DDraceNetwork - off-topic

chancho 2023-09-18 02:55:00Z

@jao's hairy kebab

2

08:18

Ceemsti

Click to see attachment 🖼️

RonIn

Click to see attachment 🖼️

Ceemsti

@jao's hairy kebab

@Ham5terzilla (Макс) Ты уверен, что тебе хватит 128 мутов?)

Mʎɹ シ

@Ham5terzilla (Макс) Ты уверен, что тебе хватит 128 мутов?)

да, вполне

Would you guys recommend Radeon RX 6750 XT or rtx 4060ti?

18:23

The price is the exat same (edited)

4060 ti

jopsit werent u pro amd?

a 6750xt for same price is scam

18:31

it should be at least 30-50€ cheaper

18:31

https://www.mindfactory.de/product_info.php/12GB-MSI-Radeon-RX-6750-XT-Mech-2X-OC-Aktiv-PCIe-4-0-x16--Retail-_1452947.html

12GB MSI Radeon RX 6750 XT Mech 2X OC Aktiv PCIe 4.0 x16 (Retail) -...

Radeon RX Serie von MSI | 12GB MSI Radeon RX 6750 XT Mech 2X OC Aktiv PCIe 4.0 x16 (Retail) :: Bestellt :: über 1.330 verkauft :: 26 Jahre Kompetenz | Hier bestellen

18:31

cheapest 6750 xt

someone explain to me

18:31

the trick to finding a flat fast

18:31

i hate this

https://www.mindfactory.de/product_info.php/8GB-Gainward-GeForce-RTX-4060-Ti-Pegasus-Aktiv-PCIe-4-0-x16--x8-Retail-_1493696.html

8GB Gainward GeForce RTX 4060 Ti Pegasus Aktiv PCIe 4.0 x16 (x8) (R...

GeForce RTX für Gaming von Gainward | 8GB Gainward GeForce RTX 4060 Ti Pegasus Aktiv PCIe 4.0 x16 (x8) (Retail) :: Bestellt :: 26 Jahre Kompetenz | Hier bestellen

18:31

cheapest 4060 ti

Ryozuki

the trick to finding a flat fast

be white, have a native sounding name, be rich. have a gf have a baby

i have the first 2

18:32

and my salary is decent

18:32

actually p good for spain

need gf now

ok, should i find you one?

ye

alright

lets hop into tw

18:45

and find gfs

xd

Jupstar ✪

it should be at least 30-50€ cheaper

But the performance is basically the same if not better on Radeon RX 6750 XT and more vram lol

pilonpl

But the performance is basically the same if not better on Radeon RX 6750 XT and more vram lol

it should be bit slower

18:58

but it's also cheaper

18:58

if u buy it cheaper go with amd

18:58

else with the nvidia

But it isn't slower?

not as far as i can see

19:03

it's basically same speed

19:03

sometimes bit faster

19:03

6000 series is not current gen

Does that matter tho

no

19:05

but e.g. a 7700 xt is faster than 4060 ti

That's interesting

What movie i already watched 200 times should i watch today, because all new/other movies are trash

Alright so actually the only issue with rtx 4060ti is the vram

19:51

Because rx 6750 xt has more vram for the same price

19:51

So it seems kind of better

why even same price xd

19:52

6750 xt is cheaper

And the rx 7700 xt seems like a really good deal because it has comparable performance to rtx 4070 apparently which is insane

yes

19:54

i'd go with that tbh

19:54

if u have the budget

And the issue i see with amd is that it is not nvidia and doesn't have cuda cores and nvidia is generally more standard

with amd u have rocm

Basically nvidia doesn't seem to have any disadvantages beside price

nowadays yes. esp for windows users

19:56

few years ago, amd still build in HBM2 memory into their GPUs

19:56

which is now used for AI cards mostly

19:56

sad they dont do it anymore

Jupstar ✪

with amd u have rocm

What's rocm lol

19:57

And more importantly, why should i care about it lol

it's amd's answer to cuda basically

And even more importantly it isn't better then cuda i suppose

it's built into blender for example

Because nvidia is bigger and more standard i guess

yeah depends on the software u use

Jupstar ✪

it's built into blender for example

Is it now?

yep

What about ai?

well neither AMD nor NVIDIA have smth for u

19:58

for a fair price

19:58

AIs need huge amount of memory

But like isn't cuda for ai basically?

not even nvidias best card is good enough actually

Stable diffusion runs on cuda

yes, but u can also use rocm

19:59

and rocm xd

I know it can but i think it's not official

19:59

https://pytorch.org/

PyTorch

Okay so basically amd is just better??? Wtf

but as said

20:00

u wont get a good AI card here

20:00

not with AMD not with nvidia

Yeah i know

i tried to run a few models, but it has to partially live in RAM instead of VRAM and that makes it very slow

Small models can run on cheap gpus

true

Like stable diffusion

20:01

And small models are the future

no

20:01

the human brain has 600 trillion synapses, and 90+ billion neurons

20:02

we are actually still far away from having the hardware requirements

Humans might not be very well optimized

20:02

Some animals have larger brains than us

hard to say

pilonpl

Some animals have larger brains than us

well but not as many synapses

20:03

and also most of their brain is used for movement etc.

20:03

even humans use like 30% for eye sight

The human brain is made for survival

20:04

Overall ai doesn't have to be as complex as the human brain is to still be better than us

20:04

I think

"“A toddler's brain is forming more than 1 million connections every second.” Research shows that they'll have roughly 15,000 neural connections per neuron by the time they are age two or three. They started out with approximately 2,500 per neuron. In total, a toddler's busy brain has about 1,000 trillion synapses." baby brains are actually insane

pilonpl

Overall ai doesn't have to be as complex as the human brain is to still be better than us

they are better in a single task