Thread 106481978

>UPGRADE & BUILD ADVICE
Post build list or current specs including MONITOR: https://pcpartpicker.com/
Provide specific use cases
State BUDGET and COUNTRY or you will NOT be helped
Guide: https://igwiki.lyci.de/wiki/Build_a_PC

>CASE
mATX: AP201, Lian Li A3, O11 Air Mini, XT M3, CH260
ATX: XT PRO (ULTRA), AIR 903 Base/MAX, Lancool 207, Flux Pro, Y40, Meshify 3, 4000D FRAME
Dual Chamber: Y60/70, O11 Vision (Compact), Antec C8

>CPU
Budget (inc. gaming): 7500F, 7600/X, 9600/X
Gaming: 9800X3D, 7800X3D, 9700X
Workstation: 9950X, 9900X
Premium: 9950X3D
Intel: If deeply discounted (265K for value & reliability, 14400F budget, 14600K midrange)

>COOLER
AIO: Thermalright Frozen Edge, Cooler Master Atmos, Liquid Freezer 3/Pro (unhandy mounting solution)
Dual tower: A620 PRO SE, Peerless Assassin 120 SE V3, NH-D15 G2
ITX/>42mm RAM: Mugen 6 Dual Fan, AXP120-X67

TIM: MX-6, Duronaut, PTM7950 (phase change)

>MOTHERBOARD
Avoid ASRock AM5 boards, CPU degradation issues still reported.

>RAM
DDR5: 2x16GB or 2x24GB, 6000CL30 (AMD), 6400CL32 (Intel)
Workstation: 2x 32GB (budget), 2x48GB or 2x64GB (high-end)

>SSD (Keep firmware updated)
Budget: SN7100, NM790
High-end: SN850X, 990 Pro (Windows)
Premium: SN8100
https://borecraft.com/

>GPU
Budget: Arc B580 (for new builds), used card
Midrange: 9060 XT 16GB, 5060 TI 16GB
High-end 1440p & Midrange 4K: 9070 XT, 5070 Ti
Workstation: 5090, used 4090 or 3090 Ti

>PSU
Buying guide:
https://hwbusters.com/best_picks/best-atxv3-pcie5-ready-psus-picks-hardware-busters/

>MONITOR
1080p: KTC H24T09P, ASUS XG259CS
1440p: KTC H27T22, Gigabyte M27Q, ASUS XG27ACS, KTC M27T6 (miniLED)
4K: Gigabyte M27UA/M27UP, KTC M27P6 (miniLED), ASUS PG32UCDM (QD-OLED), XG32UCWMG (WOLED)

>OS
Activate Windows @ >>>/g/fwt

>CASE FANS
Meta: Case with good stock PWM fans
Midrange: Arctic P12 Pro, P14 Pro (5-pack, loud @ higher RPM)
High-end: Fractal Momentum, Noctua G2 (140mm or 120mm)

Prev: >>106472558

>News & Rumors
RTX 50 Super: might launch in Q4 '25, 24GB VRAM for 5070 Ti & 5080 Super, 18GB for 5070 Super.

>Omissions
RTX 5070: buy if it's reasonably priced & if you plan to work around potential VRAM bottlenecks.
RX 9070 non-XT: worth considering if it's significantly cheaper than the 9070 XT.

Intel 13 & 14th gen i7/9 -K CPUs: BIOS update required due to reliability issues. Recommended to set a CPU VR voltage limit (1.4 or 1.45V).

>Product Notice
PNY RTX 50 cards have QC issues (rattling fans, reports of 5070 & 5090 with blown capacitors)

Gigabyte RTX 50 & RX 9000 video cards: thermal gel has QC issues, Gigabyte claims only early production batches were affected.
https://www.igorslab.de/en/update-on-thermal-putty-on-gigabyte-graphics-cards-after-further-storage-oil-separation-to-an-alarming-extent-and-silence-as-a-strategy/

>PSU Reviews
https://www.tomshardware.com/pc-components/power-supplies
https://www.lttlabs.com/
https://www.kitguru.net/reviews/?category_name=power-supplies
https://www.techpowerup.com/review/?category=Power+Supplies

For tier-list enjoyers:
https://docs.google.com/spreadsheets/d/1akCHL7Vhzk_EhrpIGkz8zTEvYfLDcaSpZRB6Xt6JWkc/

>Storage Reviews
https://www.phoronix.com/reviews/Storage (Linux performance)
https://www.tweaktown.com/cat/storage/index.html
https://www.techpowerup.com/review/?category=SSD
https://www.tomshardware.com/pc-components/storage/ssds/reviews
https://www.servethehome.com/category/storage/

>Monitor Hunter
https://docs.google.com/document/d/1illeNLsUfZ4KuJ9cIWKwTDUEXUVpplhUYHAiom-FaDo/

>X99 + 9070XT
only in Brazil would you think of this

Is it just me or do a lot of PC parts tend to have issues ? It's not just Nvidia, AMD and Intel have them too. Sometimes it's CPUs, sometimes it's motherboards, sometimes it's GPUs or SSDs. It feels cyclical and like nary a generation of any product really releases and doesn't have some kind of issue.

>>106482121
its just you

>>106482121
yes. the last time a highend gpu didnt have any issues was the 3090ti, and that was only because it was basically a second revision 3090 that fixed all the flaws of the original like overheating vram, failing vrms and high temps/noise
tl;dr always avoid the 1st revisions of any products to avoid design flaws

So apparently jedec 7200 cl64 is a thing now

It's over for Radeon

RTX 5080 AND RTX 5070TI to be discontinued NEXT MONTH

Don't get too much buyers remorse now

>>106482230
>discontinuing cards that just came out
How about not releasing gimped gpus in the first place.

>>106482283
Thanks for beta testing!

if they have access to 3gb vram chips why would they make worse gpus with 2gb ones

so that they can sell to the consumer twice in one generation, think of the shareholders you anti-semite

>>106482296
>why would they make worse gpus with 2gb ones

it's crazy that the 5090 is stuck with 2gb modules and they want 4X the 5090 MSRP for the one with 3gb modules

>>106482330
The greediest companies make the most amount of money while offering the least in return. Every single time. Though it turns out in the PC market space that's everyone.

>>106482296
Those 24Gb GDDR7 chips weren't readily available last year, reserved for 5090 laptops and some workstation cards.

>>106482121
Not having one company like Nvidia release an all in one pc on the same mobo. APU, GPU, CPU, you put a flat vapor chamber with air cooling going right through it, in a small box.

>>106482359
nvidia prob has a warehouse full of 3gb vram chips that they coould've used but opted not to

>>106482336
>Though it turns out in the PC market space that's everyone.
Except thermalright
But yeah out of literally every hobby I've been in. The PC market is the absolute greediest.
There is also something about the PC market where it seems to have the biggest group that gets hung up on minor performance differences.

Any other hobby I've seen debates literally spanning a decade about what's the best only for the very vast majority of people to buy from whoever is giving the best value at the time.
God I wish PCs where still like that.

>>106482368
Samsung's press release for 24Gbit GDDR7 said they're ready to produce those ICs in early 2025.
The launch batches of RTX 50 cards were being made in Q3/Q4 '24.
https://semiconductor.samsung.com/news-events/news/samsung-develops-industrys-first-24gb-gddr7-dram-for-next-generation-ai-computing/

>>106482398
I dont understand what "24Gbit GDDR7 " is exactly
5090 launched with 32gb gddr7 using 16x 2gb modules
then they have the pro AI chips using 3gb modules

>>106482396
>the PC market where it seems to have the biggest group that gets hung up on minor performance differences.
cars are the same and there's a lot of weird overlap between these spaces

>>106482330
if only there was a leaked 5090 vbios that supports 3gb modules so it can be modded in easily

>>106482440
>I dont understand what "24Gbit GDDR7 " is exactly
The GDDR7 chips with 2GB capacities are 16Gb models released in 2023, 24Gb GDDR7 chips with 3GB capacity are a different product.

>5090 launched with 32gb gddr7 using 16x 2gb modules
>then they have the pro AI chips using 3gb modules
The RTX Pro 6000 Blackwell cards were launched later in March '25.
NVIDIA was prioritizing limited 24Gbit GDDR7 supplies to workstation & high-end laptop models.

>>106482487
>16gb models
is it not limited by how many 2gb modules you stuff in? Like 5070 has 6, 5070 ti has 8
what if you put in 10 to make a 20gb chip? Does it stop working? but 32gb doesnt because its 2 times 16 somehow? Whats the 16gb module?

>>106482230
Wtf would the performance uplift be coming from? Is he just making things up again?

>>106482226
i've seen a ramkit come with 884 trfc and im like damn they will really do anything to get ryzen to boot

>>106482553
overclocking the core (upto +15%) and shipping with faster vram (upto +5%)

Treat that VRAM like its gold and product time limitation.

>>106482318
Of course they have, 5090 is an AI card that is selling like hotcakes, especially to China, and it's included in that revenue. Also a lot of people just can't be arsed waiting anymore for an upgrade.

>>106482553
>Wtf would the performance uplift be coming from?
magic. like how 4070 ti super hyber turbo was so good
https://www.youtube.com/watch?v=lDpr84DK7xQ

>>106482514
The number & type of GDDR chip you can pair to a GPU depends on the GPU's total memory bus & the PCB design.
The GPU's memory bus width determines how many memory lanes (& physical chips) can be set.
A GDDR memory chip has a standardized 32-bit memory interface width.

So a 5090 with a 512-bit total memory bus can be specced with 16 GDDR7 chips (16 x 2GB = 32GB model).
For the RTX Pro 6000 Blackwell, NVIDIA is using clamshelled 24Gb (3GB) GDDR7 chips = (16 x 3GB x 2 sets = 96GB)

For your 5070 example, this GPU has a 192-bit total memory bus, so it's expected to use six GDDR7 chips (2GB or 3GB variant).
That's why the 5070 has a total of 12GB VRAM, and the 5070 Super is rumored to have 18GB.
Increasing memory capacity by clam shelling is out of the question for the 5070.

Asked a while back about a deal for a 5060 ti. Got it but I wanted to make sure about the PSU since the MSI variant needs a 12 pin connector instead of 8. I have a Straight Power 11 850w power supply, is the adapter that comes with the card enough and or should I upgrade the PSU as well?

>>106482553
There's a decent amount of overclocking headroom left in those chips, which would explain at least 5~8% of the attainable performance.
The 5070 Super is rumored to be getting a slight GPU upgrade.

>>106482646
>The number & type of GDDR chip you can pair to a GPU depends on the GPU's total memory bus & the PCB design.
okay so design it to allow 20gb using 10x 2gb? No?
>The GPU's memory bus width determines how many memory lanes (& physical chips) can be set.
same as before
>A GDDR memory chip has a standardized 32-bit memory interface width.
hmm

>5090 can be16 GDDR7 chips
>6000 clamshelled 24Gb (3GB) GDDR7 chips = (16 x 3GB x 2 sets = 96GB)
neat

>For your 5070 example, this GPU has a 192-bit total memory bus, so it's expected to use six GDDR7 chips (2GB or 3GB variant).
they can design it differently then?
>5070 Super is rumored to have 18GB.
a bit late now that millions of people have a shitty 12gb gpu
>Increasing memory capacity by clam shelling is out of the question for the 5070.
so they designed it wrong and now they're stuck selling wrong gpus

>>106482666
The adapter bundled with the card is fine; the 5060 TI doesn't draw enough power to be a danger to most people.
Server cards with those power connectors consume more power than a 5060 Ti.
Read the recommended bend clearance on this page, it applies to adapters too.
But Seasonic is sharing conservative guidance.
https://knowledge.seasonic.com/article/72-psu-recommendations-for-nvidia-rtx-4000-cards

>>106482680
>>106482553
its only 10% uplift from oc. this has been common for gpus for a long time. it wont change anything unless your gpu is already hitting below 60fps as a base because that means framegen is dipping below 120.

>>106482697
>okay so design it to allow 20gb using 10x 2gb? No?
The GPU chip needs a 320-bit memory bus for that 10x 2GB configuration.
Even the 5080 doesn't have that much bus width.
The memory bus is a physical interface in the GPU.

The goal of every GPU designer is to sell the smallest possible chip, you cut down on memory bus width to achieve that.
AMD has similar plans with their next set of GPUs according to >>106475685

>>106482723
Thanks anon.

https://pcpartpicker.com/list/ZKJZYd
heres babies first build
im open for 1440p monitor suggestions

>>106482794
so what was amd smoking when they are designing the xtx

>>106482697
>so they designed it wrong and now they're stuck selling wrong gpus
The 5070 is the most popular RTX 50 series card according to the latest Steam Hardware Survey.
It's fine for normie gamers, even for most people in /pcbg/.
The settings that usually push the 5070's 12GB VRAM over the edge are too demanding for the 5070 to performantly render.

Watch NVIDIA market the expanded memory capacity for local AI projects.
They have been pivoting GeForce to be relevant to non-gamers.

>>106482831
AMD was prioritizing a GPU design that minimized the size of the 5nm Graphics Compute Die while trying to compete against the 4090 & 4080.
Smaller 5nm chips needed for Radeon cards means more TSMC 5nm wafers are available for Epyc & Ryzen CPU chiplets.
AMD gave more VRAM than the 4080 because that's how they compensate against GeForce cards (RX 480 & 580 8GB vs 1060 6GB etc.)

>>106482857
If you go into any GPU heavy college course in the country they basically forced you to buy nvidia and that was before the AI boom it's 10x worse now

>>106482831
the good shit that said 8K gaming was what it was made for

>>106482889
rdna3 was a mistake
rdna4 should have been released 2 years ago

>>106482121
Because PCs are made of a bunch of different parts that need to work together and the people making all the parts don't communicate very well with each other. Also they want to cheap out to make more money.

>>106482831
they used chiplets instead of monolith and gddr6 instead of gddr6x to cut down on costs.

>>106482805
Region? US?
9600X doesn't need an AIO, I'd get an aircooler and spend the savings on upgrading the storage capacity to a 2TB NVMe drive.
https://pcpartpicker.com/product/MFvD4D/id-cooling-frozn-a620-pro-se-58-cfm-cpu-cooler-frozn-a620-pro-se

You might have to flash the board BIOS before the board can work with a Ryzen 9000 CPU.
Not difficult, but follow the instructions carefully.
https://www.youtube.com/watch?v=iTkXunUAriE

There's no way the 5070ti super will go for 750$, the cheapest model in western Europe is like 830€ or something

>>106482976
stop kvetching about prices and just pay for the premium

>>106482943
Top 10 gpus on the planet speed wise:

GeForce RTX 5090
152%
GeForce RTX 4090
116%
GeForce RTX 5080
100%
GeForce RTX 4080 SUPER
89%
GeForce RTX 4080
89%
Radeon RX 7900 XTX
88%
GeForce RTX 5070 Ti
87%
Radeon RX 9070 XT
83%
Radeon RX 7900 XT
GeForce RTX 3090 Ti
75%

>>106482995
lopsided performance between raster and rt/pt and lack of upscaling destroyed any value proposition of rdna3

>>106483003
lack of ai upscaling*

>>106483003
>>106483007
transistor count is king
thats why 5090 and 4090 leap ahead
those are all the top 10 fastest gpus on the planet

>>106483003
people who spent 1k € on an amd gpu that couldnt do anything besides run games at native at higher fps than competition were insane

>>106483015
you buy based on transistor count. anything more or less is stupid.

>>106482994
Pay for my premium dick nigger

>>106483013
>>106483019
>transistor count is king
>you buy based on transistor count
Except the 7900 XTX has more transistors than the 4080/Super & 5080, and the 7900 XTX is slower than all three of those cards according to that "fastest gpus" post.
In fact, the 7900 XTX's GCD itself has almost as many transistors as the entire 4080 or 5080 GPU.
Inefficient.

https://www.techpowerup.com/gpu-specs/radeon-rx-7900-xtx.c3941
https://www.techpowerup.com/gpu-specs/geforce-rtx-4080.c3888
https://www.techpowerup.com/gpu-specs/geforce-rtx-5080.c4217

How much performance am I losing without Above 4G and Resizeable BAR? Win10 RTX 4080.

>>106483062
they must have wasted a lot of transistors on the IF hubs in the gcd and mcd which did zero for performance and only increased power draw, just to save a couple of pennies

>>106482994
He was totally right though. A 2080ti or really anything from the 2060 super onwards is still good to go today, 1080ti and 5700xt or vega not so much.
Blackwell is going to be the new baseline like Turing was
>>106482995
You max out the memory slider on a 4070ti or 5070 and they beat the 3090ti at 4k. At 1440p and 1080p they're already faster by default.
>>106483003
No DLSS alternative is/was a way bigger deal.

>>106483082
Depends on the game, but GeForce cards are generally less reliant on ReBAR because of their larger on-die L1 & L2 cache.
You won't lose much performance running the card @ Gen 3 x16 either.
https://www.techpowerup.com/review/nvidia-geforce-rtx-4090-pci-express-scaling/28.html

>>106482230
>rdna5 / udna planned for april 2027 mass production
holy shit radeon is dead

>>106482121
>>106482964
It's something that people need to realize.
No one selling you components guarantees you a working PC ultimately.
Companies will do whatever to get away with selling things that may not actually work correctly with other parts because as far as their concerned their liability ends on the specific component.
Mobo vendors for example doing whatever with CPU, power limits, voltages, etc. As long as what they are doing isn't damaging their motherboard, they couldn't give two shits about how the CPU or RAM is doing, it's not their problem.

You also see this with RAM, you'll find kits that are rated at a much higher speed than what any OEM is willing to offer and it's on the basis that the DIMM vendor isn't responsible for the stability of your PC. Their liability purely ends at the RAM itself and it's up to you to determine if their RAM is going to be an issue.

>>106483124
amd is a console silicon manufacturer
all gaming related improvements are done to appease sony, i guess microsoft too but they might not even make another console

ps6 comes out in 2027 2028?

>>106483124
but he did say the (3.7ghz+) 9080xt with 24gb of gddr7 will be released in the meantime as a stopgap until rdna5 is ready

https://starforgesystems.com/products/voyager-iii-ultimate

So guys I haven't had a gaming PC in about 10 years and I'm too scared/retarded to build my own.
The very best one is 1000$ off for the labor day sale, just wondering if this is actually a good deal by any means.

For $5,299.99 USD:

CPUAMD Ryzen 9 9950X3D
CPU CoolerStarforge Custom Bitspower 360mm Liquid Cooler
RAMTeamgroup Delta RGB 64GB DDR5 6000 CL38 (2x32GB)
MotherboardMSI B850 Tomahawk Wifi
Graphics CardGeForce RTX™ 5090 32GB

Its probably overkill, but I just want something that is future proof and can run anything for the next 5 years. I finally have the money and ready to pull the trigger on this. Bad idea?

>>106483158
whats your annual income?

>>106482970
germany
the aio was pretty cheap for around 70 bucks here, so i thought why not
i have another 1tb nvme ill plan to use for secondary storage
i have this amazing luck where the big local hardware store has all monitors on display turned off, and the ones that are turned run some crappy 720p advertisement video so i cant really compare them by eyegpv

>>106483158
>64GB DDR5 6000 CL38
For that price? Seriously?

>>106483158
get 6000 cl30 ram instead of cl38
be careful of the 5090's power draw since it is very prone to melting from heavy use. the only real way to prevent this currently is to get an astral with per-pin current monitoring. its more sensible to just get a 5080 that pulls 2/3 as much power which doesnt push the 12v-2x6 connector to its limits

>>106483185
>5080
and thats also a poor choice when the 5070ti is 85-90% as fast for a $250-400 less

>>106483158
>streamer pc company

>>106483158
you should ask the prebuilt PC general

>>106483165
About 25k a year as a line cook, but my rent is cheap at 600 a month.
I just really want a good PC more than anything in the world right now.
And im planning on getting into making Indie games as a hobby (and to maybe get out of my shitty career choice)

>>106483222
wat
buy a <2k € pc with 5070 ti and 9800x3d and build it yourself

>>106483185
>since it is very prone to melting from heavy use
More ones are working fine than ones burnt, + source on claims?

>the only real way to prevent this currently is to get an astral with per-pin current monitoring
literally only warns if uneven power draw? + source on claims?

>its more sensible to just get a 5080 that pulls 2/3 as much power which doesnt push the 12v-2x6 connector to its limits
even lesser power hungry 5000 series have burnt connectors too + source ?

>make an all black linux build
>get cucked by the GPU RGB not being controllable in any way

feelsbadman

>>106483158
I genuinely stopped reading at B850 Tomahawk

you must be genuinely retarded to pay $5000+ for a prebuilt with a B850 in it

>>106483229
i have a bridge to sell you if you think a 5090 can last 5 years with how its built

>>106483236
just disconnect the rgb

>>106483250
literally everything he/she/you/xir/xem said is verifiably false though
you can begin with that

>>106483260
you even contradicted yourself in your own post. you have nothing to add to the discussion

>>106482805
What other anon said. Cursed price though, change something to fix that please. I tried 2 1440p KTCs for my AM5 build and they're solid monitors. The Miniled and IPS ones

>>106483192
Can say the same thing about the 70ti vs 70 (if you ignore the vram)
>>106483185
>be careful of the 5090's power draw since it is very prone to melting from heavy use
You're exaggerating, how many cases in the wild are there of burnt cables this gen now? I bet you can still count them on one hand.

>>106483115
That's good to know, thanks.

>>106483266
your concession has been accepted

>>106483158
I'd rather ask a technician @ Best Buy, Microcenter, or a local PC shop to build the PC from the parts you've bought (and get the 3/5 year repair/maintenance warranty).
https://pcpartpicker.com/list/kVDyC8

>>106483177
Check reviews from these options, someone recently bought the G50D and he really likes it.
https://de.pcpartpicker.com/product/9xLdnQ/samsung-odyssey-g50d-270-2560-x-1440-180-hz-monitor-ls27dg500enxza
https://de.pcpartpicker.com/product/k8GnTW/gigabyte-m27q-270-2560x1440-170-hz-monitor-m27q
https://de.pcpartpicker.com/product/jXmNnQ/lg-27gs75q-b-270-2560-x-1440-180-hz-monitor-27gs75q-b
https://de.pcpartpicker.com/product/MsQKHx/asus-tuf-gaming-vg27aql3a-270-2560-x-1440-180-hz-monitor-90lm09a0-b013b0
https://de.pcpartpicker.com/product/JPKnTW/dell-alienware-aw2725dm-270-2560-x-1440-180-hz-monitor-aw2725dm
https://de.pcpartpicker.com/product/pGqBD3/asus-tuf-gaming-vg27aq-270-2560x1440-165-hz-monitor-vg27aq

>>106483185
>>106483228
>>106483248


Ok thanks guys I guess Ill hold off, like i said im retarded and havent built a PC in 10 years.
But whats so bad about the B850 Tomahawk just curious.

>>106483158
Is this bait? That's a 3000 dollar PC for 5000+ dollars, and most of that cost is just the GPU. That RAM is garbage, that motherboard is a budget board and a 9950X3D will toast the VRMs on that thing up

>>106483158
It's not a good deal, but it is a pretty killer PC which is what you want. Who cares about the mobo, he's not going to be tinkering around inside anyways

>>106483222
>>106483228
Yeah, I don't know how it works in the US but in the EU, you can just pick your parts and have the different pc stores assemble it for you for like 70-100 euro. But I would advice to just assemble yourself, it's never been easier and theres millions of videos on how to do it on youtube. Put the rest of your money in an indexfund so you can buy a robowaifu in 15 years.

>>106483283
>You're exaggerating, how many cases in the wild are there of burnt cables this gen now? I bet you can still count them on one hand.
I was thinking long term like >3 years on. We're already seeing a lot of reports of burnt connectors on 4090s that were working fine the last 3 years and were only found out because their owners taking out their cards. With regards to connector's safety margin, the 5090 is much worse than the 4090 as the latter only really used 410W at worst and averaged in the mid/high 300s in power draw while the 5090 is often running at 550 if not maxed out at 600w while gaming in high loads.

>>106483311
>>106483323

Ok thanks all check that out. Seems like a way smarter idea.

And not bait, Im retarded. Thanks for saving me from doing something stupid.

>>106483315
no idea, its a great board. it just doesnt look that nice compared to fancier ones on the market

>>106483258
theres no rgb specific connection

>>106483380
Are you competent enough to install the CPU cooler & video card?
If you get someone else to build the PC, you need to safely transport the completed build.
Make sure the PC is securely packed so that the video card won't sag/break while transporting it.
If you get an AIO, the weight of the CPU cooler is less of a risk.

>>106483236
are you super sure? what did you try?

>>106483236
Try openrgb
But if it's a newish model or one of the less popular ones, it won't be supported

>>106483478
running the gigabyte software through wine led nowhere, then I tried openrgb and it doesnt recognize it. Also shutting down rgb in the MB does nothing

>>106483505
yeah its a 9060xt, guess im just stuck waiting for an update

>>106483454

Id probably have someone else do it, but ill take my time and research a bit more. And Ill keep those things in mind.
For right now I just wanted to know if the Starforge is a good deal even though its 1000$ off and the answer seems to be no.

>>106483511
try opening an issue on their gitlab then. all you need to provide is the pcie id and someone will add it to the whitelist.
since its a gigabyte and not a lesser brand like pny or kfa2, the rgb logic is already implemented and the device just needs to be whitelisted

MY COMPONENTS ARE GETTING MORE EXPENSIVE BUT I'M GETTING PAID NEXT WEEK
REEEEEEEEEEEEEEEEEEEE

What are the benefits here and can I have a modern Mobo and CPU, SSDs and all that but also have the old GPU?

>>106482995
*sad RTX PRO 6000 96GB noises*

>>106483607
2 months for "Black Friday" my guy.

>>106483642
I ain't waiting that long
plus it's a meme outside of US

My 5070 is obsolete

>>106483315
It's a fine board for a $1500-$2000 build. Great mid-range. A $5000 build with a 5090 should be on an X870E board for that. Gen 4 SSDs when Gen 5 is available in that build too, the RAM...basically they put a 5090+9950X3D in that build, then nickel and dimed you on every other part in it.

>PS5 raises price and lowers the size of SSD from 1TB to 825GB
ah good, they are getting screwed too, it's not just us

at this point, 5k 27" qdoled might be the first usable oled on the desktop due to the ppi
like literally you need this minimum or better to use it on the desktop, nobody should have even made oled monitors before then
maybe woled is less retarded but people complain about it too, maybe it does not fry your pupils and just looks bad idk

>>106483716
the old one was 800 the new slim was 1tb though or did the swap back

>>106483684
literally unplayable

>>106483757
Yes

>>106483777
why do you post unoptimized settings for low end gpus
this isnt /v/

>>106483796
because it's ogre

>>106483829
guess I'll be playing the game with mere >100 fps on my system

>>106483841
my god how will you last with that torture

How do I go about attempting to setup a new monitor (AOC Q27G4ZR) without a colorimeter? I don't do graphic design or video editing but can appreciate somewhat accurate colours. The monitor is kinda new and not got a lot of reviews (normally one of the good monitor sites has a "recommended settings" bit in a review). As for the monitor itself, it's a nice budget 1440p 240hz ips screen, you get what you pay for (£170). It's a massive upgrade over my old XG2402 so I'm happy, even if it's not "amazing" when compared to slightly more expensive offerings.

>>106483777
>>106483829
What is this AAAAAAAAAAAAA game?

>>106483964
Cronos: The New Dawn

>>106483311
Thanks anon

>>106483829
around 30fps is still perfectly fine to have silky smooth gaming experience

I have:
14700K
64GB DDR4
3090
4K 60Hz TV

what would you all upgrade if you were me

>>106484087
buy a 65" qd oled tv thats on sale rn

>>106484087
Get a high refresh rate oled display

>>106484112
which one, father?

>>106484127
s90d

>>106483902
Go through a bit of these tests
https://www.lagom.nl/lcd-test/

>>106484005
If you're the German monitor shopper, consider this AOC monitor as well.
The parent company of AOC also owns the right to produce Philips monitors.
https://de.pcpartpicker.com/product/bGVfrH/aoc-q27g4xn-270-2560-x-1440-180-hz-monitor-q27g4xn

>>106483902
https://www.youtube.com/watch?v=avJTz1JhkR4

>>106484087
go outside nigga

Cumpleted it on Intel Arc.
Im tired of bad games.

>you need 2000$ gpu
I dont trust this slop.

She is dreaming for your Big White PCBG.

>>106484277
why would you ever play a game soo bad the studio that made it was closed within a month of launch? i hope you pirated that goyslop.
>>106484283
she's already balding lmao.

>>106483902
just use your eyes, pick the game that looks the best to you, that has the nicest, and most robust color schemes (Mordhau for me), and pause it in-game and simply adjust the colours until it looks good to you.

Problem solved, you're wildly overthinking it.

>>106484087
300 buck ips 4k 160hz monitor and go to sleep

>>106484073
lmao poor, brown and coping, get a real job Ranjeesh.

>>106484087
your monitor sucks, for browsing the internet alone its awful at 60hz.

>>106483684
>>106483777
>>106483829
The stunning and brave gt 1030 can run it at 720p 40 fps no upscalers.
>>106484300
>closed within a month
Thank God.

>>106484345
>stunning and brave
Truly
>CPU : 100%
Kino

After all this coping, research, meticulous autism, we all went full circle and went back into IPS being the gold standard, default choice

>WOLED
chrominance overshoot or diagonal dithering + fringing still due to non traditional matrix

>QDOLED
insane fringing, something will always be wrong, no ease for eyes focusing due to diamond pixel substructure

>VA
slow smear slop and miniled blooming lag cope

>TN
only zowie 600hz here for 1500€

>IPS
You were at my side all along.... my guiding moonlight

CRT/Plasmafags need not apply

>>106484404
In 2020 onward there was fast VA panels like g7 available and that was the best image quality for gaming and IPS became entirely redundant.

Now the qd oled monitors were much better image quality, hdr + high color volume and richfags bought those immediately.

Because the oleds entered the market and become cheap really fast it didnt allow va minileds time to shine. People tried to sell those for >1k € which made zero sense.

TN remains viable for cs2 players thanks to strobing I guess. IPS is deader than a dead dead guy.

My friend gave me his aurora r10 but the cooling sucks. So I'm thinking of taking out the gpu (rtx 3070), cpu (ryzen 9 5900), ram and drives and making my first pc build. What psu, motherboard, and cpu cooler, and fans should I get?

>>106483902
i just went around looking at pictures and screenshots where i knew what the colors were supposed to look and tried to match those.

>>106481978 (OP)
Finally some good fan LEDs that arent just red.

https://www.techpowerup.com/forums/threads/system-stutters-with-ryzen-7-9800x3d.338907/

things like this scare me about ryzen

>>106484467
its a waste to buy new components to support a mid tier parts from 5 years ago
you'd be better off buying entirely new pc.

>>106484467
the cheapest stuff
the most ram

>>106484467
Learn to undervolt and repaste/repad your parts. Not worth changing it out like other anon said

>>106484480
Should've just manually tuned his 13700K.

>>106483684
What’s the point of the 5070 when the 9070 xt exists?

is the ghosting on this bad boy really that bad ?

>>106484603
https://www.rtings.com/monitor/reviews/samsung/odyssey-neo-g7-s32bg75

seems fine

>>106484432
>In 2020 onward there was fast VA panels like g7 available and that was the best image quality for gaming and IPS became entirely redundant.
Slow + Bent + VA gamma shift view angles + 32" + coatings + Delaminating panels due to Samsung + samsung QC + overpriced

>Now the qd oled monitors were much better image quality, hdr + high color volume and richfags bought those immediately.
I bought two so far, Fringing was concluded being the eye strain problem + if you bought a bad firmware once you get additional aids
Now you have to wait for 220ppi+ to cope, maybe then they will be good

>Because the oleds entered the market and become cheap really fast it didnt allow va minileds time to shine. People tried to sell those for >1k € which made zero sense.
VA remains a highly suspicious buy, they can't match budget IPS, I would never

>TN remains viable for cs2 players thanks to strobing I guess.
strobing also kills eyes, allegedly

>IPS is deader than a dead dead guy.
budget? IPS
mid tier office/Gaymen? IPS
mid tier office/Gaymen+? 4k IPS

to be concluded: Does WOLED fuck up eyes? Now that I've concluded that everything is shit next on my potential tries before I throw in the towel are: LG C4 42", 4k 32" 240/480hz WOLED, 4k 27" QDOLED

I am getting fatigued, I don't wanna try more shit and return it, might aswell pass up on the 27" qd considering it already has fringing, wonder if WOLED isn't cucked, but the TV will have dithering which already looks like ASS, the 32" WOLED will still have worse rendering than a normie RGB IPS and the price is kinda still high too even with codes

>>106484622
>>106484603
that is the 43" VA thingy
and yeah the ghosting is bad

>>106484570
A 9070xt is $150-$200 more expensive in the United States. Was more like $300+ when I bought this, but even if I was buying now I would probably still do a 5070 over anything else.
And almost every single game I play is Nvidia favored (plus I really really like Nvidia pt tech demos) or old/light enough that 10%-25% extra performance means dick because I'm already at 8k internal 100-zillion-fps.

is this a good card

>>106484679
if I with hold my opinions on msi you can consider if you want more than 1 hdmi output because it only has 1
asus has 2
and so on
other than that guess so

>>106484679
5090 FE, TUF, or that Gaming Trio would be my choices.

Does the waterblock in a Gigabyte Aorus 3090 Xtreme Waterforce cool the backplate mounted VRAM or do it not cover that?

>>106484697
>>106484705
using display ports more than hdmi but thanks
what's a good motherboard to pair with it?

>>106481978 (OP)
wot fans? looks cozy

>>106484666
all you play is dx11 shit? because those are all OLD. dx12 and vulkan amd has the 100% advantage.

>>106482396
>Monopolies
Abolish IP and solve this problem. The declaration of intellectual independence will also likely solve the lithography bottleneck for modern civilization as well
>10000 original AMDs all competing with each other and the big firms
As per opposition i could think of;
> The normgroids and the monopolists
Will buy fake apple silicon and be happier than now.
The monopoly firms will go bankrupt and sell their stuff so better hardware can be engineered by actual professionals rather than legal teams with token businessmen and engineers sprinkled in to produce a barely acceptable product.
Respectively

>people calling other poor because they didn't blow as much on PC as them
nobody is obligated to match your retardation

>>106484744
Cyberpunk + Silent Hill 2 + RTX remix mods are most of what this gpu has been used for. So 2 games that are significantly better on Nvidia + an entire category amd is realistically locked out of.

>>106484570
drones who slavishly paypig to Njudea because that's who they've been buying gpus from since forever. Also the 5070 should be cheaper than a 9070xt, they're in different performance brackets.

>>106484808
>Abolish IP and solve this problem.
you fucking retarded commies are as obnoxious as you are economically ignorant. While jews and globalists have made a farce out of IP+trademark laws, they are necessary to stimulate economic growth, when it protects the R&D of inventors/corporations that is.

>>106484922
>Stimulate
>Economic growth
That's called malinvestment and it's advocacy is the reasoning that makes sure there's a massive fuckoff economic crash every decade desu.
>R&D
The reason normal people R&D is to make a better product to sell, not become modern disney, you don't need IP to invent anything, infact it is that someone can go to an authority and claim your invention as their monopoly in fraud defended by an army of lawyers that would be a threat against R&D.
Abolishing IP declaws these legal monopolists from staking a claim to your thoughts
AMD would be sued to oblivion and we would likely be using dual cores right now if the "economic stimulators" didn't break their rules regarding Intel in the 70s

Bros Im an idiot. Today my new build was stuck while booting with orange light on the "DRAM" diagnostic led for a few minutes. Idiot like I'm I switched off the psu and when I switched on the PC resumed the POST process without me pressing the power button which is odd. Found out it could be my dram training on AM5 that could take several minutes sometimes. PC seems to work fine now but I'm worried I've fucked up something?

lol
literally every single AMD build

>>106485039
the more ram you have the longer it takes. if you have 32gb on average it takes about 2 minutes. after the first dram training the next boot time is much shorter if you have memory context restore enabled in the bios.

>>106485066
People actually turn off their PC when sleep mode is standard for phones/laptops/etc?

>>106485074
I have 32GB and it was a little longer than 2 minutes. Yeah it boot normally and fast now but if I accidentally turn off power once would it mess something up permanently?

>>106485037
IPs provide financial incentives & protection for R&D.
People want to be well-compensated for their products/services.

>>106484404
alright autism central
recommend me a 27" 4k 160hz monitor with high color accuracy for color sensitive work

>>106484404
>WOLED
>QDOLED
>VA
>TN
>IPS
Aren't you forgetting someone?

>>106485066
AM4 boots in 5 seconds

>>106484743
Looks like Corsair fans.

>>106484743
yeah, I can see the logo
its corsair rs120 argb

>>106485128
IPs provide financial disincentives & monopoly rights for the first legally recognized version of a product, effectively making it anti-capital to be hoarded
People want to be monopolists who ban their competition so that they don't actually have to work again

>>106485339
No one else made a serious alternative to CUDA when NVIDIA was all-in on it because it wasn't profitable enough.
NVIDIA is making a lot of money atm because they were the only ones crazy enough to set themselves up for this moment.

Should i downtier my prospective build from a 9700x to a 7700 and buy a printer or a phone/tablet on the side?
I can also buy a cheaper power supply or mobo
>Downtier PSU is Gamemaster 750w from thermaltake 750w 80+ bronze
>Mobo is MSI 850M-P WIFI to Asus Prime B650M-R

>>106485371
Well, clearly IP hasn't given anyone the incentive to do anything about it while it's abolishment would have almost assuredly resulted in CUDA being open sourced for the benefit of all with Nvidia still making the best products related to it since they have literally invented it and have the first mover advantage

>>106485407
I would be careful with the PSU you choose.
Buy one with at least a 5-year warranty available in your region.

>>106485230
almost choked on my food, thanks anon lmao

>>106485471
There's an incentive, but the rest are too far behind.
NVIDIA essentially created the machine learning hardware market, and they get to reap all the profits until everyone else catches up.
Intel didn't give a fuck about this market, including workstation GPUs in general.
AMD prioritized CPUs over GPUs, and it was the right decision given their limited resources.

Blame everyone else for being hyperfixated on short-term profits; IP is not the main cause of NVIDIA's dominance in AI.

I have a feeling manufacturers make bad amd mobos on purpose. Ever since I switched 2 generation ago I had nothing but problem and it's all on the motherboard's side. My new b650 Tuf come with a audio chip dead one arrival. I use usb DAC so not bother returning but it kills me inside that my brand new build has a pretty major fault.

>>106485553
feels like every amd motherboard either has awful bluetooth or audio

2 weeks until reflex 2, there also files about it on the recent GRD update. AYMDrones cant catch a break lmao

>>106485593
And don't forget the fucking random usb disconnection on AM4 and long ass boot time on am5 in fhe beginning. RAM compatibility can be ass sometimes too. Avoid Asus Asrock Niggerbyte.

>>106485548
Sure, as i said
>They would still have invented it and have the first mover advantage
What I'm saying is that tech in general would be significantly better if there weren't IP. There wouldn't be proprietary, barely compatible standards, software lockdowns or DRM, not to mention things like stop killing games and right to repair movements would be actually solved.
AI for example would be more accessible, better and with more alternatives

should I fork out more for riptide or is assrock b850m pro enough for 9600x?

>>106485663
Except IP protection laws gave NVIDIA the incentive to invest so heavily into CUDA and ML GPU development.
Blame Intel and every other major tech company that could've easily outspent NVIDIA in this market, but conciously decided it wasn't worth the effort (until it was).

anyone idea what keyboard?

>>106485197
>color accuracy
dont u need to calibrate that garbage
idk
I might buy the gs27u

>>106485230
already covered in there

>>106485066
my 96gb system powers on in 3 seconds.

>>106485611
>Avoid Asus Asrock Niggerbyte.
Who is left?

>>106485947
SUPERMICRO

>>106485967
please saar direct me to an affordable AM5 Supermicro board for a 9600X
thank

>>106485986
No.

>>106485066
i unironically made mine do that intentionally. i couldnt keep my toight timings otherwise.

>>106485996
I will do the needful and order an ASRock motherboard, ASUS GPU, and a Gigabyte SSD just to spite you then.

>>106486013
Go for it. Honestly they're all fine if you pick the right model.

>>106486013
dont buy asus

none of you guys told me how hot a 5090 would be

just spent fucking ages trying to get this piece of SHIT to work, the MPG 321URX QD-OLED 27"
long story short, when it powered on all it would say is 'no USB-C cable detected', no matter what I did, whether I connected an actual USB-C, a DisplayPort, HDMI, nothing worked
the obvious solution is to use the manual menu button on the back of the monitor, but that did nothing
am I safe to assume it's a faulty monitor, or did I overlook something? what a waste of time and money

>>106486131
It's a 550W space heater.
You can even feel the heat generated by a 200~300W card.

>>106486259
have you tried restarting the monitor

I thought amd could do ai and that vram was all that mattered? Why was I lied to?

>>106486310
yeah, didn't work. chatgpt said some bullshit about unplugging everything, holding up on the back button, keep holding it while powering it back on and yeah it did nothing

>>106486348
and if you have dp cable on, you press the menu button nothing changes or comes up

>>106486319
btw series 50xx is even faster after the last comfyui update

>>106486357
correct, tried with all different cables, no matter what the menu button did absolutely nothing

>>106486359
Why are ai niggers told to buy a 5060ti over a 5070 then?

>>106486364
rma time

What Build do I need to play silksong? Will a 5080 be enough?

IPS and AMOLED don't fuck up my eyes so I like using those. There is also 80Hz+ CRT but scarce.

>>106485812
IQUNIX

>>106486422
its steam deck verified

>>106485947
AsRock Nova is safe & effective. Works like a charm here

>>106486378
You cant run large models on 12GB, vram is king for any LLM or AI related task. even if the 5070 is faster it basically cant do the same things a 5060ti can do even if the 5060 ti takes longer on simpler task or running smaller models

>>106485759
>nobody would invest in the long term if they don't have monopoly rights
Anon, Nvidia didn't need IP to capitalize on an idea, They would do so because it's a good idea. Ideas aren't scarce resources to be property.
What IP was and is, is a detriment to everyone as even the monopolist will lower it's capabilities as monopolistic practices tend to lower the quality of a firm itself. The moment their IP expires/loses relevance, they would be devastated more so than an actually competitive firm.
Which ties into my first reply as;
>This sort of reasoning leads you to total economic collapses every 10 years or so
Because it's essentially the same thing as
>Nobody would spend any money long term if the money wasn't inflationary
This warrants it's own thread de kashira

>>106486721
Can the 9060 XT be used in this instance?

>>106486917
Depends on what kind of AI task do you want to do, you'll have to run on linux and emulate zluda as rocom sucks, even then anything beyond simple image/text generation will be terrible because most of the nodes and wrappers use cuda, even then the speed on the 5060 TI is leaps and bounds you can check the benchmarks here

https://www.youtube.com/watch?v=BzLaALDc1bw

>>106486969
for local LLM my 9070xt works just fine

>>106486989
i mean of course it works but its just straight up worse than a 5060 ti on anything optimized for CUDA and its only gonna get worse as more models get retrained to use nvfp4, only a complete retard would buy a 9070xt for AI just look at this >>106486319

>>106485037
>massive fuckoff economic crash every decade desu.
again, you're economically illiterate, stop making assertions about a subject you do not understand. Just because you can identify a problem, doesn't mean you're smart enough to think up a viable solution. Keep your commie rants restrained to the tranny subr*ddit you got them from in future please.

>>106486422
you can probably play it on a switch, stop being retarded.

>>106487003
you're pretty much clueless in your assessment. 10% faster gen speed is basically irrelevant to the general experience. it's not something that matters for your normie user for local LLMs. that shit matters at large scale enterprise applications with millions of users, not for home, you clueless moron
it's like arguing 8gb/s ssd speed is a better experience vs 7.2gb/s. it's the same shit moron

>>106487038
are you retarded? is not 10% faster gen we are talking about being able to fit larger models, a 16b model does not fit a 16gb card but a quantized 8b does. Same goes with nvfp4 it reduce a 32gb model to 13gb, im sorry you got scammed

>>106487003
that's comfyui image gen not LLMs. people don't run image gen on their PC, to that degree where twice faster means anything, they are mostly using local LLMs for coding and random shit, not for olympic level benchmarking. they solve some issues, and they're fast enough no matter which ones you use. unless you're making money from that, which is a different application, not normie local use for random needs

>>106487053
I can fit larger models in 16gb it's fine, I even spill into RAM to fit even larger models, at the end overall speed is good enough for coding and shit.
the gains are not linear, once it gens faster than you can read it doesn't fucking matter, practically, if it's faster. diminishing returns and all that you low IQ brainlet

>>106487068
You are stilll stuck using shitter quantization models, no one and absolutely no one would ever buy an AMD gpu for any LLM related task they are that bad

>>106487079
I literally got it for exactly that, runs great on linux vs that shit nvidiot hardware, and is fast enough that I don't really care, and models are already good enough that it's irrelevant, solves my needs as a regular normie with random bullshit coding tasks

>>106487075
Running a Llama-70B inference, NVFP4 uses 20GB VRAM vs. 70GB for FP16 you have no fucking idea what you are talking about

>>106487088
>I literally got it for exactly that
That because you are a moron, you'll be stuck without ever being able to run any good large models for the next 5 years

>>106487089
but the difference does not warrant paying for more that's what I keep saying. it's better marginally for normie needs. holy shit you're so fucking clueless lmao
universities use enterprise pro hardware to run full models lmao

>>106487031
>Problem
Are you saying malinvestment doesn't cause malinvestment which then causes economic crashes once people realize this? Or maybe that these policies take roughly a decade before they affect the entire economy and people realize it was a mistake?
>Solution
The solution is the stopping of economic stimuli, from IP to money printing.
just because it's simple doesn't mean it's wrong, tech would be better it there was no IP
>>106485548
>NVIDIA essentially created the machine learning hardware market, and they get to reap all the profits until everyone else catches up.
Except they banned everyone else from actually catching up with their IP, you have to go through hoops and bounds just to emulate their stuff so things people use work on your stuff as well

>>106487097
a 5060 ti 16gb is cheaper than a 9070 XT you got scammed

>>106487102
but runs games like shit, so I get AI + nice gaming experience + nice linux experience. it's just the smarter buy. morons waste their money on nvidia. but hey that's your choice you do what you want with your money

>>106487097
>hat's what I keep saying.
you have no idea about the difference because you are dumb retard that has not compared what difference running a larger model makes

>>106486906
Without IP protection, businesses would just waitfag and steal someone else's invention. Bad economic incentives.

>>106487107
or maybe you are just a moron that got scammed and has no idea what he is talking about and unironically said retarded shit like this
>I can fit larger models in 16gb it's fine, I even spill into RAM to fit even larger models,

>>106487108
I tried larger models because I have plenty RAM, I put up with shit gen speed just to see the difference, and it was literally irrelevant for me.

>>106487117
ye I have qwen3-code which is 18gb model and gen speed is fine for coding you moron lmao, you're clueless you're a shill you dont' even do any real work with them

>>106487118
maybe a slightly larger model but not one to matter, if you go OOM for even 1GB of vram you go from completing a task in an hour to 2 days, how the fuck can you say retarded shit like this and expect someone to take you seriosuly

>>106487131
>ou go from completing a task in an hour to 2 days
if you have more demanding needs you use free chatgpt and still be cheaper than paying like a pig for useless shit

>>106487125
so we are talking about nvfpe reducing a 70GB llama model to 20gb and you say retarded shit like this because you used a 18 gb model? lmao

>>106487100
>Except they banned everyone else from actually catching up with their IP, you have to go through hoops and bounds just to emulate their stuff so things people use work on your stuff as well
The inventors of CUDA can set the terms of how it's meant to be used; everyone else can make their own programmable shader ecosystem (which they are doing now).
Stop acting like Intel, Microsoft, Google, and Amazon didn't have the capital to make an alternative. Those last three companies are large enough to design their own server chips.
They didn't want to foot the bill in developing the ecosystem for AI GPUs because none of them believed the market would grow this big.

>>106487138
>70GB
>get 5060ti
these are the idiots shilling it here

>>106487137
you are just moving the goalpost because you cant admit your purchase made no sense if you wanted to use AI for the next 5 years.

>>106487145
models are only getting smaller and better that's pretty much clear from last few years, seethe more for wasting your money on irrelevant gains

>>106487143
shilling what? you are an utter moron that was so stuipid that could not differenciate quantized models and thought your card was just slower

>>106487154
thats because they are quantized which is exactly what make your card a terrible purchase for AI holy shit is like talking with a mentally ill child

>>106482121
I've spent at least 8 hours this week troubleshooting when I just want to unwind and decompress by playing video games.

I'm about to give up technology for anything beyond work or absolute needs.

I'm not kidding when I say the last 2 years has seen the most insane dive in quality of products in both software and hardware I have experienced in my life.

>>106487169
works on my machine

>>106487169
what issues are you facing ?

>>106487114
They would make stuff that do stuff
See
>People wouldn't buy anything if the money wasn't inflationary
No they want stuff to use, this would only result in reduced malinvestment and a slower but actual growth of the economy
>>106487142
>They can set terms
Not to everyone, their idea isn't scarce, they can set terms for their own use of their idea but not anyone elses. It should be perfectly legal to reverse engineer, manufacture or modify software and hardware without legal gymnastics besides outright fraud which is ideally covered by laws against fraud rather than IP.
If say, anyone can make CUDA cores that work on CUDA applications but are otherwise completely unrelated to Nvidia, they should be able to do so legally rather than having to come up with a redundant analouge that likely requires the entire software side to be rewritten.

I'm maxed out on my SSD drives and I kinda want to build a NAS at the same time. I had pretty bad experiences with cloning a drive so I sorta want to add on but not sure the best way to do it.
>1. Get a seperate PC to act as a NAS
Put all my movies/shows/tv shows/low spec games/emulator games onto this. I kinda want to build a homelab so this is an option to actually do shit but then I'm like why the fuck would I buy a new PC when I can just buy a new mobo/CPU/ram, and then just add more SSDs to that but it's also kinda like why the fuck am I spending an addition $700+ just to add fucking 1-2tb of space.
>2. Buy new SSDs and clone them
Not a big fan since then I'll just have a bunch of 256/512GB SSDs just laying about
>3. Expand mobo with an SSD hub with a TB3 or USB 3.0 connection
Not even sure how this'll work or even if it'll be fast enough for games.

Any other options?

>>106487169
i learned that most of my winblows 11 problems are solved simply by turning on compatibility mode for any indie or Unity engine game thats giving me issues. Solved them almost entirely.

free AI's like Grok and duckduckgos auto-AI search feature were big helps aswell.

If I upgrade from a 4070 ti super to a 5090, will my computer stop crashing whenever I have browsers, discord, and Unity/ Unreal open at the same time?

should i shill out $700 on that new fangled asus tandem woled? when is it coming out?

>pc hard crashed
>boot file is just straight up gone
>having to jump through hoops using my server to make a recovery drive since I didn't have one on hand
Cool
At this point I think I'm just gonna spend the weekend getting my shit updated to Win11 with a clean reinstall, this Win10 one hasn't been refreshed in ages and it really does show

7900 xtx or 9070 xt?

>>106487665
why are you using discord

>>106487773
cause all the stupid people I know use it.

>>106487770
5070ti

>>106487777
if these stupid people jump off a bridge, would you do it too?

>>106487808
no I'd laugh at them. and finally go back to using proper comm stuff like slack.

>>106487814
nope you'd do it

>>106487773
because after the age of 20 or so most people are over their leet h4x0r phase

>>106487858
its not even that.
no joke i think every channel in discord just had mentally ill people even the private ones. you really want to be john therapist because you fought with his dad again?!

>>106487867
you talk about discord the way tumblr girls in 2014 talked about 4chan

>>106487871
it's totally the case though im tired john man. can we just play some freaking 2k

>>106487242
Windows enumerating my secondary monitor as monitor 1. Applications (that are poorly programmed I assume) read that rather than the primary display flag. It also fucks with other stuff like HDR toggle, refresh rates randomly resetting, etc.

-DSC is fucking with MPO and causing display issues.

-MPO doesn't function when set to 10 bit color depth in SDR

-DSR breaks with DSC in HDR 10 bit and only displays the top left quadrant of what it's rendering.

I basically have a 2nd monitor that causes me grief for no reason. I use it for work, otherwise I'd get rid of it.

-DSR will not function in certain titles, or sporadically. Display turns off, game window shifts to my 2nd monitor then it spergs out and shifts the window back to my primary and undoes the DSR resolution.

-Elden Ring has insane stuttering that was caused by it rapidly enumerating over my USB devices that I finally figured out after having to look at the threads and see what was stalling the main render. Uninstalled some stuff and dropped my mouse polling rate.

-HDR does not enable properly in some titles like resident evil and outputs what I assume is SDR instead. All colors are fucked and washed out.

-random driver crashes if I have two monitors connected. For no reason.

-windows randomly enabling audio on my monitors over and over and setting it as default

Typing this out I realize the majority of the issues are Nvidia driver related. Fuck you Nvidia.

>>106487773
if you play games online discord is the defacto public looking for group location to meet people to play with.

>>106487983
this >>106487875

>>106487980
what nvidia gpu do you have? they reworked the dsc display tech on the blackwell cards.

>>106487254
You haven't figured out anything; people & businesses are driven by self-interest.
Any entity can invent their own programmable shaders and do whatever the fuck they want with them.
It's not NVIDIA's problem if AMD can't get enough developers to support ROCm, etc.

Stop wasting people's time with your low-effort rant in /pcbg/.

>>106487438
the usb 4.0 / usb 20g external nvme enclosures can get good speeds for file transfers. I've never tried mine for playing games.
option 2) get one of those x8 pcie bifurcation cards that let your run 2 more nvme drives, or get the x16 one?

finally, the low hanging fruit is to just get bigger ssds. 256 GB ssds are from 2013; 512 GB 2015; 1 TB 2019; 2TB 2021; We are on 4TB drives now.

>>106487993
3080. I tried HDMI to try to get around DP, but HDMI has its own issues.

The 5000 series has absolutely fucked drivers from what I've heard, so I'm not sure if it would even matter. I'd trade one problem for another most likely. Not would I pay for one.

I just rolled back to 366 to try to fix my problems.

As a quick aside, I'd honestly rather run mGPU to simulate BFI at lower fps than buy a 5000 series. The value is hilariously enough better. Plus I'd get like a gig and a half of VRAM back from moving the swap chain off my rendering card to my display card if I did it.

My eyes feel much better using BFI too, and it looks amazingly fluid.

>>106488052
>get one of those x8 pcie bifurcation cards that let your run 2 more nvme drives, or get the x16 one?
would those even work if I'm maxed out? my mobo is the tomohawk pro max. man i just scared of cloning drives and things get fucky. maybe it's real time to just do a complete refresh.

>>106486259
There might be some way to enter a debugging mode to do a factory reset on it or flash new firmware to it?

>>106488071
all i can tell you is that they enabled DSR over dsc with the blackwell cards.
If it is causing all these problems for you, maybe you could just buy a card on amazon and return it if it doesn't work.

>>106482994
most vindicated man of all time

ok /g/entoo men, how do I get faster speeds using topaz upscaling / interpolation?
My 5080 is chugging along taking forever to process the workloads i'm throwing at it.
Is a 5090 the only way, or is there some kind of specialized server card that would go way faster?

>>106488076
If you have no lanes left no. you can't drop your gpu to x8 or something? It barely matters on gen 4 plus afaik.

You could break out the x8 you get into two more x4. If your Mobo supports bifurcation ofc.

>>106488089
I cannot justify spending that amount of money on a graphics card, since the 5090 is the only one that really makes sense for me.

But ,I appreciate you telling me that. I'll keep it in mind if I get the next generation of Nvidia. I'm kind of leaning towards AMD just because of drivers now. (lmao)

>>106488076
>would those even work if I'm maxed out?
with that motherboard, no. If you have a board that can split the pcie lanes to x8/x8 on pcie 4.0 or 5.0 then one of those pcie add-on cards would work.

>>106488129
damn, maybe i'll just start nuking everything and starting from sq1. ugh im scared of deleting something that I actually need though.

>>106488129
>5090 is the only one that really makes sense for me.
5070 super with 18GB of ram should be a good upgrde over the 3080.

Is "Flower Leader" some china brand PSU that will make my PC explode?

>>106488076
>man i just scared of cloning drives and things get fucky. maybe it's real time to just do a complete refresh.
i've never had problems cloning the drive. If the cloned drive doesn't work then you can always pop in the old one.
what exactly is you drive layout? You can just copy the files from your second ssd over to a bigger one and put it in. You can get usb 10g nvme enclosures from china brands for like $15 now days I think.

>>106488162
are you talking about super flower? they are a Taiwanese brand that is legit.

>>106488168
https://www.newegg.com/super-flower-atx-3-0-compatible-1300-w-80-plus-gold-certified-power-supplies-black/p/1HU-024C-00058

>>106488106
memory overclock / bigger memory bus

>>106488177
>memory overclock / bigger memory bus
I'm terrified of this shit software crashing and losing my render.
It crashed earlier when I opened a video in mpc-hc and I lost like 8 hours of render time.

>>106488173
A+ unit

>>106488173
https://docs.google.com/spreadsheets/d/1akCHL7Vhzk_EhrpIGkz8zTEvYfLDcaSpZRB6Xt6JWkc/edit?pli=1&gid=1973454078#gid=1973454078

A+ rating on here. Super Flower makes their own psu and they supply other brands.

>>106488197
Thanks anon.

>>106488188
gddr7 has ecc. instead of crashing you just get worse performance. you likely just got a driver error.

>>106488039
why don't you stop sucking njudeas dick cockhead and reevaluate your entire life and stop being a useless waste of space to society

last question 1200W Plat for 10$ more or 1300W Gold?

>>106488152
Had no idea there was gonna be supers. I looked and 192 bit memory bus makes me wonder.
Not sure how much that matters on newer architectures, but I don't want bad frame time issues from stalling. I play at 4k.

>>106488222
>Privatize the risks & losses, socialize the profits
Nah. This isn't about NVIDIA. There's a reason why major tech companies are concentrated in the US.

>>106481978 (OP)

NEW TIER LISTS UP!

remember to add his tier lists, he offers a great breakdown of why each and every one is good or bad in clear way, that spreadsheets fail to convey

https://www.youtube.com/watch?v=4yUDKbF9-ZY

https://www.youtube.com/watch?v=hrydVwdhMD4

>>106488246
well the 5080 is a good upgrade over the 3080 if the price comes down on it especially when the supers release.

>>106488297
That's a good point too since the used market might be even cheaper.

Or not, based on the prices I've seen people asking for 3090s and 4090s lol.

>>106488039
Well if you do know how these work better than i do, my main questions would be:
>Can anyone legally modify, copy and/or produce/remove CUDA cores besides Nvidia?
>Can anyone legally do the same on ROCm, especially with firmware?
>with what right do these companies enforce a legal monopoly on thought? The protection from fraud is already a law in most countries, Their incentives are perverse ones meant to bottleneck the sector in the future purely by legal fiat and they, at times, actively keep new things unavailable since they themselves don't have to innovate

And for your information, my low effort rants are exquisite and qualitative. How very rude desu

>>106488314
90 class cards always retain unusually more value than the othera down the stack

>>106488314
I just sell my used stuff on leddit. you only have to do 3% gratuity to paypal + shipping.
I sold my 4080 on there for $1100 in april (i think) at the peak of gpu pricing.

Gonna ask here. What keyboard do you guys use?
Mine just died and I'm looking for something like this :
- Linear Switches
- Wired (I don't mind Wireless, it's always on the desktop anyway)
- Numpad (Boomer, I know, I know...)
- anti-ghosting
- Quick delivery (amazon would be nice)
- Brand : Logitech? Razer? Steelseries? Other?

I heard that Logitech 915 X durability is dogshit. Razer huntsman V3 Pro is good and Steelseries Apex 3 Pro might be the best.

>>106488348
RedDragon

>>106488348
get something where you can swap out all the keys. Buying a keyboard with locked in keys in this day and age is special ed.

If you want a numpad you are going to want some kind of wagie keyboard, not a gaymin braned one.

>>106488348
Don't get that normalfag crap. Just get a keychron, it's on sale right now. https://a.co/d/3sKGrx5

>>106488359
>>106488360
>>106488362
Thanks for the feedback, gonna look at it.
Keyboard being above $150 nowadays is insane.

>>106488278
They would have private profits, just not a legal monopoly over the product. A firm would have actually compete after their invention with their frst mover advantage and expertise as ideas aren't scarce.
R&Ding becuase you get a monopoly is perverse and makes Disney like companies. R&Ding because you get a better product to sell or use is the actual growth of the industry.
Synthetic economic stimuli are harmful policies desu

>>106488375
Try looking through here. .
https://mechanicalkeyboards.com/

I'm looking at my order from this place, the zeal pc blue zilent v2 switches were $13.00 for 10, and one of them didn't even work.

I think there is another keyboard store i've ordered switches from in the past but I cant find them on my email archives.

>>106488329
NVIDIA's not selling a public utility.
If people don't agree with their license agreements, they can take their business to AMD or someone else.
NVIDIA's major customers happen to be some of the wealthiest businesses in human history, and they're willing to pay the premium to train the best AI model asap.
NVIDIA spent the past 20 years or so investing in CUDA & AI training infrastructure, so they get to reap the benefits while the AI boom continues.

Every major tech company is planning to wean off from NVIDIA & CUDA in the long run, NVIDIA knows they're on a timer.
AMD, they're not as ambitious. They are fine with playing the follower. Lisa Su admitted it's fine to be second best in this segment because the market will be extremely lucrative for the #2 player.

The X86 license has been held by a duopoly since 1978, companies like Apple and other mobile device manufacturers eventually developed profitable alternatives.

>>106488391
NVIDIA doesn't have a monopoly over programmable shaders, everyone else just doesn't want to put as much effort into growing their ecosystem because they're in a race.

Hey guys. Long story short, idk how or why but my PC is still working after all these years. It's a RX 580 + i7 6700k setup I built literally a decade ago.
It's a very dated setup and I'm already noticing signs that something might break soon (PC freezing for no reason, hard stutters, dead CPU temp sensors, etc)

I have some money to spare now, but it's limited. I'll have to buy a new gpu now and get a mobo/ram/cpu combo later or vice versa.

For GPU, I think I'll get a 5070. For CPU, no idea.
Should I go for CPU etc first?
Now, I know that no matter what, I'll have a severe bottleneck. But it will probably be a couple months - 1 year period before I buy the rest of the hardware and I rather go for what will give me the most noticeable performance boost now.

I've been so disconnected from pc hardware in general that I feel a bit overwhelmed with all the current options I guess.

>>106488410
So NVIDIA's success is because they made the best AI products up until today with 20 years of work rather than IP and/or the best possible product has been banned by IP and Nvidia is wasting literal decades of time from anyone involved by having a legal monopoly just like intel&AMD with X86?
Their licence agreement, alongside IP law is illegitimate and unethical as ideas aren't scarce. They can actually produce a better product, which they likely would given that they have spent 20 years on this project but not ban their competition to what is mostly a redundant waste of time to avoid having to make said best product and instead do the bare minimum until the wastes of time have been completed

>>106488466
>>106488466
>>106488466

>>106488461
Most bang for your buck CPU at the moment is a 14600K, but the promotional discounts are running out of stock.
https://pcpartpicker.com/product/jXFmP6/intel-core-i5-14600k-35-ghz-14-core-processor-bx8071514600k

If you live near a Microcenter, buy one of their CPU/memory/motherboard bundles.

>>106488419
There shouldn't be an ecosystem for purely legal reasons, it's a hinderance for both the business and the tech

>>106488518
The system is driven by competition, innovation, and the potential for profit.
NVIDIA created a market that didn't exist for that profit potential, which is protected by law.
AMD's Ryzen & Epyc CPU IPs are also protected, and AMD gets to profit off a product no one else in the world can design.

>>106488605
Yes and this is wrong, they should build these things protected by their competitiveness, not by legal Fiat.
To claim that a market based on legal fiat is competitive is oxymoronic

>>106488655
Except these inventions happen to originate in markets that protect and reward the inventors.
You're out of your depth.

>>106488660
*More competitive is oxymoronic
>Hypotheticals don't exist and legal monopolies are good for innovation
How would you have felt if you haven't raytraced your graphics last time you gamed?
Inventors are rewarded by the company doing well, a company does well by competing in inventing a new product that is better than its competitors.
Not by threatening legal action against any possible competitors so that they don't actually have to compete to "do well", this incentivizes rent seeking behavior

>>106488692
This retard keeps posting images of copyrighted characters, and he's arguing against IP laws.
Those characters wouldn't have developed to be as memorable as they are if their creators weren't financially secured by copyright laws.

NVIDIA isn't rent-seeking or resting on its laurels atm, they're releasing new data center chips at an annual cadence.
They are producing GPUs & services no one else can match.

Grow the fuck up.

>>106488719
>IP made cartoon characters what they are, which is why the biggest cartoon firm, Disney famously used old fairy tales which should have been IP protected from them
>Touhou project, the famous legal firm with artists tacked on that destroys it's derivatives, the paragon of IP as a concept, no one should think differently from Zun about Touhou, this will surely aid creativity and not push artists to pirate the copyrighted material as much as they can get away with because it's an illegitimate law
To mock your second point;
>Nvidia totally aren't abusing their monopoly as much as they can, just look they're doing the bare minimum
Anon your first point is an insult to art while your second point is delusional, legal monopolies are bad for an economy, competition and creativity. You advocate for an industry of wasteful redundancy and rent seeking

>>106488605
>>106488660
they didn't invent shit. they simply provided a cost benefit to HPC customers which has always existed and will exist even if cuda wasnt created

>>106488782
Disney's interpretation & retelling of those common fairy tales are protected by IP.
NVIDIA's interpretation & implementation of programmable shaders are protected by IP.

>legal monopolies
It's not a legal monopoly; the nature of AI training incentivizes wealthy businesses to buy as many highly performant GPUs.
It's worth paying NVIDIA's premium while alternatives develop over time.
NVIDIA happens to be the company that sells the best products & services in this segment because they created this market while everyone else neglected it.

>>106488789
You're underselling CUDA and compute-only GPU programming. NVIDIA filled a gap that everyone else had neglected.

>>106488810
If someone doen the line gets inspired by the disney renditions of the fairy tales, they can be sued, unlike Disney which had no such problem because the old fairy tales didn't have IP despite the work put into them by countless story tellers, should the old fairy tales not be protected by IP then?

That's the
>Not a legal monopoly part
The
>You can't make CUDA
Is the legal monopoly part
Companies might as well pay premium for the jensen touch, but Nvidia shouldn't be able to ban anyone else from making CUDA and forcing them to make redundant analouges is

>>106488849
>should the old fairy tales not be protected by IP then?
Those fairy tales originated at a time when IP laws weren't protecting their original creators' rights.
Apples to oranges comparison.

>Nvidia shouldn't be able to ban anyone else from making CUDA and forcing them to make redundant analouges is
As the creator of the product, NVIDIA's entitled to set whatever terms they're protected under law.

>>106488872
Why aren't they retroactively applied?
Should they be retroactively applied?
Should they be applied at all?
IP law was a mistake and disney is both proof of that it was and the results of said mistake
>They invented it so everybody has to think as the way they want about their invention
No, the law is illegitimate in this regard
Nvidia is entitled to do whatever it wants with their own instance of their product.
I should be able to copy, modify and produce as i please as long as i learn of it. Otherwise is a legal monopoly for Nvidia, on my and everyone elses ideas about their products

>>106488926
>Why aren't they retroactively applied?
>Should they be retroactively applied?
>Should they be applied at all?
Just arguing like the retard you are, those folktales don't even have an original copy to trace the owner/lineage.
Regarding CUDA, you're acting as if Intel should be allowed to just reverse engineer and copy AMD's Ryzen & X3D chip designs.
That's not how any of this works.

>>106488959
>We can't find someone to give the monopoly charter to
You can make another fiat to say that someone, say the public owns it and goverment gets to decide who can and how they can use it, you then increase this everytime something is asked until you reach tech like fire and it's literally unenforceable due to the laws stupidity being on full display at that point
>Intel should make X3D chips
Not necessarily just intel, anyone who can and would
The way it doesn't work is illegitimate and should be abolished. It's stifling creativity and making monopolies out of legitimate sectors

>>106488999
The patents don't last forever, that's why NVIDIA's constantly adding new features to CUDA.
The same goes for the X86 licenses.

>>106489037
They shouldn't exist in the first place desu

>>106489101
With no patent protections, companies won't attract investors and the funds needed to bring new products & services to market.
Patent holders are also required to disclose the technical details of the invention.

>>106489130
No, investing will likely just return to pre IP norms like investing in a more competitive product or competing in a new market. You will just get rid of the ban all competition button

>>106489151
You keep missing the point that NVIDIA doesn't stop AMD from developing ROCm. They didn't stop Google & Intel from starting UXL.
All those guys gave NVIDIA a 20-year head start. Blame those guys.

https://x.com/BigJohnnyArcher/status/1963606484723306818
I FUKKEN NEEEEEEEEEEEEEEED IT /g/