Discussion in 'Frontpage news' started by Hilbert Hagedoorn, Nov 9, 2018.
technology doesn't advance by asking you what you want.
@Hilbert Hagedoorn Could you maybe add a Crossfire configuration to the test? I am wondering if Crossfire is properly supported in this game.
as i think i posted already, EA said RTX will be added in a patch closer to the deluxe edition release not "scientists have no idea" lol
and tomorrow we are due a patch for pc that weighs in at *unsure about size*. so if i were to bet when RTX is being added i'd say tomorrow at 1am pst.
here is where they posted that info:
RT at this point is just PhysX 2.0 it’s a really cool feature but there isn’t any hardware capable of using it and maintaining performance. By the time there are enough games that use it the next generation will be out it will be refined and along with hardware that will make it viable feature and perhaps a mainstream feature and not just limited to flagship products.
Yeah, people saying 70%, or 100%, or so many wild numbers are over-exaggerating the price difference, or comparing MSRP of previous generations to non-MSRP prices of current generations which has nothing to do with Nvidia.
GTX 1080 ti MSRP $699 - RTX 2080 ti MSRP $999 = 43% price increase (With a die size that no one talks about increase from 471mm2 to 754mm2, an increase of 60%, which directly affects prices of products, again, that no one wants to talk about, they are too busy worrying about complaining)
GTX 1080 MSRP $549 - RTX 2080 MSRP $699 = 27% price increase (die size increase from 314mm2 to 545mm2, an increase of 71%)
GTX 1070 MSRP $379 - RTX 1070 MSRP $499 = 31% price increase (die size increase from 314mm2 to 445mm2, an increase of 41%)
In times of movement from HD 3xxx to HD 4xxx to HD 5xxx transistor count almost doubled with each generation as SP count doubled. Prices did not skyrocket in process.
Same goes for previous generations. You got gradual increase in performance and beefier GPU without severe increase of price.
Yes, there was 40% increase in launch price going from GTX 680 to 780Ti, but that was GPU with double transistor count on same node and almost double area per chip.
GTX 780 Ti launched at same $699 USD price, 980 Ti launched at $649 USD.
2080 (545mm^2) is smaller chip than 780Ti (561mm^2) and 980Ti (601mm^2).
Those Ti cards of past were definitely ones with lowest yields. So 2080 being smaller can easily be considered as having reasonably fewer defects since 12nm is tweaked 16nm and much bigger 2080 Ti is around in large volumes.
In other words, pretty tiny performance increase this time around at same price points.
I bought my GTX 1080 TI for 760 dollars and at the current price point where i live in europe the 2080TI is 1672 dollars that is around 50% my bad still to expensive all things considered in my opinion, i did actually pre-order the 2080 TI but decided to wait for reviews and once they came out i when out and got one but after 2 days of playing around with it, i could not justify the price vs performance but then again i only play in 3440x1440 might have viewed it different if i was playing at 4K. So back it whent.
Early days for a brand new tech , hopefully within the next 3 years we should have affordable video cards on all price range that can do RTX properly , we are not there yet.
Too bad, your individual prices based off of where you live or GPU manufacturers inflating prices has nothing to do with MSRP and nothing to do with the 2000 series or nvidia.
This is fact.
If you don't like this fact and can not deal with this fact then you're going to be sorely disappointed throughout life when things just like this happen continuously. There are items in Europe that cost more in the USA, either due to the fact they do not exist in the USA, import fees, local shops deciding to charge more "because they can", among so many other reasons. I do not go to the manufacturer of said products and complain their items are too expensive because of 3rd party decisions to make it more expensive.
Radeon HD 3870, die size 192mm2, price $219
Radeon HD 4870, die size 256mm2, price $299
Radeon HD 5870, die size 334mm2, price $379
From HD3000 series to 4000 series, price increased by 36%
From HD4000 series to 5000 series, price increased by 26%
And considering the fact that the 3000 series and 4000 series were mostly the same technology, a refresh, unlike Pascal to Turin, with both the 3000 and 4000 series being TeraScale 1, and the 5 series being a new(ish?)generation, TeraScale 2:
Terascale 1 to terascale 2: price increase of 73%
This isn't even taking into consideration the:
Radeon HD 5870 Eyefinity Edition @ $479
So i don't quite understand what your point is when you're trying to say the 3000, 4000 and 5000 series did not have any "jumps" in prices when the jumps obviously shown here are extremely similar to the jumps in prices from the nvidia 1000 to 2000 series and don't even offer a brand new technology that is finally allowing something that has been sought after in games for 10+ years.
I can handle it if i wanted to pay the price and felt like i was getting what i wanted i would just buy it end of story no point bickering about it, i wont pay twice the price for 30% performance increase just not worth it.
Performance per $ improvement with each generation. What else you want to discuss? What else is there to care about? It is GPU and user expects it to render frames at certain rate depending on price.
So, take some price point like $500 and pick best GPU priced at that point every year. See how performance increases.
You can do it from layman's perspective and include inflation. Or you can do it from business perspective where inflation is ignored as one expects increased efficiency delivered through technological improvements.
(So, yes. Your example of $479 HD5870 is kind of irrelevant when it is 26% above regular HD5870. Try to at least simulate professional who removes distraction signals instead of adding irrelevant data into statistics.)
Otherwise you'll end up with very bad results. Like, doing statistics about $2000 cars over time and seeing their horse power, mileage per gallon, and QoL improvement for passengers. And then you go and start including $5000 models of same car.
= = = =
Please note that I am not saying that RTX cards are bad. I say exact opposite and anyone thinking about getting GTX 1080 vs RTX 2070 should get RTX because it is going to deliver similar improvement/$ over time as cards in past due to new and currently unused improvements. (That's unless RTX goes Vega and some of them will prove to be DoA.)
Cards are good, but year-2-year improvement in performance at any given price point is far from being as good as before.
= = = =
And there is another thing to consider:
- 2070 came almost 2 and half years after 1080 which allowed it to have 1.5times more transistors (10.8B which is just tiny step from GTX 1080Ti having 11.8B) and still be cheaper to make => Efficiency through technological improvements
=> So, would you rather have 12nm $500 RTX 2070 or 12nm $550 GTX 1080Ti clocking 10% higher than 16nm version?
=> nVidia decided to bring RTX with its small improvement in performance/$ instead of delivering you Titan Xp performance at $550 mark
I'm trying to understand what you're saying here. I didn't add the $479 into the statistical information, but it IS a card that is released, and it WAS massively more expensive then the rest, and better yet? All because it has 6 mini-display ports, there was no other difference. So now you're saying reality is not....professional? or, a distraction? How is reality a distraction? Just because it upsets you that the list of GPUs you posted don't actually agree with what you stated?
As to the rest of your comment, everything i replied to, was a direct statement to what you replied to me, if you want to discuss other information that's fine, but then we'd be talking till the end of time and i don't have a desire to do that.
My entire point was, in my original post: Die size matters, die size increases prices, and thought it does not ALWAYS increase price due to competition forcing them(either of them) to charge less to be competitive, it should not be unexpected that a die size increase will increase the price of a product.
My only other point was, in my original post: People are over exaggerating prices, and/or blaming nvidia for third party companies price inflation. The prices of the 2000 series, non-FE cards, are $499, $699 and $999, anything other then those prices are not nvidias doing. We didn't blame nvidia or AMD for releasing cards for $499 and having 3rd party companies sell for $549-649 in the past, there's no reason to blame nvidia now for the same thing happening, blame the 3rd part companies.
Any other discussions are irrelevant to my original post and what you replied to me in the first place.
Specific cards always had higher price for one or another reason. 6x DP is increased value for target demography and 26% more is not that bad if you consider that you would have to buy 2 cards otherwise.
I used MSRP for reason. Because if I went and took most expensive non-reference 2080 for example, it would have 34% higher price than cheapest one. (Not single extra port, only 4% higher clock out of the box.)
And I can ensure you that they are not exactly over exaggerating those prices. Because 2.5years between release of 1080 and 2070 is plenty time to have technological improvements like in past.
1080 was 33% smaller than 1080Ti. 2070 is 41% smaller than 2080 Ti. If we say that both top GPUs were feasible in similar fashion, then 2070 is much more defect free than 1080.
Why? Because not only 1080 is closer in size to 1080 Ti on same manufacturing node, but it took almost one more year till nVidia released 1080Ti. (yields were probably not so good)
And with RTX 20x0 (Ti), nVidia easily released all GPUs in close succession with great availability. That 12nm is doing well even on bigger GPUs. (And why would not it have smaller defect rate when TSMC is already at 7nm. They are rocking their game.)