s what i’m on
they really need to let us disable liquid glass
liquid ass
i dont think thatll ever happen, we'll just have this until the next redesign
liquid glass is nice and stuff for the wallpaper, camera app
it doesnt make me mad or anything, im not a raging anti-26er
liquid glass is actually good on TV os though
same me neither but for my personal phone i think it’s a smart idea to pass
i just think that the previous ui design was better, especially the more squarish corners
12pm, thing is ancient antiquated tech
yeah it's not deal breaking or anything, it just seems half baked to me
yeah i think youre making the right call
likely
18.7.tenmillion is very polished
i have a 13 i’m going to repair but it’ll probably be used as a camera
(the webcam clearly needs upgrading)
I will upgrade every 4 years as god intended
(it’s to have 2 camera angles)
i only upgraded early to change sizes
bro i just wait for someone in my family to not need a phone anymore realistically
not a mobile computing kinda guy
like 95% of my screen time is a at a real desktop like god intended
i used to buy a new phone every 6 months but now i stopped caring
Same that’s why I just wait 4 years
tc1 is awful
👀
Yeah I try and use my phone as little as possible now
it’s not even that i think all people are “addicted” to their pgones
i just don’t have a large use case for computers not at home
if i want to read an article im not doing that on my phone i dont hate myself or my eyes
I just try and use it for staying in contact with people now that’s about it
Texts + phone calls, occasional email
And then just generic phone stuff
it is half baked in comparison to ios 18 which is not hugely different from ios 11 (launched with the iphone X)
we had the same sort of design and controls for a long time
How you handling with battery life on it?
Lol I still see people complaining about it being terrible, but was comparing it from ios 18 and both are unchanged on my end
Yeah it seems like they just wanted this huge redesign but came up with nothing
At least with my 16 and for a bit I tested it on the 12 I use to main.
i didnt realize that we had the same design for that long until i just went back and looked at each version
ive only used ios 26 on my 17 pro max so my perception is skued, but this phone has the longest battery life out of anything that ive experienced
my 13 pro max on ios 18 was dead by the time that i got home from work, i was used to charging it mid-day
To 80%
I charge mine every day but to 80% max
i charge mine
Just over a year of use now, still 100% battery health
92% for me
i charge mine during lunch, and while i sleep
wow mister min maxer here
Dropped to 97% for me lol. I don't care otherwise
What can I say
that 12% is crucial for more reels
My 12PM dropped from 100% to 79% in the first few months and stayed there for the next 4 years. Idk. I think it was defective.
mines cooked
im not even gonna show or say just know its cooked
My old 11 was shot
i replaced it when i got it 3 years ago but now the health down again
I got some other shortcut they analysis to a bigger detail but not bad for the 12.
There’s already a fuck ass mosquito in my house man
oh right it's skewed not skued, idk why i used the medieval spelling
You wish for a simpler time
What is this?
I’ve never seen that before
he yearns for monolithic cpus and no such thing as infinity wires
Neat
@edgy laurel mf just ripped overhead running no lights
i gotta learn from uncs before to stretch before i go into work
Gives you a bigger read out on the charger used, battery temps, and a logged history of mah drops over the months.
my back hurtin
doesnt help that i probably look like a bridge goblin while building puter or fixing puter
You just gotta get bigger
nah its not the lifting i just have terrible posture
i have to not do that
heaviest computer ill ever build is probably 80 pounds and i definitely got that
Back extensions at the gym help so much
noted
lock in season
I gotta get back on the grind, I took the week after finals and this week off from the gym
repost but i’m mildly excited about this board potentially having copper heatsinks
i hope it looks cool
I love the look of copper so hopefully it looks good
I got my unc shoes on
cannot wait to see this irl
Missing the 4090 part of it
I’ve got a pair and they are nice
Cozy outdoors
i like my feet warm
i have hardwood floors so theyre always cold so mocassins would be peak
See I don’t like wearing them outdoors cause I don’t believe fluff should go outside lol
and i can walk outside in them which is even better
I’ve got an indoor and an outdoor pair but I just never wear the outdoor ones


