Previous /sdg/ thread : >>107831056>Beginner UIEasyDiffusion: https://easydiffusion.github.ioSwarmUI: https://github.com/mcmonkeyprojects/SwarmUI>Advanced UIComfyUI: https://github.com/comfyanonymous/ComfyUIForge Classic: https://github.com/Haoming02/sd-webui-forge-classicStability Matrix: https://github.com/LykosAI/StabilityMatrix>Z-Image Turbohttps://comfyanonymous.github.io/ComfyUI_examples/z_imagehttps://huggingface.co/Tongyi-MAI/Z-Image-Turbohttps://huggingface.co/jayn7/Z-Image-Turbo-GGUFComment too long. Click here to view the full text.
>>107844919because he's doing a rare on 4chinz and thats being blunt honest in a way you cant handle. move aside nerd so someone that can help him can.
how are zoomers STILL this clueless about technolo/g/y and how it's used to spy on you in the big 2026?they put their head in the sand and pretend none of this affects them or even exists. https://files.catbox.moe/cmcsap.mkv
this thread was BUNGLED
>>107844960it's not exclusively a zoomer issue, it's more of a normie problem
>>107845604hell, i don't even get why do millennials like to pin every normie thing to zoomers? i mean i am a zoomer, sure, but it's not like that automatically disqualifies me from ever dealing with tech, in fact i am the first person people informally ask for tech support, and the people in question mostly consist of people who are not of my generation
>>107844960>>107845246they probably already have lasers in space (or use HAARP) to send some kind of non-ionizing radiation or neutrino through our lunix and windows computers to feed into a mossad AI
>>107845678Because Zoomers were the first gen to grow up with iPads and smartphones, and there have been numerous articles written about how younger people who exclusively use mobile devices are totally flummoxed by anything even slightly deeper than surface level traditional desktop UI, let alone understanding how anything actually works on a technical level. It’s generalizing to say “all Zoomers are tech dumb” but it’s based on the fact that there are simply more layers of abstraction between how we interact with devices as opposed to how things were for Millennials. Keep in mind that iOS devices didn’t even have a Files app for a long time, and the iPad only recently allowed for desktop-esque windowing of applications; that’s the sort of stuff I’m talking about. What might have used to require inserting a CD, opening the D: drive, double-clicking an EXE, and going through an install process is now just a web page. Software is just less difficult to use, and therefore doesn’t necessitate developing tech literacy even to the limited degree that it used to.
Why has no big brand succeeded in the e-ink tablet space?Huawei and Lenovo tried and abandoned their productsAmazon has the Kindle Scribe series but it's AmazonBoox has nice hardware and features but typical Chinese brand software longevitySuprenote is mehRemarkable is trying to be Apple too hard
>>107843954it's better.
>>107844302Anything more specific?The advantage I see for Neoreader is that it's more directly integrated into the system and therefore less ghosting and faster page turns. You can refresh the screen with a single tap instead of needing a gesture.
>>107826750>software longevityFor my readers this means nothing since its purpose is to reads books
>>107840421I bought a boox poke for like 60$ four years ago, works great
>>107826750don't have one yet but viwoods seems like a decent middle ground between supernote and boox, besides the slop focus
C++ eternally BTFOhttps://www.youtube.com/watch?v=7fGB-hjc2Gc
>>107843995>languages that require a GC backed runtime>C++ replacementholy nocoder.
it's just C with classes and extra libraries stop seething
>>107841864have you ever used another compiled language, retard? header files are 100% unnecessary.
>>107844970b-but headers are based or something. modules are tranny. you can't just, define a binary format for exported symbols, name mangling and linker behavior.honestly I fucking hate headers too and all the insane copes people do to make them work well, like precompiling them and shit. combine that with C++ basically have templates + header bullshit and you wonder why you haven't suicided yourself in cringe. One of the reasons I refuse to learn OCaml is because of the retarded header garbage, especially since headers are the only way to actually type ocaml code. wtf?
>>107838681>breakdownkill yourself
What's a good mouse for gaming? I'm rolling with that HP
>>107845472Any particular model?
>>107845539
>>107845449i use this one because i don't have impostor syndrome like most other people
>>107845602i bought an m3k and a model f and thought i was baller but now 64gb of ram costs more
>>107845788ddr5 is so slow that i might straight up switch to intel just for the ddr4 support
previous: >>107833909#define __NR_mmap 9this is probably my favorite syscall of all time. we could spend weeks discussing this alone. it is extremely powerful, versatile, and widely used. not to mention, it's one of the (somewhat) rare six argument syscalls. some potential points of discussion:> the addr argument, and its use without flags, with MAP_FIXED, and with MAP_FIXED_NOREPLACE> file-backed vs anonymous mappings> the concept of pages, page sizing, and alignment > guard pages and PROT_NONE> the actual meaning of SIGSEGV, and how there's more to segfaults than simply process crashes> core dumps and stack traces> other related signals, such as SIGBUS> MAP_GROWSDOWN and the stack> the use of mmap (as opposed to brk) for allocation via the *alloc family> manual memory management vs an allocation schemeComment too long. Click here to view the full text.
#define __NR_mmap 9
>>107845701I suppose your approach is superior if there's just one connection or one thread doing stuff. What I was thinking about is a scenario where you're using some sort of I/O multiplexing or running one thread per connection, you're expecting lots of connections to be established and you can guesstimate that all processing as part of a single request should not allocate more than X bytes of memory (but you don't want to commit the pages instantly because the overwhelming majority of requests will require less). mmap(2) with its lazy commit-on-write semantics and explicit munmap(2) (or something to that effect) seemed to me like an efficient way to manage memory for this type of applications.
>>107845761i meant reuse the entries. so you have X slots open and then just assign a transaction to an open one whenever you need it>>107845764you're really just reinventing an allocator at this point. i guess the one difference is, depending on the specific implementation, they don't all handle large data types very gracefully
>>107844659Nobody has mentioned the most significant advantage and disadvantage, I think:Advantage:You are letting the kernel do its job. You express the intent to read a file, but the kernel decides when to read that file into RAM, when to evict it, how much to read ahead (although you can give hints). Since you're given a direct view of the page cache, there's no copying needed. In contrast, by using read(), you force the kernel to immediately copy a part of the file into your own buffer, which gives the kernel no opportunity to optimize anything.Disadvantage:There is no non-blocking way to access the file you mapped. When the part you're reading is not in the page cache, your thread just gets thrown out the window. This is an ISA-level problem, unfortunately, as there is no instruction to read memory in a non-blocking way. To be fair, it's not even clear what non-blocking would mean in general (an NFS mount is surely blocking, but what about a fast SSD?). You'd need some instruction like "If I were to read from this pointer, would it likely complete in X nanoseconds?", which is not something that exists in any ISA as far as I know.
>>107841306something cool about the BSD-style vm is that mmap doesn't actually need to set any PTEs - the vm fault handler can simply lookup the vm_entry structures (representing each mapped virtual address range & its flags) for the process' vm map at the faulted virtual address and install the PTE when the page's first access' fault occurs. even when a process forks, pretty much all that needs to be done is give the child separate copies of the parent's vm entries, but doesn't need to copy any PTEs over since the child process will only map them into its page table at faultthe vmm also maintains vm objects each with a listing its physical vm_pages, and "shadow" vm objects that can act as per-process containers for its modified copies of pages. these objects are chained on top of each other with the bottom-level being the original backing object which (for example) might be a file's system-wide page cache that is trivially shared between both mmap()'ings and the vfs-related syscalls (read/write/etc) - also called 'unified buffer cache'SHARED mappings will be flushed to disk if flie-backed and never shadowed/copy-on-write since they just share the same object across all processes. the kernel maintains a NEEDS_COPY flag per-entry which is set in both the parent and child's PRIVATE mappings at time of fork() and both process' PTEs are setup for write fault, so either process' modifications to its private pages are kept separate and stored in shadow objects which are created at copy-on-write if the entry's NEEDS_COPY flag is sethttps://docs.freebsd.org/en/articles/vm-design/if you want a proper description of the shit i'm sperging about
>>107841306>>107844659>>107845777"Memory-mapped" files were just the normal way files worked in Multics and object-based operating systems. UNIX people were against the whole idea, the same way they didn't like dynamic linking, threads, or strings besides null-terminated strings. Then 30 years later they finally realized that those people were right all along.
This is a thread for all AI CLI related discussionClaude Code: https://code.claude.com/docs/en/overviewGemini CLI: https://geminicli.com/OpenAI Codex: https://openai.com/codex/OpenCode: https://opencode.ai/New:>Anthropic Introduces Cowork: Claude Code for the rest of your work>https://x.com/claudeai/status/2010805682434666759>aka a gui for Claude Code
>>107845078>Claude is now capaple of replacing office monkeys who have no particular skills outside of Word, PowerPoint, Excel, wikis, ticketing systems, etc.>Claude can use tools to remote in to computers, change passwords, reboot them, and install software>self hostable models will catch up to this in under two yearsIt's over. Indian offshoring/bodyshopping is finished.
I like CC a lot but god damnit why did they write it in javascriptI found out that if i make a certain request to analyze my codebase it completely hangs the session and i have to SIGKILL it to close it
I have used gemini 3 to auto generate in ~40min what I came up with myself in 3h its not bad but code design wise it its quite annoying. Was a simple sensor monitoring application with a drop in viewer all in python. One time scripts are great I guess but it doesn't handle edge cases also its obsessed with crash testing via Ctrl+C for some reason.It does feel nice to have a quick prototype of what you want instead of having a to fight the slog that comes with proper design is Claude Code better?
Vibecoded OShttps://github.com/kaansenol5/VibeOS
fixed the link for youhttps://www.office.com/
imagine how many vulnerabilities this shit has
I started writing my wanna-be OS code (by far an overstatement to call it anything but a mess) a month before that guy and got not even my IDT working properly, let alone interrupts to get keyboard input ;_;;;;; I feel somewhat demoralized.
Is this a good option to escape (((Intel ME))) and (((AMD PSP)))?Yes, I'm a targeted individual
>>107840951>is the chip developed by the country known for mass surveillance a good way to escape backdoorskek
>>107841099kek
>>107844992found it, its too good to not post it
>>107840951it doesn't exist, AI generated specifically targeted for you, it's intel inside
>>107845763Loongson is a real thing thougheverbeit.
Popular chatting platform Discord has confidentially filed for an initial public offering (IPO), reports Bloomberg, citing sources close to the matter.The company has been considering an IPO for years. Last year, sources told Bloomberg that Discord was working with Goldman Sachs, and JPMorgan Chase on the listing.
I could have sworn they already IPOed because it went to shit like 7 years ago.
>>107845573Switch to Matrix?
>>107845573I was so pissed when I found out about this. I get it, their platform costs a fuckton of money to run and their attempts at monetization aren’t making cash fast enough, but making your platform even worse by being beholden to investor cucks isn’t going to save it. Discord will be Skype within 10 years, I’m sure, while some other company makes yet another chat app that is essentially Mumble but better.
>>107845703whats there to offer that normies care about and discord doesnt offer?
>>107845726800 BILLION TIMES MORE AI SLOP
for what purpose?
>>107844575>womenthere's the problem
>>10784306Literally let me root it and I'll buy one. They won't so I won't.
>>107844642meant for >>107843062
>>107843413my titan pocket has 3.5mm and SD card, even IR remotemajority of unihertz have that (or at least used to)
i use colemak-dh and i attempted to use this keyboard layout on a touchscreen. it's not very goodi think maybe qwerty is better for small keyboards than colemak, which was a shocking realization for me because on a full physical keyboard colemak-dh is far superior
Nuked by an escaped /v/ Schizo editionPrevious Thread: >>107781231>Links:>DALL-E 3https://www.bing.com/images/createhttps://designer.microsoft.com/image-creator>4ohttps://chatgpt.com/https://sora.chatgpt.comhttps://copilot.microsoft.com/>Imagen 4 and Nano Banana (Pro)https://gemini.google.com/apphttps://labs.google/fx/tools/image-fxhttps://labs.google/fx/tools/whiskComment too long. Click here to view the full text.
>>107845728Oh, we have *that* anon who lurks here just to repost *that* screenshot, one should never underestimate mental illness. It's your turn to bake anon, by the way.
>>107845728baking. just looking for an appropriate OP.BTW, is the /vg/ thread gone?
>>107845752It's still there I think.
>>107845752Nope, they changed it into>>>/vg/vigwhich works better/more-easily when linking.
>>107845752>>107845760yep, updating now, thanks.>>107845785>>107845785
>Read the sticky: >>105076684>GNU/Linux questions >>>/g/fglt>Windows questions >>>/g/fwt>PC building? >>>/g/pcbg>Programming questions >>>/g/dpt>Obsolete laptops >>/g/tpg>Cheap electronics >>>/g/csg>Server questions >>>/g/hsg>Buying headphones >>>/g/hpg>How to find/activate any version of Windows?https://rentry.org/installwindowsPrevious: >>107789452
>>107845559>>107845524>Pic related are my global settings for my 240hz monitor.
>>107844998>line interactive>sub 6ms latencyThey actually made a real UPS. Costs about what I would expect, but good to see.
>>107845579Yeah like I said, on paper it's good specs, I was just not keen on relying on the basic bitch windows power plan settings to control everything in windows with it.Since it just reports itself as a "Smart Battery" (essentially making windows think it's a laptop battery) windows doesn't REALLY treat it like a proper UPS.Using NUT/WinNUT-Client and some custom configuration files and it at least has fairly robust automatic shutdown capabilities. Sadly the USB doesn't seem to give access to any of the deeper details (battery voltage, input/output voltage, current load, etc) but it does at least give the UPS state (on battery, or online (AC power)), and the battery charge level (as a percentage) which is enough to get what I wanted set up. While i'd love to see more of those deeper statistics, im not gonna pay APC $1900 for that.
>>107845559>>107845571Thanks anon. Just tried this and so far I'm not seeing any screen tearing.
>>107844931>Nice blog, upvoted! What's your question?Wrong thread, wouldn't let me delete
World of Tomorrow Edition Previous Thread: >>107818694>Links:>DALL-E 3https://www.bing.com/images/createhttps://designer.microsoft.com/image-creator>4ohttps://chatgpt.com/https://sora.chatgpt.comhttps://copilot.microsoft.com/>Imagen 4 and Nano Banana (Pro)https://gemini.google.com/apphttps://labs.google/fx/tools/image-fxhttps://labs.google/fx/tools/whiskComment too long. Click here to view the full text.
As soon as games stalled and were turned into pseudo social media the miasma of conflictivity began to invade the rest. Polcrap found fertile ground in that mindset.
