#vram
"64" is the minimum VRAM
November 19, 2026 at 1:56 PM
I like pronouncing VRAM as one syllable
November 18, 2026 at 7:48 PM
~100t/s prefill and ~20t/s on my 9060XT with Inkling mini, IQ3 with 128k context runs fully in vram. It's ok. Definitely like it better than qwen, but it's not amazing. Don't expect miracles from 16GB of vram yet.
November 20, 2026 at 10:13 PM
Moore Threads teases new GPU with “highest VRAM capacity” on a consumer graphics card
Moore Threads teases new GPU with "highest VRAM capacity" on a consumer graphics card - VideoCardz.com
Moore Threads claims its upcoming Lushan GPU offers 50× faster ray tracing than its previous generation. New hints suggest it could carry more than 32GB of VRAM.
videocardz.com
November 21, 2026 at 10:10 AM
right up until until we all started needing 400GB VRAM for models at least
November 18, 2026 at 4:02 PM
KAMEN•RIDER•VRAM - Rakia Amarga é sem sombra de dúvidas um dos melhores Riders terciários da franquia.
#tokusatsu #toku #kamenrider
November 17, 2026 at 3:01 PM
small VRAM humiliation
November 20, 2026 at 9:45 PM
free VRAM für alleeeeeee
November 18, 2026 at 1:50 PM
It’s not often I need to use RF output so brought my trusty KV-9PT50 down from the office to test this Colecovision before and after VRAM replacement 👍
November 16, 2026 at 6:58 PM
Intel Preps Linux 7.5 Change To Better Handle Multiple GPUs With Lots Of VRAM

https://www.phoronix.com/news/Linux-7.5-DRM-8TB-mmap
November 20, 2026 at 8:00 PM
what quant (if any) and any SSD streaming? 5 GiB VRAM is NOTHING if that's a reasonable quant wtf?
November 21, 2026 at 12:25 AM
Intel Preps Linux 7.5 Change To Better Handle Multiple GPUs With Lots Of VRAM - https://www.phoronix.com/news/Linux-7.5-DRM-8TB-mmap
Intel Preps Linux 7.5 Change To Better Handle Multiple GPUs With Lots Of VRAM
A change set to land for the upcoming Linux 7.5 kernel cycle expands the mmap range size for GEM/TTM buffer objects in the kernel graphics drivers from 1TB to now up to 8TB...
www.phoronix.com
November 20, 2026 at 7:48 PM
Oh, I also figured out how to render in 480i and it looks so high-quality... 💟

Sadly, 640x480 framebuffer occupies almost half of VRAM and you can't double-buffer, so it's very limited...

Basically if your game doesn't run in 60 FPS, you're screwed. :D
November 19, 2026 at 1:10 AM
Played all the way through it on a repurposed Dell mini pc with 8gb of ram, 128mb vram. Zero issues
November 20, 2026 at 9:36 PM
6gb vram is crazy you could maybe play Borderlands 2 with that much
November 21, 2026 at 1:42 AM
Awesome open-source AI: FreeVideo runs MiniMax H3 video generation locally on 8GB VRAM. Full setup & guide in Telegram: https://t.me/synthesis_feed #ai #tech #opensource #coding #devtools #automation
November 20, 2026 at 12:36 AM
Just noticing that Amazon listing show 8gb vram, but Google says 16gb...
November 19, 2026 at 6:09 PM
ran my ray tracer and somehow it was 4x slower which made me panic. it turns out it was because I ran out of VRAM because I had 4 instances of Blender open 🥲
November 21, 2026 at 1:44 AM
Kai - Video
"54GB VRAM For $35 The Cheapest Local AI Hack?"...

youtu.be/oUsXv7ZoYQg

==========================
#librecanada #politicscanada
54GB VRAM For $35 The Cheapest Local AI Hack?
YouTube video by Kai
youtu.be
November 18, 2026 at 6:55 PM
Крутой открытый софт: FreeVideo генерирует видео MiniMax H3 локально на 8 ГБ VRAM. Разбор в Telegram: https://t.me/the_sintez #софт #ai #opensource #нейросети #разработка
November 20, 2026 at 12:36 AM
I know gating cores/etc is common to handle varying batch quality, but shipping 80GB VRAM with all but 8GB fused off is *wild*. In this economy??
November 15, 2026 at 6:05 PM
OVHcloud Olala mini 7A1B (8GB VRAM / CPU) ($0.03/$0.09) vs GPT-6 Luna Pro ($0.1/$0.5) at 10K DAU:

OVHcloud Olala mini 7A1B (8GB VRAM / CPU) comes out 76% cheaper ($8,208/yr saved before caching).

which one are you routing primary traffic to?
November 20, 2026 at 1:45 PM
corrupted vram always makes for fun texture
November 18, 2026 at 2:32 AM
So you're estimating VRAM from model weights alone — how do you account for KV cache overhead across different context lengths?
November 22, 2026 at 12:17 AM
oh wow one of these: wiki.preterhuman.net/NEC_MonoGrap...

512KB VRAM, 1024x1024, one of Intel's early graphics controller chips, the 82786.
wiki.preterhuman.net
November 21, 2026 at 12:16 AM