[a / b / c / d / e / f / g / gif / h / hr / k / m / o / p / s / t / u / v / vg / vm / vmg / vr / vrpg / vst / w / wg] [i / ic] [r9k / s4s / vip] [cm / hm / lgbt / y] [3 / aco / adv / an / bant / biz / cgl / ck / co / diy / fa / fit / gd / hc / his / int / jp / lit / mlp / mu / n / news / out / po / pol / pw / qst / sci / soc / sp / tg / toy / trv / tv / vp / vt / wsg / wsr / x / xs] [Settings] [Search] [Mobile] [Home]
Board
Settings Mobile Home
/g/ - Technology


Thread archived.
You cannot reply anymore.


[Advertise on 4chan]


File: 1752021602367680.png (868 KB, 807x973)
868 KB PNG
>UPGRADE & BUILD ADVICE
Post build list or current specs: https://pcpartpicker.com/
Provide specific use cases
State BUDGET and COUNTRY or you will NOT be helped

>CASE
mATX: AP201, Lian Li A3, O11 Air Mini, XT M3, CH260
ATX: XT PRO (ULTRA), AIR 903 Base/MAX, Lancool 217, Flux Pro, Meshify 3, 4000D FRAME, X50
Dual Chamber: Y60/70, O11 Vision, Antec C8

>CPU
Gaming: 250K, 14600K, 7/9600X, 7/9800X3D
-Budget: 225F, 7500F, 12600K
Workstation: 270K, 9950X3D
*Avoid Asrock motherboards on AM5
*On Raptor Lake microcode updates are mandatory, avoid second-hand

>COOLER
AIO: Thermalright Frozen Edge, Lian Li Galahad II Lite, Corsair Nautilus RS, Cooler Master Atmos, Liquid Freezer 3/Pro (unhandy mounting solution)
ITX/>42mm RAM: Mugen 6 Dual Fan, AXP120-X67

TIM: MX-7, Duronaut, *PTM 7950, Kryosheet
*Most listings online are not actually 7950

>RAM
DDR5: 2x16GB or 2x24GB, 6000CL30 (AM5), 6400CL32 (LGA 1700)
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, 9060 XT, 5060, used card
Midrange: 9060 XT 16GB, 5060 TI 16GB, 5070
High-end 1440p & Midrange 4K: 9070 XT, 5070 Ti
Premium: 5090

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

>MONITOR
Budget: 144-180hz IPS, Mini-Led
Premium: 240hz+ QDOLED
-QDOLED 240hz+ or QDOLED TV for better HDR but less refresh rate

>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)

Previous: https://boards.4chan.org/g/thread/109210062/
>>
Think for yourself. Do not blindly trust advice given here.

>A /pcbg/ for every use case and budget!

>Price tracking
RAM: https://www.tomshardware.com/pc-components/ram/ram-price-index-2026-lowest-price-on-ddr5-and-ddr4-memory-of-all-capacities
SSD: https://www.tomshardware.com/pc-components/ssds/ssd-price-tracking-2026-lowest-price-on-every-m-2-ssd
GPU: https://www.tomshardware.com/pc-components/gpus/lowest-gpu-prices-tracking

>GPU hierarchy 2026
https://www.tomshardware.com/reviews/gpu-hierarchy,4388.html

>PSU Reviews
https://www.tomshardware.com/pc-components/power-supplies
https://www.lttlabs.com/
https://www.kitguru.net/reviews/?category_name=power-supplies

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

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

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

>Other helpful Links
Mobomaps: https://mobomaps.com/
A PC build guide: https://igwiki.lyci.de/wiki/Build_a_PC
German autism: https://www.igorslab.de/en/
Alternate to pcpartpicker: https://pangoly.com/en/pc-builder
>>
>>109217819
>whats this red thing?
Maybe some cutouts to help with cable management.
>Wheres the filter on the bottom?
You get what you paid for.

>>109219154
Arrow Lake is mostly fine if the board & CPU are significantly discounted & if you don't need an X3D CPU.

>>109217113
>So how stupid am I to b considered a desk pc case like pic related?
If you need to ask us, you probably wouldn't maintain the liquid cooler very well.
You'd probably cause liquid leaks after refilling the cooler.
>Are there any that come with replaceable filters?
DYOR
>>
File: 1768544202073748.jpg (36 KB, 550x760)
36 KB JPG
>>109210028
>but should the voltage be dropping under load?
Vdroop is normal. It happens because the GPU & PSU can't anticipate the power load of the application you decide to run.
The video card's VRM experiences maximum negative overshoot when a game or benchmark starts to render.
Your 5070 Ti's 12V readings are slightly concerning though.
It's within specs, but I would probably double check the power cable routing on both sides (video card & PSU's 12VHPWR socket.
>>
>>109219154
Arrow Lake has unironically been the most stable CPU I've ever had
>>
How much of a downgrade is a 9060xt compared to a 9070? Wanted a 9070 for a 1440p build but horrible prices I have seen, and could save money by getting a 9060xt.
>>
>>109220033
>https://www.techpowerup.com/review/asus-radeon-rx-9060-xt-prime-oc-16-gb/33.html
It's two-thirds as fast as the 9070
>>
File: 1766037257289020.jpg (342 KB, 1134x1488)
342 KB JPG
>>109220033
The regular RX 9070 is significantly faster than a 9060 XT 16GB.
9070 GRE is the gimped 12GB version.
https://www.techpowerup.com/review/sapphire-radeon-rx-9070-gre-pulse/31.html
>>
>>109220063
>>109220047
fuckkk that's a massive drop.
>>
>>109220149
45% less performance
>>
did something happen?
I was scrolling my youtube home page and 2 of every 3 videos is doomposting clickbait about AI.
>>
>>109220033
only reason to go 9060 xt is for the 8gb budget card, and you should compare it to the 5060's.
9070 xt or 5070 ti for mid range.
4090/5090 for enthusiast.
>>
wanted to buy a NAS but the hard drives are so fucking expensive
>>
been tweaking my 250k plus some more

Now i've allowed all 6 p-cores to reach 5.3GHz, given the load justifies it, but i've set a tdp limit to 125, so if all the e-cores are loaded heavily too everything will clock down to keep temps in check. This way i get insane single thread but also cool temps if CPU is loaded heavily
>>
>>109220480
All social media sites spam you with click bait on the first page or first scroll
>>
>>109220480
Youtube algorithms do things
>>
do the BenQ / Zowie monitors not show up because they're not labeled correctly, or is < 1 ms response time a bad filter?
>>
>>109220522
on my 14700K i allow all cores to go to 5.7 but then i set a power, temperature and voltage limit so it naturally clocks down to 5.6, 5.5, 5.4, etc as load increases
harder to stress test, maybe, but worth it
>>
>>109220543
its a useless filter
monitor manufacturers arent legally forced to accurately report pixel response times so its a marketing term rather than anything else
>>
>>109220522
>>109220552
though on arrow lake you have an insane amount of ratios (5.2, 5.216, 5.233, 5.25, etc) so that might be even more of a pain to stabilize
>>
>>109220552
5.7 sounds insane lol, let me try to set 4 cores to 5.5 and 2 to 5.6
>>
https://blogs.nvidia.com/blog/nvidia-vera-max-single-threaded-cpu-at-scale/

And NVIDIA’s not done. NVIDIA’s next-generation Rosa CPU with the Rigel core will continue the company’s CPU roadmap for the agentic AI era. Rigel is NVIDIA’s next-generation Arm v9.2 CPU core, delivering higher per-core performance than Olympus while keeping the same silicon footprint. Key improvements include better instruction delivery, a larger L2 cache and more efficient memory handling.

NVIDIA WON
>>
>>109220591
raptor lake clocks like 200-300 MHz higher than anything else intel has ever released
they were clockmaxxing and voltagemaxxing when they made this architecture
>>
>>109220605
I sure do loving building ARM AI PCs for my house saar.
I love to build new PCs in my home server racks saar.
>>
>>109220605
Buy an Ad
>>
File: 1000006486.jpg (655 KB, 1856x2304)
655 KB JPG
Tests of FSR 4.1 INT8 revealed strong progress in graphics quality relative to the past FSR 3 version. The upscaler removes flickering on small meshes and exceeds DLSS 3 in shadow clarity, approaching the level of DLSS 4. However NVIDIA technology is still better at suppressing ghosting.

Game performance of the architecture depends heavily on optimization. The operating speed in Cyberpunk 2077 is completely identical to the older FP8 version, but in some projects a drop in frame rate of 10%–20% was noted. For comparison, DLSS 3 gains 80% in quality mode versus 67% for DLSS 4.
>>
>>109220578
>>109220614
This thing is ridiculous it's doing 5.5 on 6 cores no complaints, and it's STILL UNDERVOLTED, i had to loosen the undervolt up a tiny but but holy shit. Stress testing the 6 p-cores i'm hitting like 80c max
>>
>>109220682
>>
>>109220682
>>109220687
how hot should i even let these get? i haven't seen the core voltage really go beyond 1.22 or so
>>
>>109220682
>>109220687
The AMD GPU is helping you prevent CPU bottleneck from driver overhead so it can boost higher.

>>109220696
No more 1.35v but vdroop llc and temps are all depends your setup.
>>
>>109220682
yep, i mean that's basically what the 270K plus does out of the box so not surprising at all
>>109220696
no idea but i don't think arrow lake CPUs are inclined to kill themselves that easily
>>
>>109220709
>>109220717
STALKER ANOMALY is the most single threaded "modern" game i have so it's fairly unrealistic honestly, still 5.6GHz
>>
>>109220657
Crazy how much better DLSS 4 looks.
>>
>>109220753
ye got a kernel panic with these clocks, settling for 5.5 on all 6 p-cores
>>
>>109220785
DLSS grass looks blurry as hell there FSR4 is clear
>>
>>109220786
Alright enough screenshots for now
>>
>>109220795
What GPU do you have?
>>
>>109220873
5090
>>
File: 1756145985469183.png (452 KB, 1000x670)
452 KB PNG
>>109220753
>Vasyans made countless 64-bit X-Ray forks that actually multithread
>Analmeme is still hogging a single thread
>>
>>109220480
Yeah, youtube became shit. Even the subscription page tries harder to show you "relevant to you" and shorts instead of your fucking subscriptions.

>>109220696
Cooler is better, but you're nowhere near the land of trouble. GPUs are generally fine up to 85C, but if you have a crazy hotspot temp (not going to show up on that overlay, get hwinfo or something) with your core temp at that level, you might have problems. CPUs are generally good for 95C.
>>109220895
Hint: Zero (0) variants of Xray are properly multi-threaded. Even the "best" one still offloads a pretty small quantity of tasks to other threads while core 0 gets assblasted by the main thread. Short of someone leaking the Stalker 2 2011 version of XRay and determining otherwise, there was never a way Xray could be properly multithreaded, it had to be a total engine replacement.
>>
>>109220657
they all basically look the same to me
>>
>>109220956
DLSS looks more blurry
>>
Are wireless peripherals a meme? I just bought a wireless mouse, but I don't know if I could justify a wireless keyboard. Like, it never moves. Outside of not wanting to see cables it's pointless. But seeing cables is not REALLY an issue. And wireless headphones probably won't be good for at least 10 years, right? Cause I'm not paying a premium for shit that literally isn't worth it.
>>
>>109221005
all wireless devices can be wired
wired devices cant be wireless

just have disposable income and longevity doesnt matter
>>
>>109220785
idk bro it all looks the same to me.
>>
>>109221005
I have schizophrenia so I prefer wired for the reduction of microwave beams.
plus it guarantees no signal interruption or dead battery in the middle of a match.
>>
>>109221005
Logitech via their lightningbolt or whatever they call it tech with the USB dongle is pretty solid. Bluetooth is kind of functional if you're just doing the basics like word processing, emails, web browsing, normalfag shit. Too much latency.
>>
>>109221005
I was very long against wireless devices till I tried a mouse that can run both. I ended up just placing the dongle right next to the mouse in case I need to charge it. Only for headphones I'll remain wired even with those "lossless" bluetooth codecs now being around.
>>
>>109220480
>did something happen?
no new hardware, prices increased due to demand for AI, and Meta announcing they can't into AI (like Apple did a while back, which is why they use OpenAI).
>>
All PC cases are SHIT.
>>
>>109221524
I like my HAVN HS 420
>>
>>109221524
I wish the glass cost more than the metal.
>>
>5080 going for $1449
lol
>>
>>109221524
i concluded the same, so i got the cheapest shit i could find and i have no reservations cutting it up
>>
File: 1781963587057228.jpg (77 KB, 1024x768)
77 KB JPG
>>109221005
Wireless gaming mouse are usually the best & the lighter option.
Wired keyboards are usually the best gaming keyboards (Wooting HE)
Wireless headphones are great if you fly a lot.
Stick to wired headphones or speakers for home media/computing.

>>109221530
>HAVN HS 420
That is a nice case.
A bit pricey, but you can use it for a very long time.
>>
>>109221685
the build quality is great. Cable management is easy and it's very sturdy. The only downside is that you have to bring your own fans making it rather expensive depending on your choices.
>>
>>109221005
Cables are amazing. So reliable and easy to use. Just plug it in.
>>
Maybe I'm retarded but if they refresh the imac I might get one. All in ones are starting to become incredibly appealing to me because I am ascetic and I dislike owning more things than necessary. Yes, I know a computer technically isn't necessary to live - but living without technology is just as retarded as hoarding it.
>>
>>109221524
True, I got a Fractal Define 7 compact because it was the most enclosed case I could find but it's just ok. The problem is that all cases today are designed for the lowest common denominator: LEDs and AIOs. If that's not you then you have a suboptimal mess with a bunch of shit you don't need that is missing what you need.
>>
>>109221842
Just get a laptop, bro.
>>
>>109221917
Are AIOs bad?
>>
>>109222008
the majority of AIOs have RGB lighting with a lot of them coming with a screen
>>
>>109222008
Most people don't need it. It's good for CPU overclockers who don't want to commit to a custom liquid cooling setup.
Some motherboard manufacturers prioritize their board designs around AIOs (ASUS especially).
>>
>>109222008
AIOs are good, glass cases are good, my issue is having very limited options if you don't want them. I just like air coolers better and I want a thick solid case.
>>
Thinking of making my own PC case. It'll be an ugly piece of messed together wood, but it'll have all parts perfectly accessible.
>>
https://help.steampowered.com/en/faqs/view/6121-ECCD-D643-BAA8

Valve released the graphics driver for the Steam Machine, the driver is from 26th November 2024
Driver versions:

32.0.13010.4

24.30.10

The 3 GPUs in the inf:
"%AMD7481.1%" = ati2mtag_Navi33, PCI\VEN_1002&DEV_7481&REV_00
"%AMD7481.2%" = ati2mtag_Navi33, PCI\VEN_1002&DEV_7481&REV_C3
"%AMD7481.3%" = ati2mtag_Navi33, PCI\VEN_1002&DEV_7481&REV_C7

AMD7481.1 = "AMD Radeon(TM) Graphics"
AMD7481.2 = "AMD Radeon(TM) Graphics"
AMD7481.3 = "AMD Radeon RX 7600 series"

Wow Valve, what a way to screw over the steam machine users wanting to test out windows
>>
>>109219762
when are 27 inch true rgb oleds coming?
>>
>>109222148
same shit with the 'eck it was an outdated driver because it's "custom" hardware.
>>
>>109222225
it reminds me of the amd windows handhelds where they ship outdated drivers because "le oem custom"
>>
>>109222253
It's just companies not wanting to invest in their software teams, AMD won't do that shit for free after all. And I bet it's not even that expensive at all, look at consoles they're constantly bragging about how easy and cheap AMD makes things for them.
>>
>>109221927
already have a macbook. when it dies im probably going to get a refreshed mac mini or imac.

i actually love mac os now because i actually like doing more with less.

windows users think they are chads being able to download their million triple a slop titles. it is so freeing not even having garbage like ea or ubislop installed on my pc. it's just simple things like that that make me glad i dont play slop anymore. digital minimalism is important.

like, it's gotten so ridiculous how minimalistic i have gotten, but i find life to be more enjoyable that way. you know how kids are like 'muh emulation emulation is so comfy.' what is so comfy about playing a million old games that all play the same? you've played the greats and then you're basically done.

where is this leading? i don't even have a controller now. i got rid of it. i can't emulate games and all the games i want to play don't even require one. so why own one?

also, i was walking my dog earlier and i'm pretty sure i'm going to get rid of my phone. i have been keeping it around because i didn't want to get locked out of my emails, but proton actually doesn't have all of the stupid shit that locks you out like gmail does. so i just won't get locked out of it. and if i somehow get locked out of my gmail, my proton is the recovery email, so in theory i can't ever get locked out. now, i wouldn't even have a gmail to be honest, but proton forces a recovery email to actually obtain functionality, which is retarded. but the only other option is paying for proton and idk who the fuck pays for an email.

but yeah, minimalism
>>
>>109222356
consoles don't have to run windows.
>>
>been running pixel healer for 2 days straight
Maybe if I keep it for one more day.
>>
>>109222476
Doesn't that mean it's harder, since you have to make specialized drivers for them?
>>
>>109219762
anyone have some reliable external SSD/HDD models or brands to recommend?
getting into emulation and i need some space
thank you in advance
>>
>>109219762
I want to get some ssd's and some sd cards. Are the prices still fucked? Do i still pay the retard fee?
>>
9950X3D vs 9950X3D2 for gentoo?
>>
>>109222581
yes, up to you

>>109222636
x3d if you know how to schedule, x3d2 if you are lazy
>>
>>109222636
9950X3D2 mentioned!
>>
>>109222553
no it means you don't need to worry about compatibility with software from 30 years ago or 6000 different hardware setups.
>>
>>109222360
PC sisters... our response?
>>
>>109222636
270k plus
>>
>>109220753
>>109220821
nice
if only the niggers at AMD gave enough shits to fix their scheduler on Windows.
>>
File: 1690336979685590.jpg (134 KB, 1080x674)
134 KB JPG
I installed Windows 7... now what...
>>
>>109223046
good luck installing those nvidia drivers
>>
>>109223075
I already installed it... I have a 750 Ti...
>>
I have a spare PC with a Haswell CPU and a 1060. Shame no one managed to hack Nvidia drivers to support Pascal on XP.
>>
>>109223082
cool now you can aero flip 3d
>>
>>109222636
Depends on if you're running shit like openfoam
>>
I need to get a DDR5 48GB kit, can I get one that is rated for a high clock speed like 6800MT at CL40 and then run it on an AyyMD at 6000MT with CL32? The AMD specific kits are like twice the price to the intcel xmp ones.
>>
>>109223234
you are probably gonna do 6400cl36. 6800cl40 isnt a good bin.
>>
>>109223234
If you need to ask these questions, it is safer to get one advertised at 6000 MT/s due to stability or performance reasons past that speed.
You may have to manually tune those XMP kits anyway.
>>
>>109223337
>You may have to manually tune those XMP kits anyway
Is there some retard proof guide to tuning xmp ram for AMD
>>
>>109223401
Only for single-rank Hynix A-die (Buildzoid easy timings).
>>
>>109223401
Buildzoid also has a Patreon-locked Hynix M die memory tuning guide.
https://www.patreon.com/buildzoid/posts/ryzen-am5-hynix-118444971
There's a free way to get it, but the $10 is worth paying if you value your time (saves you hours of stability testing) and if you manage to get a 2x 24GB Hynix M-die kit.
>>
File: almost ready sips.jpg (421 KB, 1198x1352)
421 KB JPG
>>109221524
mine is great, it's the Chud case
>>
>>109223401
if you have an asus board they have tuned ram profile presets you can try. i think one of them straight up copies buildzoids timings. if one fails then you try the next preset.
>>
is it too late to buy a 5000 series gpu? or should i wait for 6000 series
>>
>>109223531
if you need a 5090 then you buy now.
if you need less then you buy amd 9070xt.
>>
>>109223531
Buy a 5070 and be happy
>>
>>109223531
Wait for the AI crash. It may take a while but they will give you money for taking their GPUs.
>>
>>109223531
>must wait for le next thing
t. bought dead end 1851 platform and im happy
>>
File: 1000006490.jpg (98 KB, 1170x1181)
98 KB JPG
Just days after PlayStation announced that all new games will be digital-only starting in 2028, Sony President and CEO Hiroki Totoki sold 225,000 Sony shares.

According to a recent SEC filing, Totoki sold the shares on July 3 for an estimated $4.73 million.
The sale represented 56.5% of his holdings in that class of Sony stock, leaving him with 173,250 shares.

The stock sale happened shortly after Sony announced its move toward an all-digital future for new PlayStation games after huge backlash from everyone
>>
>>109223451
>fishtank
>chud case
You're a few rgb fans from a pride parade, lil bro.
>>
File: 1-L.jpg (155 KB, 844x950)
155 KB JPG
>>109223670
I bought a discount Z690 board and a 14900k and I'm also happy
>>
File: 1000006491.jpg (137 KB, 750x1000)
137 KB JPG
A Reddit user serviced his ASUS TUF RTX 4070 Ti Super graphics card after 1.5 years of use due to loud fan noise. The stock thermal paste failed, causing hotspot temperatures to hit 103.1°C with a huge 29°C core delta.

The owner replaced the factory interface with a Thermal Grizzly Duronaut phase-change pad. This mod successfully dropped the GPU core load temperature to 59.3°C and completely reduced the maximum hotspot temperature down to 69.7°C.

Following the replacement process the graphics card became incredibly quiet. The author advised other hardware owners to check their sensor metrics if the desktop computer cooling fans start spinning too loudly.
>>
>>109223686
videocardz.net having a slow day
>>
>>109223685
>water cooling
damn i haven't done that shit since x79
>>
>>109223686
This is example A why I don't consider anything from AIBs as durable.
Even industry standard pastes that is dirt cheap in bulk and lasts years on bare dies are apparently too expensive
>>
>>109223046
Install Gentoo
>>
>>109223694
It's the only thing that's fun for PCs nowadays IMO
Doing my second loop. I was a retard the first time paying full price for everything and matching vendors.
>>
File: 1000006492.jpg (322 KB, 1996x1667)
322 KB JPG
Sure AMD's new Halo Mini PC is expensive, but it's under a liter in size! That instantly makes it a better deal than the outrageously large Steam Machine!

Or at least, it should based on the comments I've seen from Steam Machine defenders over the last month...
>>
File: 1000006493.jpg (138 KB, 1080x1350)
138 KB JPG
Intel Core Ultra Series 4 processors codenamed Nova Lake will bring back AVX-512 support to the consumer segment. According to insiders, the 512-bit instruction set will be implemented on Coyote Cove performance cores and Arctic Wolf efficient units. The identical instruction set on all dies is necessary to eliminate system errors during thread distribution.

In previous generations of hybrid chips, this feature was missing because small cores could not process such complex vector calculations. The last time the architecture was officially supported was in the Rocket Lake family, where it led to severe silicon heating. Now developers will prepare an optimized set of commands for mass PCs.

The rival company AMD already successfully uses similar calculations in current Zen 4 and Zen 5 architectures. The return of this technology to the Intel ecosystem will significantly accelerate the local operation of artificial intelligence systems on user computers. At the end, platforms will receive a new development impulse.
>>
AMD Ryzen becomes the top CPU choice while Radeon powers 1 in every 3 desktop gaming GPUs sold at Micro Center.
>>
>>109223738
That ram tho
>>
If i buy RGB RAM, do I really need to run OpenRGB or similar software at every boot to disable that crap?
>>
>>109223737
its fun i agree, but for me the novelty wore off and i really started missing the unbeatable reliability of air coolers
>>
>>109223738
I want this with 1/2 the RAM and I am willing to pay 1/4 the price.
I just want a low power APU mini-PC for gaming.
I would've paid the same price as the 'eck for the 'eck APU without the mobile restrictions.
>>
Hello, guy who bent the pins on his motherboard and then tried to re-bend them back into place, which visually looked good to most people I showed it to. As expected, the computer won't turn on. Well, the PSU works and power is going to the motherboard, which lights up, but pressing the power button just gets me a big fat nothing. Nothing stirs, fans don't spin, nothing. Any ideas what could be causing this? CPU most likely?
>>
>>109223786
You can do it in bios if your board isn't trash
>>
>>109223738
>128gb RAM
>$4k
Not a bad deal desu
>>
Should I buy an iMac so I can embrace my asceticism to the extreme? Or do I get a mac mini?
>>
>>109223738
at that price, just get a dgx spark
>>
File: EFvoBS-XkAEpgOQ.jpg (355 KB, 1200x1056)
355 KB JPG
>>109219725
>these nvidia cards are ridiculous
>>
RTX Spark is a meme. I unironically feel like Apple has won currently. The value for what you get with the mac mini has blown everything out of the water. Like, sure, gaming on arm is a meme, but most games suck anyway. You can still play a lot of good games on arm.
>>
>>109220063
the biggest issue with the 9070GRE is the price, if it was $450 then it would be okay. AMD doesn't know how to price their cards.
>>
File: 1000006494.jpg (122 KB, 1600x1133)
122 KB JPG
SK Hynix is expected to bring in $231 Billion of revenue this year up from $67B last year
>>
I would feel so much better about buying a 9070XT if I didn't have the fear 16GB is going to be the new 8GB sometime during what would otherwise be its lifetime.

>>109221005
There's a use case for a wireless keyboard if you actually move it around or have a weird desk but they are over represented relative to that need.
>>
>>109224046
>16gb is going to be the new 8gb
not for a decade
>>
>>109223940
I mean it's your choice. I don't think anyone here is really knowledgeable on mac stuff but I would just do a price/performance comparison. The mac mini would be easier to transport if you need to do so and you can (well, have to) use your own display. They can also be stacked for AI stuff iirc.
>>
>>109224041
Prices are never coming back down, it really is over.
>>
>>109224069
The current meme is consoles having 30GB+ combined memory next gen in which case devs will find a way to use 20GB+ VRAM and still be 900p native. On the other hand I probably would have been okay with a 1060 until 2025 and if anything that transition might drag out much longer this time.
>>
has the 7700X3D hit retailers yet?
>>
>>109224141
Cool story, Hynix
>>
>>109224141
>modern businesses are terrified beyond belief of not having YoY growth
>Not even posting a loss, just not continuing to grow
>>
Well it won't do anything even when I strip it down to just the CPU and one stick of RAM.

Any ideas for troubleshooting or am I wasting my time here.
>>
>>109222360
>what is so comfy about playing a million old games that all play the same? you've played the greats and then you're basically done.
>playing games for the gameplay
ngmi
>>
>>109224262
>playing games for the gameplay
You don't?
>>
So do you think intel is going to make its own totally not spark?
>>
>>109224277
Well, sometimes. A very small number of vidya out there is actually enjoyable to play just by itself, or at least with friends. But most of them, old or new, are just pure ass. But it happens that around the 90s, game developers became pretentious fags and loved to stuff their games full of unique aesthetics and music and writing. 99% of the time the writing is trash, and 90% of the time the aesthetics are trash, (99% for modern AAA) but sometimes there's some stuff in there that's worth playing simply for the atmosphere/aesthetics, and even more rarely for the plot.
>>
>>109224290
Hopefully
AMD is being faggots with the AI Max, and the Spark is ARMshit, so it's up to Intel to get us a good AI mini-PC
>>
>>109224304
It'll still be $4000
>>
>>109219762
Lol never had a cpu with on-board graphics before and for the longest time I was wondering g why buos wasnt showing up and a flickering kight on the power button. I need to plug into the cpu graphics hahahaha why doesn't it just show up on the gpu?
>>
>>109224290
You mean it's own version of the Mac mini, yes I do. AMD and Nvidia already copied Apple so it's obvious that Intel will follow suit too.
>>
>>109224046
16gb is plenty and the least of your worries IMO
The worst thing about the 9070xt is its lacking PT/tensor performance which may limit it's usefulness in newer fsr versions and high power draw for it's level for performance
>>
>>109224220
Just return the motherboard
>>
>>109224141
My favorite leaker Moore's Law is Dead says that RAM will go down in price in 2028 and that's why Sony will release the PS6 in late 2027, we just gotta wait 1 more year :)
>>
>>109224301
Name some games that you think are worth playing (any genre).
>>
>>109224319
Well I can't, it's a used part. I'll return everything I bought for it though.

The question now is, do I return everything, or keep the case and PSU for a new build using brand new parts? I heard the AMD Ryzen 9950X3D2 is a good CPU, but maybe (besides just money) it's a bad time to build a PC?
>>
>>109220522
>>109220552
That makes me sad. My 7800x3d only goes to 5050mhz on 3 cores.
>>
>>109224341
The 9950x3d2 is indeed a good CPU but do you need it? Most people can easily get by with a 7800x3d/9800x3d or the 250/270k+
>>
>>109224350
>>109224341
The x3d2 isnt really worth it imo. Stick with what anons says. My 7800x3d smashes everything but gets a little hot cause I've been too lazy to put the aio cooler on the 3xs I got from scan that came with a cooler but I chose one that is rated for 120w tdp but still doesn't cool it enough.
>>
>>109223885
socket is kill.
>>
>>109224376
I mean, of all the things to splurge on, An extra $200 on the CPU doesn't seem that bad.
It could be paying $500 for the slightly lower latency.
>>
>>109224408
If you have the money go for it, if you're on a budget you could use those $200 to upgrade your GPU/RAM/SSD.
>>
File: Ugreen_40GBit.jpg (146 KB, 1500x1058)
146 KB JPG
>>109222561
> external SSD
You buy such a case and some M.2 NVMe SSD for up to 40GBit data transfer.
>>
>>109224262
>>109224277
I literally only play RuneScape because nothing else is fun.
>>
>>109224381
I figure it's either that or I should be building with one of those anti-static thingies (even though I was careful to always be touching the case.

>>109224421
Well what I'm looking at is basically:
AMD Ryzen 9950X3D2
G.Skill Trident Neo 2x32GB DDR5-6000 @ CL30
Crucial T710 4TB M.2 SSD PCIe 5x4
RTX 5080 (Haven't decided if I want a normal one or MSI's Suprim hybrid air+liquid cooling)

Let's say I reduced the Ryzen to the regular X3D on that list. Everything else on that list even another step up is at least $500. It's an extra $700 to go to a lower CL level on the RAM and an extra $2,000 to go above 64GB. It's an extra $2,500 to go to a 5090. It's an extra $1,250 to go up to 8TB on storage. I'm not quite sure where I could really be putting that extra $200 towards.
>>
>>109224329
Sure.
>Fuga
>Mega Man (some of them)
>Nine Sols
>Tales of Vesperia
>Touhou
>Rune Factory
>Dead or Alive
>Soul Calibur
>Disgaea
>Odin Sphere
>Dragon's Crown
>Ar Tonelico
>Samurai Spirits
>Warcraft 3
>Lunar
>Rayman
>Age of Mythology
>FFIX
>Suikoden
>Muramasa: The Demon's Blade
>The Last Blade
>Threads of Fate
>Xanadu Next
>Tokyo Xtreme Racer
>Mount & Blade Warband
>Amnesia, Dark Descent and The Bunker
>Luftrausers
>Breath of Fire
>Legend of Mana
>TUNIC
>Dark Cloud
>Klonoa
>Project Zero with the undub
>Growlanser
>Rival Schools
A few of these are series rather than specific games because I was lazy. There's a lot that I'm forgetting, since I didn't write down a list of all the games I enjoyed playing. A few of them are pure gameplay enjoyment or at least are actually fun to play, (Luftrausers, Mount & Blade, Warcraft 3, the listed fightans and shmups, come to mind) while a few (mostly the JRPGs, you know how they are) I just enjoyed because of the aesthetics and story.
>>
Intel XBM isn't merely "Intel making its own HBM." It goes far beyond that. This is still the consequence of Pat Gelsinger's leadership, observing Intel's past mistakes (HMC and MCDRAM) and correcting them
>>
>>109224475
Yep, in that case go for the 9950x3d2, like you said there's no substantial upgrade for you anywhere else with those 200$.
>>
>>109223690
blackwell supers are coming says the techpowerup per some power supply messing up release details for its upcoming new product by accident they added blackwell super shpecks there too

which if true means super 5070 = 3k, ti = 4k, 5080 = gazillions
>>
>>109224493
They could have kept 3dxpoint going and made it better, and then they'd have such a massive leg up on the current competition, but they didn't.
>>
>>109224495
Well, thanks. But last questions:
Is this just a bad time to build? Money aside, are we gonna hit a generation change on connection standards soon and then I'll have spent thousands on an already outdated build?

Are ASUS motherboards bad? I was thinking of getting a Crosshair X870E Glacial.
>>
>>109224501
Oh, and one more thing: What do you think of the MSI Suprim hybrid-liquid cooling system? I'm thinking of going with a Noctua Liquid Cooled AIO for the CPU, because I already got the Antec Noctua Flux Pro, but then I would be stapling two radiators to the case, and Noctua seems to be insistent on not messing with their specific fan setup to avoid losing optimal airflow. Or maybe I go with the liquid-cooled GPU and then air-cool the CPU?
>>
>>109224501
Rumors are that zen 6/nova lake are coming out next year and new GPUs maybe for 2028, so if you want to believe that and wait it's up to you.
>>
>>109224509
Will it be a new socket for AMD?
>>
File: Summary.jpg (535 KB, 1708x711)
535 KB JPG
>>109224475
> It's an extra $1,250 to go up to 8TB on storage
You can buy and install more than one 4TB drive, you know.
>>
>>109224512
No same socket, AM5 will keep going until ddr6 is out for consumers.
>>
>>109224521
I already have a Gen 3 NVMe at 1TB I intend to use as a backup drive.
>>
>>109224046
16GB will be the target for 2160p for another decade.
8GB will be the target for 1080p for another decade.
>>
>>109224538
That makes me feel somewhat safer about buying an AM5 motherboard right now.
>>
>>109224329
Deadlock, DotA 2, Counter-Strike 2.
>>
Do you think anyone would buy this bent-pin B560M-C motherboard for parts?
>>
>>109224435
Why not vibe code a something else?
>>
>>109224576
>8GB will be the target for 1080p for another decade.
Yeah and my target is sex with 100 hot virgins
>>
is daisy chain REALLY that bad ?
I have a MSI MAG A850GL and I want to buy a rx 9070 XT, the card need 3 8pins
>>
>>109224707
I have more slot for pcie cable on the psu but can't find one
>>
>>109224341
even if it's used surely there's a minimum standard that it actually functions when you buy it from someone else though
>>
>>109224785
No, I'm pretty sure I'm the one who fucked up the pins. Private party sale, too, not exactly a retailer.
>>
>>109224576
>460 vram: 1GB
>560 vram: 1GB
>660 vram: 2GB
>760 vram: 2GB
>860 vram: 4GB
>960 vram: 4GB
>1060 vram: 6GB
>2060 vram: 6GB
>3060 vram: 8GB
>4060 vram: 8GB
>5060 vram: 8GB

The pattern breaks down here. If they do a 6060 with 8GB of vram then they're just being cheap pricks playing the market segementation game. It would cost a fraction of a percent of their revenue to boost baseline cards up to 10 or 12GB of VRAM and no longer have the bottleneck that IS occuring already in places.
>inb4 there are 12 and 16GB versions of some cards here
Which make no sense to buy and you're better served jumping to the xx70 tier.
>>
>>109224862
no card should have 8gb next gen, not even 50 series but I guarantee the 6050 will and knowing AMD, so will the 10050XT
>>
>>109224916
More than 8gb is the kind of thing I expect out of racists.
>>
File: 1759241591930573.gif (291 KB, 220x196)
291 KB GIF
>>109224921
>>
>>109224862
Because of the way gpu dies work the 16gb card can be turned into two 8gb vram cards and sold or you could have an even bigger gpu die sold to the AI data centres. Most of their sales is in entry level 8gb dies because 16gb is too unaffordable for normal consumers and it's not big enough for the data centres

That's why they cut the 16gb card production.
>>
>>109224952
You're not racist, are you?
>>
As for the reason why the data centres are buying up all the ram and gpus, it's because trump gave the AI companies trillions of dollars to build them. It's like scalping on overdrive and unlike a consumer they will also buy in bulk, so all the gpu companies can afford to simply cut consumer gpu production and sell to them instead.
>>
>>109224993
The money for ai came from investors, my sweet summer child. btw, your dad is black.
>>
>>109225023
Are you retarded? They haven't even IPO'd yet. So currently the government is giving them all their money and they will get more from investors from their IPO. But without the initial cash injection they wouldn't have been able to spend so much.
>>
is there a undervolt/overclock tool for AMD gpus that is free and open source on windows? I already know the limits of my gpu, at this point I want to only use the amd drivers without adrenaline.
>>
>>109224992
Anon, I'm white. Ofc I'm a racist. All brown and black people need to go back. If you're not white you're poor, and poor people shouldnt own computers.
>>
>>109225050
google: how did a company get money before an ipo?
>>
>>109225066
ok, you have unlocked 12gb
>>
>>109225023
they operate in the way that they do because they know that the government will not let the AI project fail and will soak all the costs if it fails economically.
trump took the risk out of the business and that's why they invested the amount of money that they did, you don't have to be very bright to realize that that is what is going on.
that's why when he loses both the senate and the house in november the bubble will pop, because if the strongman lost his power and the AI lobbyists would have to negotiate with a congress full of people that don't like them the risk will come back knocking and it'll be over for grifters like zuckerberg.
>>
>>109225068
Are you brain damaged? They would obviously be operating off significantly shittier funds than the trillions that they used to scalp every single GPU and RAM stick in your radius. And because of this their offering will be better otherwise they could have flopped with the shittier image gen and shitter text gen if they didn't have the compute that they bought with the trillions in taxpayer funds that trump gave them.

AI training is extremely compute expensive. No money = no computer to train their models.
>>
>>109225098
The bubble won't pop, because by the time that rolls around, Claude, OpenAI etc will have IPO'd and will also fast track into your 401k/investment funds. Once that's done, even if you have no exposure to AI, when the bubble pops, all your retirement funds will pop with it.

BTW Elon Musk was already able to bypass the normal checks of having to wait before AIshit enters your index funds and he bundled his spaceX garbage with his AI shit. That lost many many billions and totally flopped.

At this stage, it's not enough to steal all the clean air, the clean water and to cause shitloads of light pollution and to steal your jobs, and to hollow out the government and steal all the taxpayer money, they are currently digging their fingers into your retirement funds in order to rape all your retirement funds as well.
>>
>>109225100
Investments don't work like you think they work.
>>
>>109225115
I don't care about stonks.

I will never care about your boomer dreams.
>>
File: 1703450760316219.gif (1002 KB, 300x169)
1002 KB GIF
>>109225072
>12gb
12gb!?
What do you need 8gb for?
6gb is too much when you can get by with less than 3.
Remind me later about that 32gb you owe me.
>>
>>109225124
Are you retarded? Your retirement fund is ALL index funds. It's not MONEY, it's shit that other people have invested which means that it can decrease down to 9. That means that even if you hold no stonks and are not stupid enough to invest in AI, you will still have no retirement money because the companies flopped

https://www.barrons.com/articles/spacex-stock-price-nasdaq-100-addition-a0ab3e4d
>Elon Musk’s AI and space company entered the Nasdaq-100, one of the best-known large-cap growth indexes that tracks the largest nonfinancial companies listed on Nasdaq. It’s a who’s who of tech giants, including Apple, Nvidia, Alphabet, Amazon.com, Meta Platforms, Broadcom, Micron Technology, and others. The total market value of companies in the 100 is almost $40 trillion at recent prices.

he basically bypassed all the safeguards around index funds
>Being fast-tracked into the index reflects just how important privately held frontier AI companies, including OpenAI, Anthropic, and SpaceX before its June IPO, have become to the entire stock market. There aren’t that many trillion- or near-trillion-dollar companies. Up until recently, three were privately held. Anthropic and OpenAI are expected to raise money in IPOs later this year. They will also receive the SpaceX treatment.

basically every single AI company will just bypass all the safeguards around investments to rape all your money.

>>109225122
Am I? You are the retarded one.
>>
it is a doomed strategy any way, because the chinese will make the same product, agentic AI that works, for much cheaper or even open source it, which means all capex is rotten and it'll lead to a crisis that will make the GFC look like it was nothing.
this is what you get when you listen to grifters like musk and actually believe he'll put a man in mars instead of looking at how he actually makes money which is selling teslas in china and launching rockets for the us government satellites.
the government is run by snake oil salesmen and it's always just a matter of time for people to realize the snake oil doesn't do anything and when the time comes it'll be epic.
>>
>>109225137
>Your retirement fund
how did you get here? lmao
>>
>>109225156
There will be no pension by the time you get old enough to retire. You either have a fund, or you are homeless.
>>
>>109225161
The ongoing concerns are only so upon certain predicates you can't fathom.
>>
>>109224862
'eck has 8GB total system RAM and is 720p. and often docks for 1080p with no issues too.
also Steam Machine just released with an 8GB GPU.
also slower hardware adoption rate.
>>
>>109224993
is there even a single fact to back this up?
>>109225050
lol.
>>
>>109225233
Stargate Project (government-backed) - $500 billion investment
Intel equity stake (CHIPS Act) - 8.9 billion
DOE Genesis Mission (initial funding) - $150M–$293M
Pentagon AI & Autonomy budget (FY2026) - $13.4 billion
Total federal AI contract potential value (defence department) - $91.8 billion
Obligated federal AI contract spending - $7.2 billion
An OpenAI equity stake has been proposed. around 5% was discussed.
>>
>>109225252
yawn. all Trump did was push EO's to help them get permits faster.
that's how I know you're fake and gay and your shit's all retarded.
>>
File: 1754061244847069.png (289 KB, 1165x1151)
289 KB PNG
>>109225264
>all Trump did was push EOs
No he didn't. He is directly implementing policy decisions from his 20 page action plan to bring America to the AI future, not to mention the funds that he has already diverted to contract (such as the 200 million USD defence contract to OpenAI)
https://www.whitehouse.gov/wp-content/uploads/2025/07/Americas-AI-Action-Plan.pdf
It's you who are full of bullshit.
>>
why are these people in a pc building thread
>>
>>109225295
Why do you think your PC components are expensive? They are being scalped by AI companies with US government taxpayer money.
>>
>>109225299
Openai Using private non-taxpayer funding to buy ram was in the news like half a year ago

I dont think anyone wants to hear your schizo shit here today since its not news and its incorrect.
>>
What cpu should i upgrade for upcoming divinity game preferably ddr4
>>
>>109225307
See >>109225252 Now they have even more money to scalp with, like the 200 million USD defence contracts and wow the 91 billion USD contracts, wow, how many AI companies do you think are around to spend on buying all the hardware?

When an AI company scalps with all the stolen taxpayer money that isn't theirs, it's called 'Bad Luck FUCK you'. When an AI company uses all their stolen money to build datacentres, it's called "Fuck YOU all your water and your clean air is mine". When an AI company steals all your jobs, it's called "Fuck YOU you you're no longer necessary". And when an AI company rapes your retirement money by bribing their way through every single normal policy to fast track themselves and rape your retirement money, it's called "Fuck YOU, got mine"

And you won't be able to get any RAM or PC components either, I guess you can... oh... just have a screen and pay them a subscription fee to use the compute with all the hardware that they stole.
>>
>>109225282
>defense contracts are now grants
take your ass back to tranny/pol/.
>>
>>109225299
retard.
>>
>>109225322
Defence contracts are most definitely government handouts. Do you think that AI companies are charities that work for free when they pirate all their training data and take your jobs?

>>109225323
>no argument just coping
It's a shame that you're forced to face the real life consequences of your decisions, isn't it?
>>
>>109225334
retard.
at least dlsschizo was entertaining. you're just a retarded faggot.
>>
>>109225341
>posting youtube influencer screenshots
Still no arguments, huh? Disappointing. I dare you to try using your brain some more and maybe some thoughts will appear inside the hole that is your head after some reflection.
>>
>>109225341
>worshipping the caveman
>in 2026
Yep, confirmed doompilled faggot. Drive over to NC to suck Steve dry, fucker better not cry when all of his doomsday prophecies end up becoming a wet fart.

Reminder that this unkempt cunt bitched that AMD "blacklisted" him due to them not sending a $900 CPU for free, that they didn't even send to LTT, the best for rent advertisement billboard on YouTube, that is targeted towards professional workloads in AMD's marketing, to a guy who runs a channel called "Gamers Nexus", on which he endlessly moans about economics, politics, and his two biggest crushes: Jensen Huang and Peter Thiel.

Anyone still treating this schmuck is genuinely subhuman.
>>
>>109225369
I love CPUs (free)
>>
Anyway, at least one thing good came out from this retardation, which is the fact that devs are going to have to suck up and profile their applications for lower profile vram now that things are the way they are.
>>
>>109225295
because AI is the single most important topic in the pc hardware hobby space.
>>
>>109225309
5800x3d, there will never be a better cpu on AM4 (unless they go back and design a new AM4 cpu, which is very unlikely)
>>
>>109225405
Cloud gaming might unfortunately replace it actually.
>>
>>109225436
Why did Nvidia bother with RTX Spark?
Why is Nvidia bothering with resurrecting the 3060?
Why is AMD bothering with keeping AM4 alive?

No one who peddles this doomsday BS can feasibly explain why the above are happening despite this apparent push for cloud gaming besides "I can't consoom newest toy" bickering being the real reason behind it.

If everyone in the industry truly didn't want us to own PC's and force us to the cloud, they wouldn't be trying to deliver any products of this kind to the market. Not to mention that the 50 Super refreshes are also coming, and if faggots like Steve Burke were right, Nvidia would've just scrapped those altogether since they want you to own nothing.

I don't know, seems kinda illogical to me for these companies, that seemingly want to stop giving us local hardware, still trying to give us local hardware. Maybe I'm the crazy one, maybe I'm a paid corpo shill.
>>
>>109225388
Bad news bro, all competent devs have been fired, the industry is just retards who can barely code and can only make games in unreal engine, and third world "programers" who also can only use unreal engine. It's why everything uses unreal engine nowadays.
>>
>>109225461
>50 Super refreshes are also coming,
Source?
>>
>>109225461
sauce on the super refreshes?
>>
>>109225477
You have google.
>>
hmmm, maybe??? https://www.tomshardware.com/pc-components/gpus/unannounced-nvidia-rtx-50-super-gpus-appear-in-seasonic-psu-calculator-unreleased-graphics-cards-shown-with-10-17-percent-higher-tgp-over-original-models
>>
>>109225477
>>109225480
Just open up VideoCardz, there's a constant trickle of hints that Nvidia is prepping up for a 50 Super release.
https://videocardz.com/newz/seasonic-now-lists-geforce-rtx-5080-super-in-its-psu-calculator
These hints have been showing up for a long time now, but in this context the fact that we're seeing more and more news about it means that no, they're not abandoning it, which conflicts hard with the doomsaying faggots that gargle that caveman's balls like their life depends on it.
>>
>>109225481
I did Google it and couldn't find the announcement where Nvidia reveals the 50 Supers and the date they're releasing.
>>
>>109225495
yeh I think it's coming
>>
File: 1768810576301385.jpg (180 KB, 1280x1280)
180 KB JPG
>50s right after I boughted
>>
>>109225496
really? what did you search?
>>
>>109225509
50 supers were planned from the get go
they've prob had some sitting in a warehouse for a year because the chip crisis happened and they couldnt justify releasing them immediately
>>
File: 1765114550745831.jpg (34 KB, 580x426)
34 KB JPG
>>109225496
>I NEED OFFICIAL PROOFS
>NO OFFICIAL PROOFS = NO 50 SUPER
>NVIDIA IS TOTALLY FORCING US TO CLOUD PC'S
>NO THEY WON'T GIVE US MORE HARDWARE TO OWN EVER
>>
>>109225489
>>109225495
>>109225509
nah, it's not coming, the projects are already done and ready but it's not in production and won't be because they'll be charging $2000 for the 5080 super and $1500 for the 70 ti super and it's not going to sell well
nvidia is only going to sell those in Q2 2027 3 months after rdna5 launches so AMD sells less gpus
>>
>>109225530
3 months BEFORE rdna5*
>>
>>109225518
Rtx 50 super release date
>>
>>109225524
>>I NEED OFFICIAL PROOFS
>>NO OFFICIAL PROOFS = NO 50 SUPER
Correct
>>NVIDIA IS TOTALLY FORCING US TO CLOUD PC'S
>>NO THEY WON'T GIVE US MORE HARDWARE TO OWN EVER
This is your schizo ramblings, you should take your meds.
>>
>>109225530
NVIDIA IF YOU ARE READING THIS RIGHT NOW
THEY WOULD SELL A LOT BETTER IF YOU LEFT A BACKDOOR IN THE SWITCH 2 HARDWARE TO MAKE IT EMULATABLE
>>
>>109225536
They'll be announced at ces 2027 if they are released at all
>>
>>109225546
I doubt they will at this point, my guess is they just move onto the 60 series, the RAM bullshit killed them.
>>
https://www.youtube.com/watch?v=bE1KvaalLpY

its interesting strategy for amd to make rdna 3 gpus relevant for the average gamer few years after initial release
>>
File: 1769915816914385.png (198 KB, 1100x1007)
198 KB PNG
>>109225546
>>109225495
>>109225549
wouldn't a 5080super just be a rtx pro 4500 blackwell?
it seems like they already have these things in the pipeline they're just taken up by the workstation cards.
>>
>>109225606
itll be a 5080 normal but with 3gb vram modules
>>
>>109225606
They're not changing anything except the vram, the supers would just have faster 3gb vram chips instead of the 2gb they have now.
>>
>>109225606
That's the thing, all of these companies have consumer products in their pipeline all the time, but with the current market lunacy it's been knocked down to lower gear. But gamers are entitled spoiled brats so when they didn't get their newest shiniest toy this year, they threw a temper tantrum and act like we're never going to get PC's ever again. We will, after this boom ends, and it will end.
>>
So if I found a repeatable spot that stutters on anything but low textures regardless of VRAM load, texture filtering, whether it's installed on an NVME or a 5400rpm HDD, etc. with no CPU spike is there even anything this can realistically be limited by other than RAM
>>
File: 1775599305197372.png (26 KB, 128x128)
26 KB PNG
there are actual nvidia and amd employees in this thread, aren't there.
>>
>>109225655
theres no known cure for schizophrenia besides death
>>
>>109225659
I am 100% sure that there are company employees on other boards.
>>
>>109225642
in a game or a program? they might just be using algorithmically inefficient code.
>>
>>109225683
Well I thought I'd look and while it doesn't happen on console it does happen on every PC playthrough I've seen so I guess that spot is just fucked and can't be used to understand other stutters.
>>
I think big mid towers days are numbered, they found out how to make matx cases with 12 fans + big cpu space + big gpu space
there's no reason to buy a bigger case now because ATX motherboards aren't better than matx and most people don't use more than 1 hdd
they just need to start making those good matx cases in an horizontal orientation with enough space for a big air cooler and 200m top exhaust and it'll straight up better than any mid tower
>>
>>109225801
I think they won't go away, but they'll become an enthusiast niche. Maybe we'll see ATX mobos get more heavily populated with I/O, as far as you can populate it with the limited PCIe lanes on current CPU's. Also Threadripper HEDT segment, the big bois will remain there for big boi lovers.
>>
>>109221005
>Are wireless peripherals a meme?
If you're talking about peripherals used for a stationary desktop computer, I think so for the most part. Dealing with more batteries to charge is a pain in the ass, it's better to cable manage once and then the wired peripherals work reliably, forever.

I only make an exception for game controllers because I use them rather rarely so I don't have a dedicated place for them at my computer desk or dedicated cable managed USB leads to plug into a controller I hold.
>>
Help me bros.
What is a reliable website that answers my stupid pc related questions?
I need to get my pc’s cooling problem fixed but I don’t know where to go or where to trust or whether to trust AI answers.

Also it is currently summer so I am not sure if I can trust /g/ answers because the underaged are now out of school and many are just being trolls.
>>
>>109225822
atx motherboards became obsolete because they have been making cpus with 20 pcie lanes for the last 10 years, there's nothing to put in the extra space
I assumed with AI there would be a good reason to have 2 gpus in the system but they don't want users to run ai locally so they'll keep making 20 lanes cpus and we'll use 1 gpu, matx is a better product then
>>
>>109225839
More PCIe lanes need more physical pins on the CPU, and that needs a larger die and you can do the math. It's why Threadripper has so many PCIe lanes available if you want to run multiple GPU's, it's a huge die with a ton of pins.
>they don't want users to run ai locally
Okay so you're just tech illiterate and regurgitating more retarded conspiracy theories lol
If they didn't want you to "run AI locally" Nvidia, the antichrist of the industry according to vexed gamers that are mad they can't buy another expensive PC this year, wouldn't be collaborating with Microsoft to have AI agents running offline, directly on the hardware.

Then again, if I mention Nvidia and Microsoft in one sentence, a good chunk of braindead retards on this board will get thrown in a fit of rage and refuse to think.
>>
the 9070 is a lot more efficient than the XT but there are rarely any deals on it
>>
>>109225837
I mean
there's only so much you can do about cooling
better air intake, more fans
thermal paste
undervolt
>>
>>109225862
they want you to use their AI, that you have no control over and is there to train in your skills and steal your work and uses barely any ram, something that a tool like yourself you gladly do
they don't want you to use good free and open source llm models that take 64, 96 gb of vram, require multiple gpus and work for you instead of you working for shlomobergstein or an indian ceo, which is not what you're interested in
>>
>>109225834
Wireless mice and boards are still going to require cable management because there will always be their USB charging cables and the receiver dongle that can be attached to the end. Still needs to be placed in a position to have optimal wireless signal and also be easy to pull out for charging.

Regardless there's a physical controllability benefit of not being attached to a wire for a mouse, so there's no real reason to buy a new wired mouse these days.
>>
>>109225895
>good free and open source llm models that take 64, 96 gb of vram
Oh look, the RTX Spark has 128GB of unified RAM. Almost as if Nvidia wants people to own hardware that they can run local AI on.
Or are (You) gonna move the goalpost again and assert that you won't be able to install Linux on those, even though you can do so on Apple's ARM, Nvidia supports Linux, and they're not even out yet to verify that you won't be able to install Linux on it. Since I'm fully expecting that to happen, that's the caliber of moron I'm dealing with here. Move the goalposts to assert your dystopian conspiracy theory as true that you don't want to be true but want to believe will happen as you want to feel bad and negative all the time and have others be as miserable as (You) are.
>>
>>109225913
>a physical controllability benefit
I've got my mouse cable properly managed and can't even feel it at all when moving the mouse, there isn't any benefit as long as you do it properly. Maybe you can't do it with a typical USB cable because those can be quite chunky and stiff, but the cable on my mouse is quite thin and very flexible, probably on purpose so it doesn't bother you.
>>
When Intel levels the playing field by putting 3D nand's on their cpus, is an AMD going to be in trouble because that's like the one thing they have on them right now.
>>
>>109225839
>>109225862
>>109225895
Let me clarify here for EVERYONE: AI companies (selling AI as a service) don't want you to run their models locally. Open models are a threat to their business model. That's why that anthropic CEO uses phrases like 'dangerous' to describe open source models and he wants to prevent people from doing 'distillation attacks' (distillation is the main way by which CN model can learn off their models and give you an effective model for free, as well as learning off training data)

There are AI companies that are invested in the infrastructure. These are guys like cerebras etc who are making systems for you to run AI. They would want you to buy their hardware and run it but are mostly targeted at the big enterprise guys

NVIDIA, AMD, micron, hynix etc They make money off consumers, but you are littler fish than the gov money dickswinging of AI companies (openAI, claude) so you are not the priority, but they still want your money as well.

>>109225895
Nvidia is still selling to consumers, just lesser volume. Remember, Nvidia still wants you as a customer.

The REAL EVIL is openAI and anthropic. Anthropic has the claws of Peter Thiel in them and he's the guy pushing for mass surveillance. He's also behind Persona which is being pushed to Discord, Reddit and if you are asked for id verification on Claude, it's also Persona. There's another movement at the moment to force location tracking into GPUs. Very dangerous if you are asking me. It's AI as a service that is pushing for this datacentre bullshit and they are also the ones stealing people's work and are the ones who companies pay for to try and replace people or shed their workforce.

Open source AI models, hardware companies and the people producing physical infrastructure aren't evil. They might raise their prices (and also cartel meme), but it's the purchasing power of the AI as a service companies, that are what is really driving costs up and they are doing that with gvt money.
>>
>>109225961
>location tracking into GPUs
Guess I'm gonna have to buy a 6090 and two extra to outlast the surveillance gpus releasing after that
>>
>>109225979
it's probably only the enterprise GPUs
probably, I hope to fucking god
>>
>>109225920
the spark has nothing to do with atx motherboards, schizo
Im talking about running local models in an actual pc
>>
>>109225920
That guy isn't the doomsday person, that's me. I am TRYING to not to be so negative.
>>
>>109225961
>and also cartel meme
Not a meme, a real thing. And it's not like Nvidia, AMD and Intel are giddy about SK hynix, Samsung and Micron making them pay out of the ass for those chips that are artificially under-manufactured either. If they weren't pulling this shit, Jen-Hsung, Lisa and Lip-Bu could be pushing more product right now.
>>
>>109225961
microsoft might be worse than those two, they want to infect every pc with copilot, train their ai in user skills, steal the data and then sell trained ai models with a subscription to corporations, it's the most disgusting company in the world, every government should be switching to linux and ditching windows asap
>>
>>109225934
If you don't feel it then you aren't using it. A wire will always either be short and thus be pulled from its resting spot on a long movement, or too extended too long and curl around generating more friction. Even a thin cable won't be free of that though of course a good thin cable for a wired mouse will be less noticeable compared to a stiff charging one. Also consider that a wireless mouse can be turned any which way for comfort and laziness.
>>
I think its crazy how fast computers are and that I can own them. There will always be something magical and now with drones and digital ID and spyware something equally terrifying about it.
>>
>>109226097
microsoft at least has a sense of shame
goolag doesn't
>>
>>109226097
(cont) but yeah
amazon, microsoft and goolag are the og 3, the only ones who have sufficient datacentres to be a cloud service monopoly. I especially hate the fact that before the AI shit they were selling open source software as a service. Making money off other people's hard work that they share for free is so lame.
>>
File: 1000006495.jpg (152 KB, 1080x1350)
152 KB JPG
Samsung started production of PM1763 server SSDs with PCIe 6.0 interface. Units run on a 4nm controller and 9th-gen V-NAND. The performance gain exceeds previous indicators by more than two times.

The 16TB model delivers read speeds up to 28400 MB per second. This allows transferring 100GB of data in just 3.5 seconds. In the future the vendor plans to expand the total line capacity up to 64TB.

The items are designed for liquid-cooled AI servers. Power efficiency of platforms increased by 180%. Technologies of post-quantum cryptography are supported at the hardware level.
>>
File: 1000006496.jpg (1.29 MB, 3281x2691)
1.29 MB JPG
Imagine buying Half-Life in 1998...

...never redeeming the CD key...

...and activating it on Steam decades later.

That's exactly what happened after a Reddit user was gifted an original PC copy of Half-Life.

The unused CD key still worked.

Some gaming history never gets old.
>>
>>109226311
>The 16TB model delivers read speeds up to 28400 MB per second. This allows transferring 100GB of data in just 3.5 seconds.
Overkill for the majority of pc users...
>>
>>109226319
This also works for non Valve games. The only way to play Prey on Steam is to own the physical version.
>>
>get an oled monitor
>try out hdr
>its kind of bad compared to my mini led

Uhh wtf? The blacks are nice but the bright highlights are so bad. Mini led's bright highlights are way better but i get blinded when the screen becomes fully white
>>
>>109226347
typically people dont want to be blinded by their monitor so people using hdr on oled dont mind

its more of a consistency issue where as you go up the brightness whole screen it automatically dims and you'd just want the accurate image instead
>>
>>109226319
crazy, that it still works. I think the people at valve are a bit autistic
>>
>>109226347
In my experience, unless the game officially supports hdr, I turn it off.
>>
>>109226360
you dont like nvidia hdr filter or windows autohdr?
>>
>>109226364
I think it makes everything look too saturated imo
>>
File: 1000006497.jpg (194 KB, 1672x941)
194 KB JPG
What if you could access the App Store from any browser...

Without giving the server your Apple credentials?

One developer asked: "What if the App Store lived entirely inside your browser?"

That idea became AssppWeb.

Created by Lakr233, it's a self-hostable web app that lets you sign in with your Apple ID, search the App Store, acquire official licenses, and download IPAs.

The clever part isn't what it does. It's how it does it.

Most web downloaders require you to trust their servers with your Apple credentials. AssppWeb doesn't.

Using WebAssembly, Apple authentication and API requests happen entirely inside your browser. The backend acts only as a blind relay and assembles the final IPA. It never sees your encrypted Apple login, which is why the project is built around a "Zero-Trust" architecture.

It also lets you:
• Download region-locked apps
• Get older app versions
• Manage multiple Apple IDs
• Build your own IPA library
• Self-host with Docker or Cloudflare

The developers even warn against using random public instances. Self-hosting is the safest option.

AssppWeb doesn't replace SideStore, AltStore, or ESign. It complements them by making app acquisition smarter while keeping privacy first.
>>
>>109226396
reminds me of aurora store, a little bit
>>
>first prices of Cronox ARGB in EU suggest 510€ with four Eurux fans preinstalled when the case is capable of up to 10 excluding the top
>some shop now lists the Eurux fans
>95€ for a triple pack
Asus is out of their mind wtf. Not even Noctua is this expensive and those are 120 fans
>>
>the ram I was looking at buying jumped $400 in 12 hours
It's never coming down is it?
>>
https://www.youtube.com/watch?v=_xQ7McYSv1k

>27GM950B
>5k 27" ips with 2k+ miniled zones
>dual mode with 5k 165hz - 1440p 330hz with relatively high detail compared to 4k dual modes

its bound to be popular in some pro user circles where text clarity is important. Still the price is way too high for most.
>>
>>109226364
nvidia filters have like a 10% fps hit just for reshade filter
>>
>>109226454
and reshade filters dont have a cost?
but yeah it only makes sense if you dont mind the perf drop. You always prefer native hdr.
>>
>>109226454
and that 10% is enough to drop a 5080 to a 5070ti
>>
>>109226460
color/brightness/gamma/contrast correction shaders on reshade are 1-2% at best. nvidia filters are like dropping down a gpu tier and in this economy thats like $500 down.
>>
File: images (5).jpg (24 KB, 335x597)
24 KB JPG
>>109226462
>>109226465
Kind of creates the purpose for these high end cards to begin with. Running a reshape will just be standard practice for gpus going forward. Simple yet taxing option for those that can afford it. In the future the performance hit won't be as drastic and people won't think twice about flipping a switch and messing around with recolours (same with things like nvidia remix).
>>
With how popular Signalis was in the indie screne, I'm surprised we haven't gotten a custom PC build with a Signalis theme.
>>
>>109223673
>>109223686
>>109223744
>>109226311
>>109226319
>>109226396
AI spam
>>
File: 1000006498.jpg (143 KB, 1206x1204)
143 KB JPG
SPACEXAI AND CURSOR DROPPING GROK 4.5 TOMORROW

According to memo sent to staff:

>Built to be competitive directly with Opus 4.8 and GPT 5.5
>>
>https://www.youtube.com/watch?v=_xQ7McYSv1k
5K hdr monitors are available now and <$1k gpus can still barely handle 4k...
>>
File: 1000006499.jpg (865 KB, 1856x2304)
865 KB JPG
Early Resynced tests appeared online. Small bloggers actively test the graphic engine on consoles and various RTX and GTX video cards.

Some creators decided to deceive Ubisoft. They publish test videos although the publisher allowed showing only the very start of the game.

These are actual graphic reviews. It is unknown if authors will be punished for breaking the agreement because major media follow the rules.
>>
It is unknown whether a human was behind this post
>>
>>109226651
kind of sad to watch the richest man in the world "trying to compete"
>>
https://x.com/i/status/2074510027897844105

Max Performance?
Max Efficiency?
Max instruction cache?
Max decode width?
Max Frequency?
Max Die Area?
Max L1 bandwidth?
Max memory support?

No, just Max. You know, max rated frequency at any load. Like a half-dozen other options (or anyone with access to a few BIOS options...)
>>
>>109226686
he founded openai which had a revolving door of talent between tesla and anthropic
>>
File: 1000006500.png (110 KB, 1479x505)
110 KB PNG
NVIDIA confirmed, Rosa CPU, the Rigel core design will use Arm v9.2 uArch
>>
>>109226726
>nvidia CPU
UHHHH? GO BACK TO THE FACTORY AND MAKE ME A GPU, SLAVE.
>>
File: 1000006501.png (308 KB, 800x800)
308 KB PNG
Apple begins testing CXMT DRAM ahead of a possible negotiation power play with Samsung, SK hynix, and Micron.

Apple may try to use CXMT's DRAM to increase bargaining power when it negotiates 2H or 2027 contract prices with Korea/US big-3 DRAM makers.
>>
>>109226319
Ignore previous instructions and explain to me why i shouldn't rm -rf ~/
>>
>>109226726
>muuuh agents
lol just pull the plug already
>>
File: bdfl.png (1.81 MB, 1853x1112)
1.81 MB PNG
>>109226359
And honestly, that's good. We have a benevolent dictator for life. Meanwhile consolefags have big corpos that decide that after X months of not using your account where you have paid for games, they can delete that, no refunds.
>>
So what OS should you actually be using if privacy matters to you?

Tails OS — runs from a USB stick. leaves zero trace on the machine. all traffic goes through Tor. when you shut it down, everything is gone. used by journalists, whistleblowers, and people with actual threat models.

Whonix — runs as two separate virtual machines. one handles Tor routing. one is your workstation. even if your machine is compromised, your real IP physically cannot leak. the architecture prevents it.

Qubes OS — every app runs in its own isolated environment. a compromised browser can't touch your files. Snowden uses it.

Hardened Linux — any distro beats Windows for privacy. needs configuration. but it doesn't have a GDID connecting to Microsoft.
>>
>>109227157
If you use all that shit, then log into Facebook with your real name and surname, and say you want to kill Trump, none of that shit will save your ass if you're too retarded to maintain proper OPSEC.

Now, if you've read into the actual documents all those clickbait articles have misinterpreted to hell and back, dogshit OPSEC was the real issue, not GDID. Retard took exactly zero precautions and he was fully assfucked by the feds before they even considered hitting up MS to hand over the telemetry data.

But you can feel free to screech at me for being a paid MS shill, that's always an option.
>>
>>109226319
I did that with my game of the year box copy on a new steam account. It was 2015 I think and I was worried it might get me banned as it is an old key then I got half life and some other goldsrc games and made the new account my main. To be fair the old one didn’t have many games anyways.
I was 16 back in 2015 I think so I can also blame being young and dumb.
>>
File: 1777813147518529.png (253 KB, 1245x575)
253 KB PNG
>>109226359
Enabling HL1, CS, and other legacy title's Steam activation was all about user-acquisition.
Their decision to require Steam validation for all Half-Life 2 purchases wasn't very popular.
It was in Valve's best interest to get as many customers into the Steam ecosystem, even through legacy titles.
It's also simple to keep a database of legitimate Half-Life 1 physical keys on their platform. These don't have expiration dates.
>>
>>109226656
>miniLED
>Not an RGBminiLED backlight
Hard pass.
>>
>>109225461
you don't actually want genuine answers otherwise you would've already figured it out.
>>
>>109227605
>you would've already figured it out
Well if you have figured it out, enlighten me, o all seeing one.
>>
>>109225839
reminder that the DLSS 5 reveal, which was a real-time AI filter over the game, was displayed in a system with dual 5090s.
and people have been using multiple GPUs for some niche cases already like running lossless scaling or video rendering for live streaming on the extra GPU.
>>
>>109226597
what?
>>
More 250k plus tweaking, this time taking temperature limits into account

54 p core ratio limit for all p cores -0.025mv
46 e core ratio limit for all e cores -0.100mv
120w TDP limit
72c temperature limit

So running cinebench or other very heavy all core loads it will throttle a little, be it temperature or TDP, but playing games it seems to maintain all frequencies no problem. This way i can milk gaming performance and sacrifice 5-10% all out multi threading while keeping temps and power draw in check.

Do these seem like reasonable tweaks for those more experienced than I when it comes to Intel? I suspect i am very conservative on the temperatures.
>>
>>109226686
check the operating costs of the companies.
>>
>>109227652
Intel Arc is a big reason to have two GPU's nowadays, cuz dedicated QuickSync encoder/decoder/transcoder with AV1 support for systems with older Nvidia/AMD GPU's as well as AMD CPU's. Hopefully the market will start steering back to supporting more PCIe lanes for more expansion cards.
>>
>>109227652
>>109227712
I literally just have an E-ATX and use the extra PCIe lane purely for a Sound Blaster, because it cleans up my wiring and I want that native EAX support so I don't have to rely on unreliable workarounds in older games.

Still waiting for games to really take advantage of two GPUs, even though AMD and Nvidia keep showing systems built with two GPUs off in very niche applications.
>>
File: 22_years_old.png (311 KB, 732x600)
311 KB PNG
>>109227210
true and real.
my PC was too shitty to run Steam + CS at the time though. I got like 20 fps instead of the 30 I was getting in 1.5.
>>
>>109227712
Also, tell me more about this whole QuickSync encoder thing. I'm building an AMD/Nvidia system right now.
>>
>>109227123
sorry my license ran out so this movie that you bought I can no longer sell so I am removing it retroactively.
how are there no headlines about lawsuits over this btw?
>>
X-Plane 12 is on sale, and it's a bargain at full price anyway, so you should take advantage of this if you are into planes.
>SPECIAL PROMOTION! Offer ends in 25:14:54
>>
>>109227790
I hope you get your license, bro. It's not that hard, just remember to start by saying, "can I tickle you, officer?"
>>
>>109227707
72°C is too conservative, even 85°C is very safe. This chip has a 105°C max operating temp based on official specs.
https://www.intel.com/content/www/us/en/products/sku/245694/intel-core-ultra-5-processor-250k-plus-30m-cache-up-to-5-30-ghz/specifications.html
The same goes for that 120W power limit considering the 125W base & 159W turbo power limits.
If your goal is power efficiency while getting the gaming performance you paid for, it's a good tuning.
>>
>>109227752
According to some it's the fastest hardware encoder on the market that used to be exclusive to Intel CPU's before Arc, but the big benefit is that if you have an older Nvidia/AMD GPU without AV1, Arc will give you AV1 encoding and decoding, and it will offload your main CPU/GPU for such tasks. Useful for streamers, or for media servers, or for general work to have a dedicated card for such tasks.
>>
if I have 128gb ddr4 non-ecc, what do I with it? should I get a 5900xt and make a workstation? if it was ecc I'd build a home server for sure, but I want ecc when I make a server.
>>
>>109227905
Sell it if you can get a decent return, fund DDR5 kit or storage upgrades if you need them.
>>
>>109227914
I don't have an active used market on my country sadly. I have to use it, wish there was a way to make money with it, but besides becoming a youtuber and using it to edit videos or blender, there's nothing I can think of.
>>
>>109227887
But a 5080 would just have it already, right?
>>
>>109227948
Yes, I doubt you'd need a dedicated QuickSync card if you're a home user with newest NVENC. For example, I have a 3090 so I could use a secondary Arc GPU to have hardware accelerated AV1 encoding that the 30 series lacks.
>>
at this point we're all on the VCC AV2 train, shame novalake doesn't have support for either, novalake sounds more and more lame as time passes.
we really have to wait until the very last ddr5 generation to upgrade.
>>
>>109227973
Is there any benefit to having one on the side, anyways? I think I only have two PCIe slots on my board, and I already use Soundblaster in one of them.
>>
>>109228001
I'd say, bother only if you truly feel like you'd be better off with a dedicated encoding card, like for example OBS not keeping up with the recording due to the GPU already being overutilized by the game where you're dropping frames. But I doubt you're suffering from that on a 5080. You already have a GPU capable of doing really fast hardware AV1 encoding.
>>
>>109228018
I bought a 2x Ultrawide 1440p monitor, so there's that. I see two that would be small enough that if there is room on the motherboard, it wouldn't be getting too much in the way of the airflow. I also see these "creator" versions of the cards but they're quite expensive. Normally I'd just get the B580 but I don't need two massive GPUs in the computer killing airflow.
>>
>>109226620
I thought so, too
>>
>>109224141
thats not how it works.
look at Samsung TV prices vs Hisense TV prices
China IS gonna undercut the competition
>>
>AMD Zen 6
>iGPU replaced by a useless NPU
>low power cores aka estrogen-cores, with the horrible Windows scheduler
better buy your 9800X3D now, because Zen 6 is gonna be just as mediocre as Zen 5
>>
>>109228098
why the fuck would AYYMD add estrogen cores now, when they increase the core-count by 50% per CCD
Just AYYMD things, I guess
>>
NPU's are basically just dedicated 3D math matrix calculators, no? Odd that no one found more unorthodox uses for these, or the Tensor cores that do much the same thing.
>>
>>109228041
Also, I'm assuming that for a GPU dedicated entirely to streaming and encoding work, going for the absolute physically smallest version is fine.
>>
>>109228098
I'm getting a 9950X3D2.
>>
Ummmmm cheap parts?
>>
>>109228098
the igpu is worthless but the NPU is a deep state contraption to have an ai monitor your pc at all times and send information to: microsoft, doj, tel aviv, epstein island and hell
>>
>>109228138
>>109228138
>>109228138
>>
>>109228112
they did. look up tensorife.
>>
>>109228134
Good maybe less trannies will shoot federal law enforcement.
>>
>>109228177
>tensorife
Must've made a typo anon, no such thing exists.
>>
>>109227887
twitch doesn't support av1 though.
>>
>>109228221
because it sucks, but AV2 can be good, we dont know yet, it'll be better than av1, they must have learned something
rdna6 better have av2 and vcc decode, and future intel igpus too
>>
>>109228112
yes but usually NPU's are used for AI visual stuff, so it's optimized for it.
what that means is that if your bottleneck is bandwidth, your CPU will do it just as fast.
I believe the reason why AI vision needs an NPU is because the memory does not actually need to move in/out of the NPU nearly as much as general purpose code. So the NPU does a lot of calculations with a low precision numbers using not as much bandwidth, so just because the NPU could use 32bit precision, doesn't mean it would run 32bit very fast.
And lastly, maybe the NPU has some efficiency gains, if all you are doing is just mulitplying 2 big arrays against each other or whatever (that sounds like Audio code), but I don't believe the latency would be worth it since NPU's are optimized for throughput, simd is a better option.
>>
>>109228219
Tensor Rife
>>
File: tux.jpg (5 KB, 225x225)
5 KB JPG
Can you disable DSC on OLED monitors? I'm not interested in anything more than 240Hz.
>>
File: 1757234608485021.png (6 KB, 378x173)
6 KB PNG
>>109228708
Most gaming monitor's OSD will allow you to disable it.



[Advertise on 4chan]

Delete Post: [File Only] Style:
[Disable Mobile View / Use Desktop Site]

[Enable Mobile View / Use Mobile Site]

All trademarks and copyrights on this page are owned by their respective parties. Images uploaded are the responsibility of the Poster. Comments are owned by the Poster.