NVIDIA 5 Series cards - Get your wallets ready boys and girls....

[Aapje]

53 minutes ago, kksnowbear said:

No reliable data means no valid price performance comparisons are possible...

What is reliable enough (or not) is opinion, but you are presenting your opinion as fact.

Again, I have explained why I consider the available benchmarks good enough to have high certainty that the performance of the new cards is not worse. This is a very modest claim, much less so than people who throw around actual percentages.

Is it 100% certain? No. But it is a core tenet of science that you cannot actually prove anything. So if you demand absolute proof to call something a fact, then facts do not actually exist. After all, there is a possibility that we are a computer simulation, and anything that we perceive to be a fact can be changed, no longer being a fact, or can be a delusion implanted in our brains.

But it is silly to have to present everything as an opinion, so in reality, people just state things with certainty when it passed their subjective threshold. You do so too. This entire idea that your personal threshold of what is valid should be accepted by others, is authoritarian and arrogant. You deny others the right to their own opinion on the matter and you act as if you are perfect.

Quote

The problem is when someone, like you, wants to post opinion and have it treated as fact.

Yet you do the exact same thing in the quote at the top of this comment. So again, you apply your double standards, where you get upset when I do the same thing that you do.

Edited January 23 by Aapje

[kksnowbear]

2 hours ago, Aapje said:

What is reliable enough (or not) is opinion, but you are presenting your opinion as fact.

Nope. The fact that you have to use the word "enough" means you're talking about an opinion. I did not use the word "enough".  Reliable data means exactly what it says; it is not subjective.  $100 is $100.  100mph = 100mph.  These are empirical fact, not subjective. 

2 hours ago, Aapje said:

Yet you do the exact same thing in the quote at the top of this comment. So again, you apply your double standards, where you get upset when I do the same thing that you do.

Nope. Again, you used the word "enough", which changes the entire context from objective to subjective. I didn't say "reliable enough".

Edited January 23 by kksnowbear

[Panzerlang]

Lol. Pretty much a debate over semantics. However, I like this line: " people just state things with certainty when it passed their subjective threshold"

Indeed. I believe it has passed my subjective threshold that NVidia are a bunch of lying crooks who are now asking obscene prices for a GPU not significantly more powerful than the previous generation. But huh, define "significantly". If 30% greater rasterization grunt (5090 vs 4090), that's a third more power. If 20%...

Then use-case. With a top-end VR headset and CPU the 5090 will be a total waste of money in DCS. The best CPUs already can't keep up with the 4090. However, a 5090 will definitely make a difference in IL2.

[kksnowbear]

3 hours ago, Aapje said:

This entire idea that your personal threshold of what is valid should be accepted by others, is authoritarian and arrogant. You deny others the right to their own opinion on the matter and you act as if you are perfect.

Again with the personal attacks.

And nope, I don't think any such thing.  I just don't call something "proof" unless it actually is proof (even though I can be mistaken, and make corrections as needed).

My "personal threshold of what is valid" is based on observable fact.  Empirical measurement.  Reliable data.  Not speculation and opinion, or bullsh*t marketing. 

I'm not sure why that offends you, other than because it's your opinion I'm not just going to accept without question.

And I already explained I have no problem with expression of opinion; I enjoy discussion involving different perspectives.

But if someone represents something as fact, they need to have data to support that.  Like the grade school math teacher tells us: Show your work.

Otherwise there's no way for the reader to actually trust anything written here;  basically *all* advice becomes unreliable.  People can just make up whatever they want, since nothing is ever subject to scrutiny.

That's why facts and reliable data are important.  They help separate bad ideas from good ideas, and what is BS opinion vs what is trustworthy advice.

Edited January 23 by kksnowbear

[Aapje]

9 hours ago, kksnowbear said:

Reliable data means exactly what it says; it is not subjective.  $100 is $100.  100mph = 100mph.  These are empirical fact, not subjective.

False. There is no objective standard for what one should consider to be reliable. Let's say that LTT tests these cards. Some people consider their tests reliable, others do not. The same for Hardware Unboxed. The same for Gamer's Nexus. Some people consider them all unreliable.

Fact is that none of these channels are without fault. They all make mistakes and their testing is imperfect. There can also be systemic issues across channels, like the vendor misinforming the channels, so they all make the same mistake. And cards can have issues that come out of left field, like the high CPU load of the new Intel cards, that led to testing changes, to expose the specific traits of those cards, which was not visible in the normal way the channels tests these cards. So at no point can you argue that their tests are perfect and their results cannot be wrong. The decision when to consider the evidence reliable enough to support a certain claim is thus subjective.

Quote

Again with the personal attacks.

No, it is not an attack, but an observation of your behavior. Not just in this thread, but across this forum, you try to take control of threads and demand that others accept your subjective opinions as the truth, rather than simply allowing them to have a different opinion from yours.

I understand that it may be unpleasant to get called out for that kind of behavior, but instead of retreating into your double standards, you can also reflect on your behavior and be a nicer forum member.

Quote

Otherwise there's no way for the reader to actually trust anything written here;  basically *all* advice becomes unreliable.  

IT IS! Again, science has realized a long time ago that the best we can do is to show how likely certain things are to be true, and even then only based on certain assumptions. Yet those assumptions can be wrong and even very likely claims can turn out to be wrong. So true reliability, in the sense that something is 100% certain to be right, does not exist in a scientific sense.

Most people do not have the ability to actually do this well. Even very many scientists fail at doing science properly. You certainly are not capable of it, as you've demonstrated. I don't see you using the tools of science to greatly increase the accuracy of your predictions, or understanding how they work.

But science is not the be all and end all, and there are good reasons to not be that rigorous in normal conversation, because it is a huge burden to be far more accurate. But the result is that normal conversation is a world of subjectivity, not of fact. You may have the delusion that you know what the truth is, but that delusion makes you less accurate, not more.

[kksnowbear]

2 hours ago, Aapje said:

There is no objective standard for what one should consider to be reliable. Let's say that LTT tests these cards. Some people consider their tests reliable, others do not. The same for Hardware Unboxed. The same for Gamer's Nexus. Some people consider them all unreliable.

What you're now describing is various people's opinions about whether something or someone is/not reliable.  Unfortunately that's not the same as the meaning of "reliable" itself.

Reliable means consistently good in quality or performance; able to be trusted.

There's no opinion in what "reliable" means.  Something is reliable or it's not.  It's consistently good in quality or it's not.

And again with the personal attacks! 

LOL How, exactly, do I prevent people from having a different opinion?  There are pages on this very thread to show you're wrong about this, too:  If it were true I'm keeping anyone from having an opinion of their own, then I'd prevent anyone from arguing with me here.

Yet that's not happening.  I'm simply expressing my own perspective, calling BS when I see it.  You're just ticked off because you had no data to back up your claim when I called you on it, simple.  All this going on about science and why numbers can't be empirical is just subterfuge to obscure the fact that you made an unqualified statement, when you had no data to support your assertion.

In this mysterious "scientific" world if yours, nothing can be quantified, making it very convenient for people like you, who want to say anything they want and have it accepted as fact.

In the real world (where I earned a living in a very technical industry), there are quantifiable values.  There are empirical proofs.  Electronics don't work on opinions, and computers don't work on guesses.  It's still the mathematical difference in a 1 and a 0.

Now you assert that all advice is unreliable.  I guess it's impossible, then, that I've been professionally compensated my entire adult life based on my perspectives about how computers work.  Fact is, many people here and on other gaming/sim forums have at times trusted my advice as "reliable enough" to spend their hard-earned money on.  It's really hard to imagine successful businesses and individuals with well-functioning machines would trust someone with a complete and utter lack of understanding about how all this works.

But since, in your world where nothing is empirical and reliable, $100 is apparently not  $100, and we can all pay Nvidia with fake bills, in exchange for their fake frames.

Now we're talkin'  

PS I'm an honorably discharged veteran who spent about a quarter of my adult life defending people's right to have and express their views, and I find any suggestion to the contrary genuinely offensive.  I'd really appreciate it if you laid off the personal attacks and stick to the actual subject.

Edited January 23 by kksnowbear

[kksnowbear]

Just read Nvidia has bumped the reviewers embargo on the 5080...(not the 5090 mind you, just the 5080). 

I bet the reviewers, who already expended resources working with 5080s and now will probably wind up starting over, are really pleased with Nvidia right now.

Hard to believe some people can't see all this nonsense for what it is.

Oh well.  Ya can lead a horse...  

 

Edited January 23 by kksnowbear

[EightyDuce]

 

 

 

25-50% raster uplift over 4090 depending on game; averages around 30% uplift at 4K. 

So TLDR its a 4090Ti

Edited January 23 by EightyDuce
More reviews added

[kksnowbear]

26 minutes ago, EightyDuce said:

25-50% raster uplift over 4090 depending on game. 

Pretty sure he states that kind of increase "at 4k".  I believe that's still not the majority of gamers by far. He goes on to acknowledge that at the more common lesser resolutions of 1080p and 1440, the advantage - even when it's not really CPU bound "really starts to come down".  (~ 00:50)

So I guess, according to Nvidia's reasoning, we all must also pay for 4k monitors to get the benefit of paying for a 5090

Would've been appropriate to show a combined chart of the games tested at 1080p and 1440...but I guess that's not how Steve at GN sees things lol...as it is, you have to plod through manually to see that the lower resolutions are...well, much less favorable, we'll say (I'm admittedly still plodding through).  I've already seen quite a few games with 6 or 8% improvement at 1080/no 'magic'.

Edited January 23 by kksnowbear

[EightyDuce]

Overall, not surprising considering its built on the same process as the 4090, with an uplift in core count, memory and power consumption.

Probably not going to bode well for other SKUs lower in the stack considering the 5090 is likely going to have the biggest uplift over previous gen. But, coming from a 30XX and 20XX generation depending in real-world pricing and availability should be a good time to upgrade. 

Could be a good time to snag a 4090 on a used market.

Some good news though, 9800X3D is still an absolute beast of a CPU... So there's that. 

Edited January 23 by EightyDuce

[SharpeXB]

6 minutes ago, kksnowbear said:

So I guess, according to Nvidia's reasoning, we all must also pay for 4k monitors to get the benefit of paying for a 5090

I have to imagine that’s what this card is meant for, it’s overkill for anything else. Also the customer for this likely owns a 4K screen already. And they aren’t extravagant anymore. 

15 minutes ago, EightyDuce said:

Could be a good time to snag a 4090 on a used market.

Or sell one for a free upgrade I see these selling for more than I paid for mine. 

[EightyDuce]

8 minutes ago, SharpeXB said:

 

Or sell one for a free upgrade I see these selling for more than I paid for mine. 

I intertained that idea but I don't wanna breakndown the Watercooling setup and get a new water-block. My loss their gain. 

[LucShep]

1 hour ago, EightyDuce said:

 

 

25-50% raster uplift over 4090 depending on game; averages around 30% uplift at 4K. 

So TLDR its a 4090Ti

Heh.... exactly as expected.
Now prepare for the even more disapointing reviews of the RTX5080 tomorrow....

Nah, no thanks.
Rather get a nice second hand motorcycle (or "whatever"!) than overspending on power-guzzling GPUs in a revolting price gouging market, which will not make miracles in VR for a game that has its own long running issues yet to be solved.

 

Edited January 23 by LucShep

[SharpeXB]

12 minutes ago, EightyDuce said:

I intertained that idea but I don't wanna breakndown the Watercooling setup and get a new water-block. My loss their gain. 

I don’t really see the need for one either. The math is just amusing. Yeah for the pain of replacing it a bigger boost would be preferable. Although I guess I gotta buy one and throw it back like a little fish to win the challenge above

Edited January 23 by SharpeXB

[kksnowbear]

Thank God for HUB...one chart to nicely summarize average performance for each resolution (why GN Steve can't get it is beyond me...then again, I consider the source).

And the results?

Improvement over 17 games, average FPS at 1080p:  204 vs 202 FPS - effectively zero and within margin of error.  No human will ever be able to see the difference in two frames when already over 200.

Even at 1440, the average improvement across 17 games was 12.2%.  192 frames vs 171; 21 frames when you're already getting 170+.  Again, no human can reliably tell the difference.

HUB Steve also discusses the MSRP vs "real" pricing value toward the end of the video.  Nice of him to include the perspective I've been discussing all along (and getting jumped on for it).

His words (21:29): "After a little over two years, we're seeing no improvement in cost per frame."  (And that's at 4k, so it only gets worse at lower resolutions).  He addresses the question of a 5090 actually being available at MSRP ("do we honestly believe the 5090 will be $2,000 US? Do we really believe that? Yeah it's probably going to be pretty damn ugly in terms of value, because at $2,000 US it's it's already pretty ugly...")

And that's a competent, highly reputable reviewer.  "Reliable enough", as it were.  (Of course, I guess his science is all wrong too )

So much for "objective gain in price performance".

1 hour ago, SharpeXB said:

Also the customer for this likely owns a 4K screen already.

Not really.  You're confining your perspective to people more like yourself, is my guess.

Of the people I build machines for - we're talking probably 100 machines over the past few years, at least...I've done exactly one for someone using 4k.  The others - and this includes older flight simmers, younger 'twitch gamers' and MMO/RPG players...were running 1080, 1440 or some wider aspect ratio variant of those.  All but 1 of 100 doesn't even want 4k, for their own reasons.

I've actually told people that even a 4090 is overkill for 1080.

But I'm still getting calls - from people running 1080p - asking when I can get them a 5090 *LMAO*  Of course, I (do my best to) talk them out of it...but the fact that I have to talk them out of it illustrates the problem.

It illustrates why Nvidia marketing BS and bragging rights are completely out of control.  People are spending tons of money that many of them don't honestly have, for crap that isn't really going to do much of anything for them.  Wanting 300+ frames when they use a 165Hz (or even 240) monitor and they're getting 250 already.

Nvidia wants people to forget empirical and quantifiable data, because then they'll spend money they don't need to, on promises that don't pan out.  They want people to buy on impulse, based on marketing blather, plain and simple.  It's wrong, it's misleading, and it's harmful to gamers in general.  And yes, that's an opinion...but it seems to agree with the data.  You know, science.

Edited January 23 by kksnowbear

[kksnowbear]

24 minutes ago, LucShep said:

Heh.... exactly as expected.
Now prepare for the even more disapointing reviews of the RTX5080 tomorrow....

Nah, no thanks.
Rather get a nice second hand motorcycle (or "whatever"!) than overspending on GPUs that will not make miracles in VR for a game that has its own long running issues to be solved.
 

This.  Indeed.  Amen.

[Aapje]

5 hours ago, kksnowbear said:

Reliable means consistently good in quality or performance; able to be trusted.

And all the words/phrases in that definition are subjective:

• consistently - subjective
• good - subjective
• quality - subjective
• performance - subjective
• able to be trusted - subjective

If you define a word based on a subjective definition, then the meaning of the word is of course also subjective.

Quote

You're just ticked off because you had no data to back up your claim when I called you on it, simple. 

No, because you are a hypocrite.

3 hours ago, kksnowbear said:

So much for "objective gain in price performance".

I was explicitly only talking about the 5080 and below and excluded the 5090 from that statement. So you are spreading a falsehood here.

Edited January 23 by Aapje

[kksnowbear]

7 minutes ago, Aapje said:

No, because you are a hypocrite.

More name-calling.  I asked if you could stop the personal attacks.  Guess not.

Well, at least now that we do have reliable data, I guess the matter of price performance is pretty much resolved.  No improvement in cost-per-frame.

Again, you can lead a horse...I am very glad that at least the majority seems to get it.

5080 tomorrow...one down, one to go

Edited January 23 by kksnowbear

[Aapje]

4 minutes ago, kksnowbear said:

More name-calling.  I asked if you could stop the personal attacks.

You guessed at the reason why I am reacting this way to you, and I told you the real reason.

If you start making stuff up, you should not be surprised when people don't want to let you get away with lies. Do you really expect that I would just let you get away with lying about me?

Quote

Again, you can lead a horse...I am very glad that at least the majority seems to get it.

What are you even talking about? I made my claim about the 5080 and below, so all these statements about the 5090 are not relevant to our discussion.

It's hard to have a discussion with you when you seem unable to keep track of the discussion and start making all kinds of false claims.

Edited January 23 by Aapje

[kksnowbear]

Seriously, could we stop the personal stuff and just discuss the subject at hand?

[kksnowbear]

9 minutes ago, Aapje said:

I made my claim about the 5080 and below, so all these statements about the 5090 are not relevant to our discussion.

Well, the 5080 reviews are coming up tomorrow so I guess we'll see.

(They're expected to be worse though, so there's that).

[Aapje]

7 minutes ago, kksnowbear said:

Seriously, could we stop the personal stuff and just discuss the subject at hand?

Sure, but only if you also stop with the passive aggressive attacks on me, while pretending that you are behaving nicely.

[kksnowbear]

39 minutes ago, Aapje said:

Sure, but only if you also stop with the passive aggressive attacks on me, while pretending that you are behaving nicely.

Sure, when you present data to back up your opinion, rather than stating it as if it's fact without any proof.

This is really simple: When you make a statement as if it's fact, be prepared to substantiate your position.  Show data; have proof.  Conversely, don't represent opinion as if it's fact.  Don't get all ticked off because someone asks you to show data that you know you don't have.

(Note there still is no data on purely rasterization performance of the 5080 and below, so...you couldn't very well have had data when you made the statement).  Unless you're a reviewer, in which case you'd likely just say you were under NDA. But I get a feeling you're not.

Edited January 23 by kksnowbear

[Aapje]

Just now, kksnowbear said:

Well, the 5080 reviews are coming up tomorrow so I guess we'll see.

(They're expected to be worse though, so there's that).

If the performance rumors are true, then the 5080 pretty much should be the worst card of the bunch. Lower gains than the 5070 (Ti), but no price reduction.

The 5090 seems to have almost no price/performance improvement, but it does gain VRAM, and in the top tier, price/performance is much less important to many buyers.

[LucShep]

Meanwhile....

TECHPOWERUP just released an interesting article about PCIe5.0 scalling on GPUs, and whether it makes sense to spend extra on a PCIe 5.0 (Gen 5 x16) motherboard just because of newest GPUs. (spoiler alert: no need, as expected - PCIe4.0 (Gen4 x16) and PCIe3.0 (Gen3 x16) dedicated slot still all good for the newest GPUs)

https://www.techpowerup.com/review/nvidia-geforce-rtx-5090-pci-express-scaling/
 

Edited January 23 by LucShep
added video