Official Final Fantasy XV Benchmarks Feature the Nvidia GeForce RTX 2080 & 2080 Ti

Those performance numbers aren't bad at all and I'm pretty excited about proper ray tracing. I'm never an early adopter though. I know this generation of the tech and probably the next one will mostly suck. I don't need a new card yet.
 
sleepybp said:
If legit, this benchmark confirms fact that 1080ti is best buy for 1440P or 1080P gaming.
Unless you own 4k display, and plan to game at that resolution, the 1080ti used at $500 cant be beat.

The 20 series appears for the 1% only who game at 4k
Click to expand...


According to this poll in the Video Cards sub, more than 1% on UHD displays. Closer to 30%.
 
Dayaks said:
If it’s not 100% usage at the standard settings there’s a CPU bottleneck which may still exist for the 2080ti at max settings.
Click to expand...


If there is a CPU bottleneck at 50% eyecandy, there is definitely going to be one at 100%. More GPU probably isn't going to resolve that.
 
alxlwson said:
If there is a CPU bottleneck at 50% eyecandy, there is definitely going to be one at 100%. More GPU probably isn't going to resolve that.
Click to expand...

I am saying both could be CPU bottlenecked at 50% eyecandy. At max the 1080ti could be going full bore but the 2080ti hits the CPU limit.

Or for an analogy you have identical cranes with a 100HP pump and another with 150HP. At 1ft/s they both are pump limited. At 2ft/s you are power limited (blow a breaker?) on the 100hp but the 150hp is still pump limited.

I didn’t work on cranes much so that could be asinine numbers heh.

I guess tomorrow we’ll find out.
 
viper1152012 said:
So the 2080 is almost the same as the 1080ti.....

Minus RTX being turned on...

So at the 700-800$ range if you switch on RTX at 1080p you should get better performance easily out of the 1080ti....

Unless you opt for the expensive Tita... I mean 2080ti... Which should do ok at 1080p....

Unless you don't want RTX then you can 4k on the 2080ti(Titan)....

GOD, What a messed up selection....

Why didn't they wait for a node shrink? This just seems like a rehash with new tricks that can slaughter performance for synthetic ray trace like graphics.

I guess if you were waiting for the Luster series card or ultra reflective everything this is your train ehh.

Here's hoping I'm wrong :!
Click to expand...
Turing is a node shrink. It is on 12nm while Pascal is 16nm.
 
Dayaks said:
I am saying both could be CPU bottlenecked at 50% eyecandy. At max the 1080ti could be going full bore but the 2080ti hits the CPU limit.

Or for an analogy you have identical cranes with a 100HP pump and another with 150HP. At 1ft/s they both are pump limited. At 2ft/s you are power limited (blow a breaker?) on the 100hp but the 150hp is still pump limited.

I didn’t work on cranes much so that could be asinine numbers heh.

I guess tomorrow we’ll find out.
Click to expand...


Haha. I don't work on mobile cranes, but a power source is a power source. Industrial EOT cranes don't use hydraulics for power transmission. Good old fashioned 3ph motors and speed reducers.
Could have a 200HP motor, but your VFD is only good for 150HP.
Either way, your analogy is correct :)
 
alxlwson said:
If there is a CPU bottleneck at 50% eyecandy, there is definitely going to be one at 100%. More GPU probably isn't going to resolve that.
Click to expand...

...no?

The CPU bottleneck will still be there, but more GPU is going to increase performance measurably if not noticeably.
 
So it appears nVidia is only competing against themselves at the upper end..

My theory as to the pricing on RTX cards is this: The situation is that they still have significant stock of 1080 gpus, and they will only be competing against themselves. So to be able to move all the 1080 parts at or near the price they've been selling at all along, the RTX had to be in a league of its own price wise.

I don't feel the price to performance scaling is in line with RTX. The new generations are supposed to come out, be priced about the same as the previous generation it is replacing or maybe be $50 to $75 more, and be anywhere from 30% to 70% faster. The prior generation cards get a price cut. All the historical releases fit into this. But since they are only competing against themselves, price the newest generation really high and the prior generation can keep on living with a slower rate of price decay.

Good business sense for nVidia. Sucks for us. There is also the possibility that the partner manufacturers will have cards after launch for prices below the founders edition pricing. The founders cards of the past were priced $100 more than the partner cards so as not to directly compete. This time around, nVidia said "hey we will go ahead and sell overclocked founders cards, and you guys (partners) go ahead and sell your cards at launch at the same price. The market can bear it." Which it did since they've all sold out. A nice bone to throw to partners who a few months ago had huge stockpiles of 1080 gpus thanks to the mining bubble popping.

Not going to bitch about everyone who is buying the RTX cards though. I think they will be pretty sweet.

As far as these performance charts go, are the numbers for the older cards an average of all the results they've accumulated? Or the highest result anyone got? Either way, the old cards' entries are almost certainly influenced by overclocked results. So the way I see it, an overclocked 1080Ti closes the gap to the Founders edition (also somewhat overclocked?). Time will tell what speeds we get from EVGA, Asus, etc. on 2xxx cards, as well as what overclocks the new silicon is capable of. Being a larger die, I think it will have less oc potential than pascal. Even though it is a die shrink, the times when a die shrink guaranteed more speed are a thing of the past.
 
Last edited:
GoodBoy said:
So it appears nVidia is only competing against themselves at the upper end..

My theory as to the pricing on RTX cards is this: The situation is that they still have significant stock of 1080 gpus, and they will only be competing against themselves. So to be able to move all the 1080 parts at or near the price they've been selling at all along, the RTX had to be in a league of its own price wise.

I don't feel the price to performance scaling is in line with RTX. The new generations are supposed to come out, be priced about the same as the previous generation it is replacing or maybe be $50 to $75 more, and be anywhere from 30% to 70% faster. The prior generation cards get a price cut. All the historical releases fit into this. But since they are only competing against themselves, price the newest generation really high and the prior generation can keep on living with a slower rate of price decay.

Good business sense for nVidia. Sucks for us. There is also the possibility that the partner manufacturers will have cards after launch for prices below the founders edition pricing. The founders cards of the past were priced $100 more than the partner cards so as not to directly compete. This time around, nVidia said "hey we will go ahead and sell overclocked founders cards, and you guys (partners) go ahead and sell your cards at launch at the same price. The market can bear it." Which it did since they've all sold out. A nice bone to throw to partners who a few months ago had huge stockpiles of 1080 gpus thanks to the mining bubble popping.

Not going to bitch about everyone who is buying the RTX cards though. I think they will be pretty sweet.

As far as these performance charts go, are the numbers for the older cards an average of all the results they've accumulated? Or the highest result anyone got? Either way, the old cards' entries are almost certainly influenced by overclocked results. So the way I see it, an overclocked 1080Ti closes the gap to the Founders edition (also somewhat overclocked?). Time will tell what speeds we get from EVGA, Asus, etc. on 2xxx cards, as well as what overclocks the new silicon is capable of. Being a larger die, I think it will have less oc potential than pascal. Even though it is a die shrink, the times when a die shrink guaranteed more speed are a thing of the past.
Click to expand...

Do realize the die is 25% larger on the 2080 than the 1080ti. I am going with a price comparison here...

So it (supposively) is faster than a 1080ti and $100 more. With DLSS it blows the 1080ti out of the water. If you don’t care about RTX features then the value plummets. So it’s either an fantastic value or meh depending.

Personally I am super excited about DLSS 2X, which takes the native res, AI super scales it (I’ve read ~ equal to 1.8x SSAA) and scales it back down to native. That’s perfect for my 21:9 ultrawide.. “free” AA.
 
Virtual_Bomber said:
What other benefits? Besides status.
Click to expand...

Status works. Drive a Camaro up to the club. Nobody gives a shit, poser. Drive a Lambo, panties drop even if your face looks like Quasimodo and Medusa had a baby.

Typically this doesn't work with GPUs, unfortunately for us. If it did, I'd have bought two 2080 Tis and damn the price.
 
Dayaks said:
...So it (supposively) is faster than a 1080ti and $100 more. With DLSS it blows the 1080ti out of the water. If you don’t care about RTX features then the value plummets. So it’s either an fantastic value or meh depending.
Click to expand...

Yeah, I will be looking forward to all those reviews where the new features are explored. If DLSS looks as good as the best AA the older cards can do but allows performance to go nuts, awesome. I hope this is the case.
 
Mazzspeed said:
The problem is, even if there was competition next generation, the prices won't come down. This isn't the price of 'bleeding edge', this is the price - Period.
Click to expand...

Says who? What evidence supports this theory?
 
So the best i can every hope for, is something that is 2x faster than my 1070 non ti...
 
DuronBurgerMan said:
Status works. Drive a Camaro up to the club. Nobody gives a shit, poser. Drive a Lambo, panties drop even if your face looks like Quasimodo and Medusa had a baby.

Typically this doesn't work with GPUs, unfortunately for us. If it did, I'd have bought two 2080 Tis and damn the price.
Click to expand...

True, true. Im not denying that but I was just talking about the cars themselves which is why I said omit status. Other than status there isn't much difference when it comes to performance in most (and I emphasize MOST) supercars nowadays (for instance corvette vs Farrari) except price.

If the only reason you are buying a supercar is to get some pussy you are wasting your money when you could buy whores for cheaper and buy a $120K corvette that's a fucking beast instead of spending millions of dollars. If you take pussy out of the equation then it's a no brainer, then the comparison becomes easy to understand.

Nvidia makes a 100FPS card at 4k at $1500.
Some other company makes a 100 FPS card at 4k...all with the same features for $899 but the nvidia card is considered "exotic" so it costs more.........so which one are you going to buy? Yeah.....

The point of my original question is if you take the 'status symbol' out of the equation for high end cars the comparison is extremely similar to that of the video card market and the end result is the same. How fast is your 1/4 mile / how many frames can your video card render in a second?

Also, I do want to make clear that I agree with you statement that panties do drop in those cases. lol
 
viper1152012 said:
Its not actually, its a density increase on the 16. They are just calling it 12nm.. Its more of a 16++..

Thanks for the naming scheme Intel.
Click to expand...
They accomplished it by decreasing the gate pitch.

At least Intel's processes are actually closest to what they advertise.
 
Armenius said:
They accomplished it by decreasing the gate pitch.

At least Intel's processes are actually closest to what they advertise.
Click to expand...
True. https://semiengineering.com/nodes-vs-node-lets/

All about how gate pitch has been bastardized by marketing.

https://www.extremetech.com/computing/245880-rumor-nvidias-volta-built-tsmcs-new-12nm-process

TSMC literally saying 12 is just 16 but revised.

They were revered to as nodelets.
Improvements on existing node. Finfet folding cell marketing. Super tall and hard to cool
 
viper1152012 said:
Its not actually, its a density increase on the 16. They are just calling it 12nm.. Its more of a 16++..

Thanks for the naming scheme Intel.
Click to expand...

Intel didn't start the PR nm naming scheme...that was litterlly the rest of the "crowd"...
 
Virtual_Bomber said:
What other benefits? Besides status.




I mean the pricing is VERY shitty, Nvidia shouldn't be charging THAT fucking much for these cards.....BUT....since they are a generation ahead of the competition now once these things officially release and no one else is challenging them, then this is to be expected. For some reason no one can understand this.
Click to expand...

The other benefits are the extra performance (no matter how small) It's the same reason if I want a 14TB hard drive today I will pay through the nose. That extra bit of density isn't worth it to most people, but to some it is.

nvidia is a for profit business, they can charge anything they want for the card. A good is only worth what someone is willing to pay this works for the craptastic garbage people try and sell on craigslist, or the high end items only the ultra rich and afford, and if people are willing to pay $1200-1400 for it, why shouldn't they charge that much? At the end of the day it's a luxury good and no one is obligate to purchase it. Any rational person running a business to make profit does the same thing with their goods/services.

It doesn't help matters that AMD can't crank out anything that competes, and we are seeing a textbook case of where innovation and pure performance gives you a supreme edge in the market. Look what happened to intel's high end once AMD finally cranked out threadripper.
 
Virtual_Bomber said:
True, true. Im not denying that but I was just talking about the cars themselves which is why I said omit status. Other than status there isn't much difference when it comes to performance in most (and I emphasize MOST) supercars nowadays (for instance corvette vs Farrari) except price.
Click to expand...

It's relative. With cars, as with computers, performance scaling is more or less linear, while price is exponential. It costs relatively little to go from a 1050 to 1060, and only a bit more to go to a 1070 after that. But as you start chasing 2080 Ti levels... price is going stratospheric. Before the 20xx series, the Titan V really demonstrated that well enough. Supercars work similarly. Camaro -> Corvette is like +$40k. Corvette -> Bugatti is like +$2 million. The Bugatti isn't going to be 20 times fast. It might be only 20% faster. You pay more to play when you're at that level. Each price bracket costs a lot more than the previous, but offers similar (and sometimes lesser) performance increases.

If the only reason you are buying a supercar is to get some pussy you are wasting your money when you could buy whores for cheaper and buy a $120K corvette that's a fucking beast instead of spending millions of dollars. If you take pussy out of the equation then it's a no brainer, then the comparison becomes easy to understand.
Click to expand...

If whores were legal. Given that they (generally) aren't in most places, the Lambo would probably do you better if you're looking for that kind of thing.

Nvidia makes a 100FPS card at 4k at $1500.
Some other company makes a 100 FPS card at 4k...all with the same features for $899 but the nvidia card is considered "exotic" so it costs more.........so which one are you going to buy? Yeah.....
Click to expand...

That's only because with computers, brand loyalty is relatively weak, and brand identity practically nonexistent. Oh sure, Nvidia fanboys exist, same as Intel fanboys, AMD fanboys, etc... But if you say "my card is an NVIDIA card" nobody here is likely to give a shit. Which card is it? How much did you OC it? What's your FPS in X? Those matter. Fanboys mostly just rationalize purchases after the fact, more or less. But say "I own a Lambo" and it almost doesn't matter how fast it is. It's a Lambo. The brand identity and brand loyalty is through the roof. Performance is like... okay, it's probably fast-ish. Beyond that, nobody cares.

Also, I do want to make clear that I agree with you statement that panties do drop in those cases. lol
Click to expand...

They do. Anybody who says otherwise is full of shit.
 
DuronBurgerMan said:
It's relative. With cars, as with computers, performance scaling is more or less linear, while price is exponential. It costs relatively little to go from a 1050 to 1060, and only a bit more to go to a 1070 after that. But as you start chasing 2080 Ti levels... price is going stratospheric. Before the 20xx series, the Titan V really demonstrated that well enough. Supercars work similarly. Camaro -> Corvette is like +$40k. Corvette -> Bugatti is like +$2 million. The Bugatti isn't going to be 20 times fast. It might be only 20% faster. You pay more to play when you're at that level. Each price bracket costs a lot more than the previous, but offers similar (and sometimes lesser) performance increases.
Click to expand...

I agree with your pricing explanation to an extent, the shit part is is that Nvidia is pricing their cards like they are ultra rare lambos. Which it isn't, they are price gouging at this point but no one is challenging them and they know they have the fastest thing right now so they do it.

When you get up to those numbers (for vehicles) you are paying for raw status and the price doesnt scale (as you said) to performance. You could buy a Corvette, drop 50-100 grand into it and it would eat most million dollar cars for breakfast while retaining comfort for a fraction of the price. You are buying the name, that is it really. In the PC world it is all about performance which was my original argument, if cars were priced according to performance (and that can mean alot of different things when it comes to vehicles) and ONLY that, 99.9 percent of people would fucking laugh at supercar manufacturers like they were idiots......like damn near everyone in this thread is laughing and shaking their heads at Nvidia right now. They do have the fastest card right now but this shit is ridiculously priced and everyone knows it.

I'm still not sure if I'm conveying my point correctly or not....


DuronBurgerMan said:
If whores were legal. Given that they (generally) aren't in most places, the Lambo would probably do you better if you're looking for that kind of thing.
Click to expand...

Just because they aren't legal doesn't mean it can't happen. :angelic:

DuronBurgerMan said:
That's only because with computers, brand loyalty is relatively weak, and brand identity practically nonexistent. Oh sure, Nvidia fanboys exist, same as Intel fanboys, AMD fanboys, etc... But if you say "my card is an NVIDIA card" nobody here is likely to give a shit. Which card is it? How much did you OC it? What's your FPS in X? Those matter. Fanboys mostly just rationalize purchases after the fact, more or less. But say "I own a Lambo" and it almost doesn't matter how fast it is. It's a Lambo. The brand identity and brand loyalty is through the roof. Performance is like... okay, it's probably fast-ish. Beyond that, nobody cares.
Click to expand...

Agreed.
 
You must log in or register to reply here.
Back
Top