#What are you using this for

1 messages · Page 1 of 1 (latest)

compact fern
#

How much watts is the other psu, how hot does it get both cpu and each gpu, what are you using this for, and how much was the whole build.

vital spade
compact fern
molten thorn
#

does 1000W suffice for that gpu?

compact fern
molten thorn
#

?

molten thorn
#

seems like an AI build or something

compact fern
#

Nvm i saw ur pc

#

What is ur cpu temps at idle

#

And underworkload

molten thorn
molten thorn
molten thorn
#

i removed these towers and fridge magnets tho

compact fern
#

Does it have 3 fans under it

molten thorn
#

under the GPU?

#

gpu has 3 fans

fathom quartz
molten thorn
#

no way 1000W can support 2 12vhpwr

molten thorn
fathom quartz
#

I mean, its obsolete with more Watts hidethepain

#

But still interesting

#

Where is the 2nd PSU?

#

And what 8K 3D VR XXX do you have to watch to make this thing sweat?

molten thorn
compact fern
#

With 2 gpus do u get more frames or something can the cpu handle that

#

Also does that mean u can fit 8 montiors and play them all at 4k 240hz if they all support dp 2.1?

molten thorn
#

usually its just for AI

#

can run models at 1.7x the speed of one GPU

#

idk of any other application other than virtualisation maybe

compact fern
#

Have u used ai before do u think my build in my bio is good for running ai and is it free?

fathom quartz
#

Nvidia has the advantage of CUDA Cores, and the 4090 with 24GB is a good entry I guess

#

Only tested AI models with the. RTX 3080, and the 10GB of VRAM are literally the minimum for AI

compact fern
fathom quartz
#

I have no experience only used 2 models so far, magicquill for example requires atleast 8GB of VRAM and CUDA cores, so no AMD possible AFAIK.
The RTX 3080 doesn't have bad stats for the process, but suffocates from the 10GB VRAM.
There are also model which require over 50GB of VRAM or Smth cryblob

#

It really depends on what you are planning to do

#

And I have used open-source models i guess

#

So I didn't pay antyhing

#

Storage depends on the model, can be a lot or less

#

And I have used pre learned checkpoints, so the AI already has a lot of knowledge

#

But required a lot more of VRAM when I am correct, so I turned that off so the 3080 won't crash

molten thorn
#

a lot of AI runs fine on 3080

molten thorn
compact fern
fathom quartz
#

They are not optimized, when I understand it correctly.
There is Flux for example, with multiple kind of "versions".
One is fast and small, and the other is larger and requires more VRAM, correct me if im wrong

#

Also the larger model delivers more quality i guess

fathom quartz
compact fern
#

Are these models like chatgpt but u can do anything u want with them

fathom quartz
#

But 10GB VRAMkek

#

ChatGPT is chat based

compact fern
#

I want to add an ai model for a homeserver project

fathom quartz
#

You can also use Pinokio, its easy for an entry

compact fern
#

Idk much about ai, what do peiple usually use it for when they have it on their storage

molten thorn
#

AI isnt just doing LLMs

#

or doing just SDXL

fathom quartz
#

Yea I mostly did LLMs

#

And SDXL

compact fern
# molten thorn well depends

Did u have ur 14900k when the 13th and 14th gens were degrading, i never played games but i do remember my cpu always hitting 100c and i thought that was normal

#

Luckily i never played games

molten thorn
#

mines not thermal throttling

#

tho PL1 and PL2 are both 253W so that probably helps it not thermal throttle

molten thorn
#

77C on OCCT extreme avx2 variable with 253W as PL1 and PL2

compact fern
molten thorn
#

usually while playhing the game too