GeForce RTX 3080

Not suggesting a provider but if you're willing to get a Chinamod GPU, go get RTX 3080 20G or RTX 2080 Ti 22G. [...] I've got a RTX 3080 20G for $450 like a year ago. With llama.cpp, bf16 kv cache, kv cache offloaded to RAM, Qwen3.8 27B UD-Q4KXL, single RTX 3080 20G, I got 10 tok/s initially and it dropped to 5 tok/s at 50k context. [...] Building a dual Chinamod GPU machine would only cost like $1000$1500. [...]

[...] I'm running quantized Qwen and Gemma on a decent, but three years old gaming rig (think RTX 3080 12GB and 32 GB RAM). Yes, it's slow, it has a small context window. But it can (given a proper harness) run through my trip photos and categorize them. [...] The cloud versions are orders of magnitude cheaper than self hosting, because sharing can utilize servers much more efficiently. Company can spend half a million bucks on a rig running GLM 5.1, and get data security, flexibility and lack of censorship, but oh it's so expensive compared to Anthropic per-seat plans.

[...] Every other update was a blank/black screen and me without a good remote config to try to recover it. After about 6 months it was actually stable, but I'd since gone ahead and paid too much for an RTX 3080, and gone back to my windows drive... [...] ran all the updates and it's run smoothly since. [...]

[...] The Dell Precision 7XXX, HP Zbook Fury, and Lenovo P5x/P7x series are the primary high-end notebook workstations, and they are all nearly infinitely configurable with a myriad of CPU, GPU, memory, storage, display, and connectivity options. I myself have a Precision 7560 from 2021 that has a Xeon W 11955M, and an RTX 3080 Laptop GPU that's roughly between a desktop 3060 and 3070 in performance due to the power limit of 90 W. [...] Both Precisions have 4 DDR4 slots for up to 128 GB, 3 M.2 slots (and an additional one for the WiFi module), and full repair/service manuals online.

I moved about 4 months ago to Bazzite on my nvidia RTX 3080 desktop. [...] One time after an update it completely broke and I had to reinstall. [...] - Once after a big update Overwatch completely crapped out and I had to reinstall the game. [...]

single RTX 3080 20G, I got 10 tok/s initially Add $200, buy a used 3060 and you won't need to offload cache to RAM. Yo'd have like 50-60 t/s with MTP enabled.

yes, I average 80-120 tok/s on my RTX 3080 with gemma 4 and faster with Qwen 3.5. The main use-case here is just code-monkey agents. I'm not looking for architectural guidance, but an agent to take a spec and complete it.

Don't really have the room for a gaming PC anymore after moving, I've got my old one but it needs some fixing up (RTX 3080). 1TB Xbox One Series X I got second hand for about $400 (also added a storage expansion for about $130) is doing me sound for any gaming needs (currency converted from GBP). All the games optimised, usually 60fps, no messing about with settings other than the quality/performance switches. [...] I do miss stuff like modded Valve games and keyboard & mouse, but can run them somewhat on my laptop.

[...] If you already have an RTX 3080 it's absolutely worth the time to just mess around and see how it does, experiment with different quants that fit in your VRAM. [...] If you are experimenting it's worth mentioning that the harness/tooling is very important to getting a solid experience. [...]

I've bought a laptop for <1500€ that came with 32GB of RAM and an RTX 3080 with 16GB or VRAM. So I don't think 2500€ device is necessary, though I'm certain it would yield better and faster results.

I did tinker a lil with mine! RTX3080 with 10GB VRAM, 5600x with 64GB DDR4 - not very good but it was very fun and exciting to tinker with :) My partner on the otherhand has an M3 Max 64GB which I've had way more success with. Setting up opencode and doing a tiny spec-driven Rust project and watching it kiiinda work was extraordinarily exciting!

Fyi, I believe --flash-attn on doesn't do anything, you should instead use --flash-attn 1. I'm getting 150t/s on a RTX 3080 10GB as well with f16 cache type.

It does work and no you don’t need to fiddle with build errors and crap unless you have some weird hardware. [...] Core i9, RTX 3080, Fedora w/ Proton on Vulkan. [...]

[...] When Windows 10 became unsupported, so did Minecraft updates. [...] The machine has a RTX 3080, which I almost didn't buy, because I've had issue with Nvidia on Linux in the past, but haven't had to do a single tweak this time.

I had said I wouldn’t upgrade from my RTX 3080 until I could run “true 4K”. I finally got the 240hz 4K uncompressed but it required buying a $1300 Asus OLED monitor and the RTX 5090. [...] I think it uses DisplayPort 2.1? [...]

[...] My most recent install from this year of Kubuntu/KDE Fedora had strange bugs where applications froze and quitting them were more painful than macOS/Windows, or that software updates through their app store thingy end up in some weird state that won't reset no matter how many times I reboot, hard crashes and so on on a relatively modern PC (5900X, RTX 3080, 32G RAM). [...] You need a reality reset if you think getting used to Linux is a qualifier to making great software. [...]

[...] My most recent install from this year of Kubuntu/KDE Fedora had strange bugs where applications froze and quitting them were more painful than macOS/Windows, or that software updates through their app store thingy end up in some weird state that won't reset no matter how many times I reboot, hard crashes and so on on a relatively modern PC (5900X, RTX 3080, 32G RAM). I had to figure out the commands to force reset/clean up things surrounding the package management in order to continue to install/update packages. [...] This is a dealbreaker for the vast majority of people but let's come to your more interesting take: If you find it hard to adjust to a Linux desktop you should not be developing software And that sums up the vast majority of Linux users who still think every other year is the year of "Linux desktop". It's that deeply ignorant attitude instead of acknowledging all these years of clusterfuck after clusterfuck of GUIs, desktop envs, underlying tech changes (Xorg, Wayland) and myriads of confusing package distribution choices (debs, rpms, snaps, flatpaks, appimages and so on), that no sane person is ever going to embrace a Linux distro as their daily driver. You need a reality reset if you think getting used to Linux is a qualifier to making great software.

[...] I broke out my little Nvidia GeForce RTX 3080 powered laptop with 16GB of VRAM and quickly started training. [...]

[...] I'd been lucky enough to get an RTX3080 for my new computer build and decided to experiment with mining ETH during downtime - it was winter and the office was cold anyway... [...] I hadn't really created anything of value so I didn't feel as angry as if I'd been cheated out of actual work. [...]

Instant image generation that responds as users type – a first in the field Stable Diffusion Turbo has been able to do this for more than a year, even on my "mere" RTX 3080.

[...] The performance is worse without being significantly prettier unless I crank the graphics up, in which case I then get 5 fps. I've got an RTX 3080, but I'm bottlenecked by my i9-9900K. [...]

I love my 65" LG GX OLED as daily driver for work and gaming. See I came from an ubutto revolution cockpit setup with 4 monitors, the ergonomics were awful. With the LG I'm about a meter or less away from screen and use window management tools to pull focus to the center lower section for any focused work. I run Win 11 from an RTX3080 card with a 2.1 HDMI cable. [...]

[...] I get like 120 FPS with a 5600X and RTX 3080. I've been dreaming of a 9800X3D based build when it comes out, but realistically don't have the time to actually play.

[...] From my pool of machines I found out only my RTX 2070 and RTX 3080 Mobile are compatible. [...] Interestingly I found no similar software to the chunked encoding in my system, but am chalking it up on me not searching hard enough or these being integrated deep in to commercial/private systems. It would make sense that big players like YouTube have something like this in place, as reducing latency for individual upload transcodes finishing is beneficial instead of limiting their speed to a single hardware encoding / software encoding node. [...]

[...] I had a not-so-powerful AMD GPU in my homelab server that somehow kept crashing the kernel any time I tried to encode videos with it and also an NVIDIA RTX 3080 in a gaming machine. [...]

In my humble opinion, a chat interface (API or not) does not a product make. Not to mention that Llama is free and competitive with both (same with Mistral, heck the 7B model works great on my RTX 3080). If you started a company that blew up because you made a badass product (and let's say you used ChatGPT under the hood), you would just eventually train and deploy your own model because an LLM is not a product.

[...] It should be using OpenGL with display lists (or some more modern variant on that). [...] With the default ribbon rendering, rotating is completely smooth; with all atoms displayed (wireframe or sphere), it's still smooth. [...] Rotating shows my GPU (nvidia RTX 3080 Ti) at about 50% utilization, and if I exit Chimera, my GPU's releases 200MB of memory. [...]

[...] Works well on a laptop with RTX-3080. [...] EDIT: The examples\screen demo almost feels realtime. [...]

I’m running Alan Wake 2 on an RTX 3080 with a 7800X3D and the first few sections in the forest (don’t think describing forest scenes in Alan Wake 2 is a spoiler here) ran around 30fps, and then the absolutely sumptuous next few levels (man the maps are gorgeous) have been running at 60fps rendered at 1080p upscaled to 1440p. [...]

You can buy an RTX 3080 for $400 or a 3090 for $650-700 right now. Those can both run the game fine. I got mine from some gamer kid in a beat up van at the gas station and it’s working great.

[...] I still have a 600W EVGA PSU from one of the bundles (RTX-3080) that I've only used once to test if my issue was the PSU, MB or something else... [...]

[...] The only compute I have at home is an RTX 3080 with 10 GB of VRAM, so I struggle with training larger models (40M, 50M params). [...]

I bought a geforce RTX 3080 at launch and boy was I surprised at the power draw and heat/noise it pumps out. [...]

I bought a geforce RTX 3080 at launch and boy was I surprised at the power draw and heat/noise it pumps out. [...] I actually run it 10% underclocked, barely affects performance, but greatly reduces heat/noise. These cards are configured to deliver maximum performance at any cost (besides system instability). My next GPU I am probably going mid-range to be honest, these beefy GPUs are not worth it anymore cost and performance-wise. You are better off buying the cheaper models and upgrading more often.

[...] I've just run the same prompt and query on my RTX3080 with openwebui as frontend. [...] I can't set the conext size any higher because my GPU only has 12GB of RAM and ollama crashes for larger context sizes. [...]

The criteria for purchase for anybody trying to use it is "run slowly but acceptably" vs. "run so slow as to be unusable". [...] Perceptually I can live with the M4's performance. [...] The 5950/RTX3080's is too slow to be even remotely usable with the 70b parameter model.

[...] It's got an i7-11800H, RTX 3080, 64GB of RAM, and a Samsung 990 PRO nvme drive. When I bought it, it came installed with Windows 11. Even with the proclaimed performance enhancements of Windows 11, this laptop seemed quite slow for the specs it was sporting. [...] I've tried on one other occasion to upgrade with Windows 11, but the performance just does not match up. [...] I recently tried out Linux Mint and immediately ran into an issue with my mouse wheel + back/forward buttons not working. [...]

[...] I had bought a 6900XT myself around launch time (the RTX3080 I ordered was not coming, it was the chip shortage times...) and it took around 2 years for Pytorch to become actually usable on it.

[...] I ran a GTX 1080 and then an RTX 3080. The performance was not very good in modern games designed for current gen consoles like pretty much any BR game for example. [...]

[...] I am currently testing the limits and got llama 3 70B in a 2bit-quantized form to run on my laptop with very low specs RTX3080 8GB VRAM (laptop version) and 16GB system RAM. It runs with 1,2 tokens/s which is a bit slow. [...] The 4070 also has the issue with quite long time for the first token to be displayed i think it was around 12s in my limited testinh. [...]

I just tried this and ran into a few hiccups before I got it working (on a Windows desktop with a NVIDIA GeForce RTX 3080 Ti) WSL outputs this error (hidden by the one-liner's map to dev/null) error: APE is running on WIN32 inside WSL. You need to run: sudo sh -c 'echo -1 /proc/sys/fs/binfmtmisc/WSLInterop' Then zsh hits: zsh: exec format error: ./llava-v1.5-7b-q4-main.llamafile so I had to run it in bash. (The title says bash, I know, but it seems weird that it wouldn't work in zsh) It also reports a warning that GPU offloading is not supported, but it's probably a WSL thing (I don't do any GPU programming on my windows machine).

I think it deserves a lot of praise, but also some criticism: While it may keep their creative vision "pure", the lack of lower graphical settings makes it run pretty terribly on an RTX 3080, which is very frustrating. [...]

This is basically telling Claude "Make Resizable Bar work on my RTX 3080" and publishing the summary of the conversation. For context, the RTX 3080 got ReBAR support with a VBIOS update. Some models ship with an updated VBIOS that supports it, some don't. [...] The GPU manufacturer only provides Windows executables to update the VBIOS, the author's on Linux. [...]

So I have a RTX 3080 10GB VRAM which I've been using with Qwen2.5 Coder and Gemma 4 E2B. Im wondering what models you have tried with which quants.

[...] The only compute I have at home is an RTX 3080 with 10 GB of VRAM, so I struggle with training larger models (40M, 50M params). [...]

I just migrated to linux (Bazzite) in March, I have a RTX 3080. The only issue I ran into was that video stream compression is not supported on linux so I can't run 1440p 165hz with HDR on because my monitor doesn't support HDMI 2.1. Either I need to turn off HDR or lower refresh rate to 120hz.

I've been toying with the idea of building a dedicated machine to continue working on my FPV flight control model. So far I've just been renting gpu time but it's getting expensive. I also occasionally use qwen3-coder in my homelab but my headless "inference" machine is quite limited with an RTX 3080. Curious if this build from llamabuilds.ai looks good? [...] I'm maybe considering buying more GPUs in the future - hard to tell if stacking RTX 6000 ADA's is worthwhile! [...]

[...] got an RTX3080 at home and tinkered around with stable diffusion when it came out. [...]

For someone jumping back on the local LLM train after having been out for 2 years, what is the current best local web-server solution to host this for myself on a GPU (RTX3080) Linux server? [...]

[...] While some of the Open Source solutions do quite well, surya took 188 seconds to process 88 pages on my RTX 3080, while the cloud ones were a few seconds to upload the docs and download them all. [...]