MoreLuckThanSkill wrote on Sep 30, 2022, 15:53:
Hopefully they get their saves, but good luck with that. I have a strong feeling that Google just said 'fuck it' and shut down the Stadia shit and has probably already wiped and repurposed the hardware.

They're gonna shut down Jan 2023, in 3 months. Plenty of time to backup all savegames.

FTFSA:

"The country is also engaged in a controversial war GENOCIDE in Yemen"

Saboth wrote on Sep 26, 2022, 15:02:
Even with water cooling, you'd better count on your office feeling like a sauna.

theglaze wrote on Sep 26, 2022, 11:51:
Normal temperature @ 95°C, max thermal is 115°C? Looks like air cooling is 100% dead

Why? Because stock it wants runs at 95c?

This is not the metric you're looking for. What you want to look at is power consumption and on that front, they are much less power hungry than Alder Lake across the board.

Also, if you don't mind fiddling around settings in the bios, you can stop that stupid 95c behavior and even get decent temps.

Also hint: Overclocking results in less power consumption and much lower temps, go figure...

Some guy got it down in the 70s. Communication was utter shite this release, I didn't even know you could by taking it off stock until I stumbled upon that Asian guy who was reviewing it on Youtube.

Not saying it looks easy, but with that THICK of an IHS, I guess AMD looked for trouble... Der8auer had a video about this too ( not linked here ).

Be warned, it's only a 7600X. So I'm not saying you should try this at home with a 7950X xD.

See:
https://www.youtube.com/watch?v=H2NW7N-pPRE

In any case... Don't think I'm gonna touch AM5 ( even future ones ) as long I can't get those thermals under control with say a 140mm Noctua tower cooler.

"Mr Snowden's lawyer Anatoliy Kucherena was on Monday quoted by Russian state-run news agencies as saying that his client has never served in the Russian army, and therefore would not be called up as part of a partial mobilisation announced by President Putin last week."

LOL, imagine granting him citizenship then forcing him into conscription.

"Mr Snowden ( ... ) has made no public comments."

Not true anymore I guess?

May they have the peace they sought.

Acleacius wrote on Sep 25, 2022, 03:53:
War crimes have been committed in Ukraine conflict, top UN human rights inquiry reveals

As with all wars, by both sides if they're honest about it.

Ofc... I wouldn't expect them to be honest about it.

As usual the winners will have done nothing wrong as they're the ones who write the fiction, err, history.

RogueSix wrote on Sep 24, 2022, 16:54:
BTW, since this is a common misconception:

- The power connector on a RTX 4090 card has nothing to do with ATX 3.0. The 16-pin (12+4) so called 12VHPWR connector is PCIe 5.0 spec.

- The ATX 3.0 spec enables a PSU to sustain up to twice the power draw for a certain amount of milliseconds.

There are non-ATX 3.0 PSUs that have a PCIe 5.0 12VHPWR connector. Most/all ATX 3.0 PSUs will probably have the 12VHPWR connector since they are new enough to feature this PCIe 5.0 connector but it has nothing to do with the ATX 3.0 spec per se.

Thanks for that clarification.

Quboid wrote on Sep 24, 2022, 15:53:

Jim wrote on Sep 24, 2022, 13:48:
This has to do with powersupplies, a 3.0 spec from February. I find one on amazon, $270 for 850W, so may be a while before they are more commonly available

Thanks. I'm fed up with JayTwoCents' clickbaiting titles, I know that's how the YouTube game works but it gets tiresome.

Hard disagree, for once it was relevant, as the problem so far only affects Nvidia's 40 series.

They could've went for 3x ( or 4x ) 8 pins instead. They KNOWINGLY CHOSE ( Nvidia was aware of the problem in their early testing ) the under-engineered crap that melts, it's entirely on them.

And even having a ATX 3.0 PSU won't fix this problem, it also happens with them.

Doesn't matter if those who made the PCIe connector specs ( PCI-SIG? ) fracked up because the connector is too small, if you know about the issue, you can CHOOSE to not use it until they fix their shit.

*****
Edit: PCIe connector instead of ATX 3.0 specs, see RogueSix's clarification above.
*****
This comment was edited on Sep 24, 2022, 18:17.

phinn wrote on Sep 23, 2022, 14:53:
Besides the rediculous price tiers at announcement, RTX 4000 performance is still way ahead of RTX 3000 at the same price points, so what is with all this complaining?

No it's not, you're comparing launch MSRP to launch MSRP, that's not how it works.

You have to compare current prices ( used included ) to current prices, and by that metric, they suck.

And if for you "way ahead" is practically the same price to perf ( or slightly slower to equal with similar or higher price comparing the 4070 12GB to the 3090Ti for example ), then you have missed a few episodes mate. Get up to speed.

bstar wrote on Sep 23, 2022, 16:29:
Nvidia is notorious for releasing questionable benchmarks.

That's way beyond questionable, they went full Apple "bazillion X faster" on this one. They're now claiming upto 6x faster in a new release, I'm not kidding, look it up.

Gamers Nexus and Hardware Unboxed are gonna tear them a new one when they're gonna review them, guaranteed.

GinRummy wrote on Sep 23, 2022, 12:58:
Yet the spammers who fill my (and everyone else's) inbox and clog my phone line walk around in their 3rd world hellholes completely free.

You're not a person, err, corporation, so they don't care.

Cops protect the upper classes ( corps included ), not the plebs.

As a matter of fact, they protect the corps FROM the plebs.

But has about the same die size as GA106, iows, 3060. That ~3x price hike... LOL

*****

"Nvidia CEO proclaims 'Moore's law is dead' over RTX 40-series GPU pricing"

No spatula man, you just want an even higher profit margin than your already obese pig 60%.

*****

"Video Allegedly Shows Crypto Miners Jet Washing Nvidia RTX GPUs"

Idiots...

Acleacius wrote on Sep 22, 2022, 12:55:
Well deserved, hope this helps fund the Coop patch!

Fix bugs > Co-op

Beamer wrote on Sep 21, 2022, 23:15:
Even the 12GB stupidity, and it is stupidity from Nvidia, is really just a naming issue.

No it's not:

-That 12GB "4080" is a on a 104 die. A die usually reserved for xx70 GPUs.
-The performance is roughly below to somewhat equal a previous gen top-end, a performance slot traditionally reserved for xx70 GPUs.
-The core count VS a 102 die is also roughly in the ballpark of a xx70 class GPU.

They tried naming it a 4080, but no one was fooled. Even if they had named it the super duper RTX 40999ti MAXX, no one would've been fooled because we have all this relevant information we can refer to the past which tells the true story despite any naming shenanigans.

They tried slipping us a quick one, they failed. It's an 80%+, 400$+ price hike ( we all know none of the AIBs will have a 900$ card as announced, unless it's a blower ).

Same for the "4080" 16GB which on paper looks more like a 4070 Ti with the enormous gap in core counts VS the 102 die. And even if you consider it a true "xx80" card, just comparing the price also reveals the plot. It's a 71%, 500$ price hike.

No one was fooled. Well except perhaps only Linus who defended Nvidia's pricing ( LOL ), but that guy has also lost contact with reality long ago too. He spouted some insane babble about inflation, as if we had a 80% inflation in the last 2 years, *facepalm*. In reality it's around 14%, which is still a lot for 2 years but nowhere near 80%, that's pure nonsense.

The only model that "kind of" makes sense is the 4090, which is "only" 1600$ ( 6%, so less than inflation, 100$ price hike ), but with about 70% raw perf boost.

Ironically, the best "value" ( perf/price increase ) option is that 1600$ GPU. That's when ofc you don't take into account the ludicrous price of electricity and the power usage that peaks around 660W according to early info, despite its "450W" TDP.

Huang has completely lost touch with reality.

We're going through a recession, stagflation, a crypto crash, ETH going PoS and a slump in GPU demand after the covid and crypto high.

Nvidia meanwhile be like: 4070 @ 900$.

If Nvidia was a person, it would get interned in a psychiatric asylum for its own good.

El Pit wrote on Sep 21, 2022, 09:47:
But then you will have to pay $2,200 for your real RTX 5080 in 2 years!

Nvidia be like: "3.14 times faster means 3.14 times pricier."

Tomshardware: "Just buy it."

bstar wrote on Sep 20, 2022, 20:59:
This whole launch oozes of hubris, especially after the Evga debacle. I really hope AMD's RDNA3 announcement hits Nvidia hard at the 4080 range of cards.

AMD will do the same thing they did with RDNA VS Turing and RDNA2 VS Ampere.

They will join them at similar prices and perf and laugh all the way to the bank.

There's no real competition in a duopoly like this, there's just price gouging.

*****

Nvidia: "LOOK AT ME !!! I'm the scalper now."

AMD: "We both are!"

That's actually something that had me interested...

Not only does it upscale textures, but also models and adds ray-tracing effects as a game mod.

Developers and publishers are gonna HATE/LOVE this one.

Free good looking remasters ( hate ) and release their own good looking "remaster" with minimal effort ( love ).

The cease & desists are gonna start flying all over the place.

"4080" 12GB has less than half the cores of the 4090.

That's usually the xx70 segment.

So that's essentially a 4070 @ 900$.

xD

Things are just getting more and more insane.

And them artificially segmenting Optical Flow Acceleration to the 4000 series is another slap in the face.

Their SDK for this was released in 2019 and ran on as far back as Turing and Volta ( as long as you had tensor cores ).

Not to mention that I have Optical Flow Acceleration running right now on a 3070 in Smooth Video Project 4

LOL, FU right back Ngreedia, I ain't upgrading just yet and certainly not at those extortionate prices.

I still use cursive in long texts... Otherwise I write too damn slow.

Altho that's a rarity nowadays with computers, internet and co.

It's mostly typing.

Kxmode wrote on Sep 16, 2022, 22:04:
I like how you're telling me my experience of hearing a loud card that ran hot is pure imagination. Is there anything else I should know about my experience with NVIDIA's products?

Don't know, I don't live in fantasyland where a dual/triple axial cooler can have worse temps and noise than a blower when every single review ever made in the last 10 years without exceptions says the exact opposite. LOL

You made an extraordinary claim, back it up with extraordinary evidence or be laughed out of the room.

Your choice mate.

Note: I won't be waiting for that evidence, because you and I and everyone else here knows for an irrefutable fact it doesn't exist.

So have a nice day and cya in another thread! *shrugs*

RogueSix wrote on Sep 16, 2022, 21:08:
And, no, the EVGA reviews were not unanimously good.

That's a nice strawman you got there, reminder of what was actually said:

"That's very strange because practically every single review of GPUs has better thermals and noise with AIBs, including even most of EVGA's lowest tiers."

Implied is that it's VS Nvidia's FE models, NOT other AIBs ( there's no way you can construe that statement this way other than with a pure strawman ) and I could link you a few hundred reviews showing it's 100% true.

It's also true they didn't have the best AIB GPUs out there, but credit where credit is due, their service is unparalleled in NA.

TL&DR:

EVGA > FE as far as thermals and noise is concerned
EVGA = #1 service by far in NA
EVGA were rarely if ever #1 VS other AIBs when noise and thermals are concerned