U guys getting pc for studies 😐

R U KIDDING ME?????? FOR STUDY??😐

😂😂😂👌yess

So I can do my homework faster

🤣

🤣

yea he can spend time with his weeb ass figurines...

He can render one in real time

He can stick his in that silly ROG logo it’s everywhere

This is cheaper anyway

Ai workstation basically any work related tasks

Definitely the only reason for this would be ai work Station a gaming server would need a server grade cpu to be able to support several different instances of virtual machines. However that is really the only difference between this set up and a true gaming server is the cpu used and of course motherboard. For a gaming server you would want a top of the line xeon or even better threadripper at minimum. The thing I don't get though is that it seems like it would be way more beneficial for ai workstations to use dedicated ai graphics cards. I'm wondering if this unit is maybe doing some ai stuff involving video and photos or graphics and hence the need for all the other none ai associated cores in the rtx 3090 gpus? Really wish this channel was in English and did some explaining cause you very rarely see builds like this even on youtube. I've only seen about four building like this using these newer rtx cards in nvlink situations and typically they're all saying they're doing so for ai workloads so it's most likely also the case here.

"Server" would be a little humble in this case.

not to mention the electricity bill behind that space heater too.

12:21 50c at full load

​@@BajaToxic It's still wattage being emitted into the room....you actually think the temperature itself is actually raising the temperature of the room? So if I took a grain of rice and heated it at 200C, my room will be 40C in no time? XD

@@InsuredFrames Most average PC run at 70c+ all the time and no one complain it warm the room.You forget there's aircond or fan running too.

​@@BajaToxic Like InsuredFrames mentioned, it's the wattage emitted vs the temp at which the GPU is running..Emitting 1000watts of heat over 500watts is double the heat being dumped into the room. The actual temp matters but significantly less.

No SLI support though

@@svartdraken will the feature eventually come or it will just never happan?

@@giannoklein2046 no they just abandoned it. I guess it makes sense but having the option would’ve been nice

@@svartdraken SLI Support what? NVLink is not needed anymore. Come out of your cave. PCIe5 gets better than NVLink speeds so they removed it.

@@svartdraken As I saw in this video the guy also skipped SLI bridges i think :)

How do I find my score

Mostly they do. U just a gay

haha, first thing I thought of was maxishine's tri sli evga hydro copper 580 build.

@@lachlanhamersley5394 i was think og the 9800 gtx or the duel 9800 cards

yeah i haven't seen a multi gpu build in years. last time i seen one was 2016-2017

@@wisdomfox857 sli 295s which were duel 280s. I miss those days, the card art and even some cool motherboard designs like the Gigabyte Assassin board.

@@UR5TEPDAD1 yap but now we have gpu thicker and taller then a ps5

Nb

It's useless and waste of money until & unless you got some nefarious plans ahead

se ci lavori si ma per giocare soldi buttati

At least you are not the only one bro 😭😭😭 wtf for me it's not a pc it's a whole server

Im just happy to buy my new 3060 tomorrow

Não é pc gamer....

Você não ia usar nem 5%😅

@@STRIKERZADA então é o que??

leia a descrição....@@DouglasRDJ

thats what in saying

if the CPU has the lanes it wont bottle neck. nothing but a bench mark will use all 4 once you do real world things your just wasting money using almost nothing.

@@channelofstuff6662 yeah, there are people that are not only gaming.

It’ll be fine! that core is xtreme series it’s a little old but pretty good, It has 18 cores with 36Threads. Im pretty sure there are newer and better cpu’s out there like the i9-13900k, which has 20 cores with 32T but runs in a faster frequency.

@@channelofstuff6662 you could use all 4 in touchdesigner or Magic Music Visualizer. They both let you keep assets in the vram for immediate access. This pc could probably run the Deadmau5 cube V3 show.

Да ладно, тыщ за 500 уложиться можно, если не ещё дешевле

Интересно сколько в кс фпс будет

@@geoph 50

ДНс один из самых богатых дядек в России Он себе может 100 таких позволить и даже не почувствует чееееееееееееееееллллллллллллллллл

Ahahha that's good one

🤣🤣🤣🤣

You’re electric bill and psu must be insane

@@stanleybochenek1862 derp

Sim

I wish there would be sli/nvlink support

@@x32i77 I don't, sli will make some rich and stupid customers confused and will take away some of that quadro shine away lol

You're badshit insane!

@@x32i77 pcie gen5 is capable of being equivalent to nvlink, said nvidia

@@user-wr5we2xg6h doesn't change the fact that we will never see sli again

It's not an overkill for his use case, I would say 2-4 high end consumer GPUs is the budget option.

@@TheGargalon Incorrect. A Tesla V100 has 3 times the floating point calculations and more Vram. This is an epic mispend. By the time you add the watercooling loop, electricity and PSU costs, 2 x teslas would be double the performance and more efficient.

@@BenState Which floating point exactly? The guy wrote that his use case is AI, and assuming he means generative AI training and inference, we are only looking at single precision performance, and the 3090 is twice as powerful and MUCH cheaper (if you can even find V100s now). You would get a V100 only if you need double precision, which is almost never used in AI. So "incorrect" back @ you.

@@TheGargalon Lol, you have no idea, stop reading redit. plenty of V100s available, and the 3090 is not twice as powerful. Lol.

@@BenState Go check the numbers for FP16, FP32 and FP64 processing power for both cards, read my previous message slowly, then come back to have an argument. So far there is no point arguing with you.

Ну это при максимальной нагрузке, а в простое он 100 ватт берёт не больше, в игре 1 киловатт, его загрузить мог только майнинг который умер

Мне видеокарта бабу заменила

Ты охуеешь когда придёт время платить электричку (если тебе для игрулек только) Если для работы, то поебать (но проводка и розетка нужна хорошая ))))

Почему мечтать если уже процы на али дешевеют i9-7900x стоит 150$ 10 ядер 20 потоков на 2066 как на видео. Осталось только мат. плат дождаться Китай и всё 2011-3 сокет не нужен станет)) Оперативка также дешевеет DDR4 так что год два и он доступный станет как в своё время 2011 сокет.

@@alexs9607 Только зачем? Если есть i5 12400f с меньшим нагревом и примерно той же производительностью )))

Deepfakes as well as blender animations.

deepmind probably killed folding at home. at least not one will be folding proteins.

Why were you building a rig with multiple 1080s? Like what was the use case I'm just wondering? From what I know most people doing this with rtx cards and aren't using server spec cpus are doing this for ai workloads. If using a server cpu there's a possibly of making a cloud gaming server also. But multiple 1080s is a weird set up since there's no ai rt cores so you wouldn't be using it for ai workloads I wouldn't think, we're you doing some super intensive video production or animation workloads? I'm just wondering because I'm still in the learning phase when it comes to the different use cases of gpus. I personally have a machine still running a 1080ti myself and it's been for a long time one of my favorite gaming set ups up until the 40series came out there was literally no reason for me to upgrade since the 1080ti still outpaced a rtx 3060 but now with the 40 series the cheapest 40 series would beat my 1080 ti so I finally had a reason to go for a upgrade. Going to get a 4080 pretty soon or possibly a 4070 super not entirely sure yet as I like to buy used and get good deals. Got my i7 8700k and 1080ti pc fully set up and working for $300 back in the day when it was likly worth over 1500 since then I told myself I'll never buy new again. First pc I bought where I felt I didn't get taken advantage of for once.

@@SpreadingKnowledgeVlogs I was watching a youtube video, maybe on jayz2cents where he built that massive Corsair case I called " the refrigerator." While walking in Fry's here in Houston, I saw the exact same case someone from California returned, and bought it. It took a year to finish since I was building it for vm's, crypto, gaming, etc. With the diminishing returns of SLI that's the only real use case. I ultimately replaced all four 1080ti oc founders with two 2080's. I ended up tearing it down since it seemed like a good idea at the time, but it was larger than a small refrigerator. It wouldn't fit under my desk and if I put it on my , it was a frig, lol. Now, I have an msi meg z790 ace now with 192gb of ram, a 4090ti oc, and a i9-14900k. Loving pci 5 and the Crucial pci 5 2tb nvme drives that scream! :) Good luck in your journey, remember it's a marathon and not a race. I've been building computers in some form or fashion since 1980 when I was nine.

@kvizena those crucial pcie gen 5 drives are insane!!!! The heat sinks on those things is daunting. I run crucial and samsung drives in my pc as well. I'm actually running nvme expansion cards in my pcie x4 and x8 slots. I have so many drives hooked up to my current pc it's kinda crazy. I unfortunately have a older set up so pcie gen 3 but still that works well for my needs. I want to eventually set all my drives into a raid 0 set up and see what kind of speeds I can get. I have several sata drives as well. Pc games take up so much damn hd space and I can't stand anything slower then Sata 3 speeds when having to transfer data around. Thankfully the only drive I have slower then sata 3 is my main back up drive for everything that runs off usb 3.2 which is slow but that's OK since it's not really running anything besides backing up stuff in the background.

concordo

Pra mim que sou pobre só a 3090 já era o sonho mesmo 😂😂😂

@@pedrohaa2012 Sim eu tambem estou pobre e esse pc do video é exagerado, pois 1 rtx 3090 já é suficiente!

Or minecraft! 🤩

@@MintCubaN why would he need all these parts for that

XDD

that pc is not for gaming, because 99% of games can run only in 1 gpu

@@flvmyyy yes, very few game can run on SLI

games will use 1 gpu and the rest will afk

@@bolsterwow hahahaha useles afk gpu

all others 3 are mining))))

and a third of a Tesla V100

только вот зачем такое делать если на 4090 нет драйверов для системы sli и значит фпс в играх будет такой же как и от одной?)

​@@user-gu9hy9iy5uэто 3080

@@user-gu9hy9iy5u ты забываешься, собирал китаец а у них есть все ))) даже если нет у производителей еще

Melting problems

just use another, and hope to remember to manually turn it off and on.

he probably has 4* 4090 by now

A 5800X3D would be bad for this build. The build needs more PCIe lanes than a normal gaming CPU. That score would be low on a single 6950XT as well. Its that bendmark was multi gpu the score would be much higher. I don't believe the build is for gaming, not saying he cant with that hardware. The 3090 has more VRAM than the RX 6950XT and thats what is needed as well. Rendering and AI workloads. This build would be cheaper. CPU might not matter too much.

​@@GodKitty677 Yeah about PCIe-Lanes and the things with doing creative stuff besides gaming, you are totally right. This thing surely is not made for gaming alone, would be stupid.

Just that? I thought would be like 2500w

Dude has 2 1600 psu installed

xD

Yeah, a newbie build for sure

So siehts aus

high??they are half the price of a year ago...

You can't use multiple cards in games

good server chips are very expensive, much more expensive than HEDT ones.. Data centers can pay those prices, but if you're just a guy with a workload that needs that type of rig, you're better off with smth like current gen threadripper pro, those beasts have 128 4.0 lanes, you can have something crazy like 4x data center grade 32x gpus running full bandwidth..

@@vinylSummer You need such a throughput only if your tasks require extensive RAM GPU RAM exchange. AI or Rendering worlds do not require such a bandwith. It is ok to wait until a scene description or AI model initial values are uploaded to every GPU and then the math started producing minimal data to put back to RAM. PCIE-E x4 or x8 is ok for this.

There are no xeon compatible motherboards with 4 PCI-E x16 slots (I mean slot dimensions, not the PCI-E lines). There is a number of motherborads with a lot of x1-x4 slots, but for such a rig you will get the problem placing all GPUs in a case and water cool them at the same time.

Watch yt videos

@@SiuuuuTaNoodle in 1080p x 720

It's made for deep learning

@@TheGargalon If it was he'd use Tesla V100s or similar. This guy doesn't have a clue.

watch gamers nexus's video about it

@@ballsconsumer77 RTX4090 wins... several games not a hav good optimization for a mult gpus

@@germano3571 oh didnt notice he said 4x 3090, yeah barely any games support SLI let alone optimized for it

MUST NOT be tested in games. Just two sticks of 16 GB of RAM and a RTX 3090 will last him for a very long time, so tests should take place in benchmarks or very demanding ones for rendering programs

@Xicore Money must not be an Issue eh?

@@xUltraShadow yeah theoretically 2x 4090 so 4k 144 should not be a problem

No simply get the MSI suprim liquid x cooled aio card

don't buy a 4090 to begin with lol...wait 6 months and it'll drop by $400-500 cause nobody is buying them

@@funtourhawk pre orders are already sold out, i dont think you know what ur talking about. All you guys say the same thing about every generation.

Rog strix

⁠​⁠@@funtourhawk1 year later and the 4090 is $1000