NVIDIA releases some RTX 2080 performance numbers and some info on DLSS

Yes, everyone is abandoning nVidia in droves in order to use AMD's FreeSync, crossfire, Tress FX, and every other copied nVidia technology (g-sync, sli, hairworks). But that i guess is to be expected from the company that got started making clones of Intel's processors (Intel 8088 and AM286).. /rolleyes

 

I do agree that the pricing is very aggressive and to the point of extreme, but I do know that GN is biased towards AMD and that's why I'm not subscribed.

 

My suspicion as well. "Sold out" in a lot of stores already? That sounds very fishy for a card without one single verifiable gaming benchmark.

On the other hand, miners all over the world have pockets deep enough to give it a try and hope they beat existing ones significantly. And if they not, who cares, they can just resell it to a gamer.

Btw, XBT and ETH seem to be slightly on the rise again (or at least not dropping anymore) since that official teaser video with "2080" all over it. Coincidence ? I don't believe in coincidences.

 

a lot of people have been holding out for two years chomping at the bit for something new.
I am not chomping that bad but show me numbers and I am ready to decide

 

You know there's an emoticon ?

That said:
G-Sync still requires expensive HW but what does it offer over FreeSync at this point? 3dfx made SLI long before Nvidia did. TressFX (Tomb Raider Mar. 2013) was out before Hairworks (CoD: Ghosts Nov. 2013), unless I'm mistaken?

Nothing stopping these stores from putting up small numbers on their store page to make it look like it's continually 'sold out' in an attempt to drive up demand.

 

Wrong thread.

 

G-sync has slightly less latency than freesync, but otherwise you are completely right. Tressfx even runs tons better than hairdoesntwork.

 

Very easy for any store to be "sold out" if they only had a tiny amount of units to start with. Hype usually builds up over any GPU release, especially one that has gone 2 1/2 years since last one.

 

In tests that matter Freesync has lower lag.
To be precise, at very low fps like 30~40fps, GSync has bit smaller lag than Freesync. Yeah, like I would care if that slideshow does respond 4ms faster to input.
At higher fps like 100+ Freesync input lag is smaller than GSync. I bet everyone understands what it means. It is better there where it matters.

 

Gsync is lower 144hz locked 142fps.


@time 14:04

Doesn't really matter as increased lag would not be perceptible with either set.

 

Do you understand that with so small difference 1~2ms in result, it is GSync which is slower?
Don't you know why?
That's because of GPUs they compared. Added lag from frametime rendering of GTX 1080 is 1/2 of added lag from frametime rendering done by RX 570.
Locked fps or not does not change it. GTX 1080 simply produces frame in 1/2 of time it takes to RX570 to make it.

Result would go other way around if you took Vega 64 vs GTX 1060. Don't they understand what entire "motion to pixel" is made of?

It is highlighted by them picking overwatch which has internal framerate limiter broken. That's why their 300fps (in-game) limit makes 35ms average input delay instead of 14 when externally limited to 144fps.

But you are right, that nobody would really notice 1~2ms difference. Thay could have made simple DXtest project which ticks internally 10000 times per second and has just black background and one polygon which changes color as it gets left-mouse-down information.

 

You did not read the graph properly.

You compare the difference of freesync off vs on.
144hz with cap @ 142fps

Freesync off / Freesync on
Max = 20.83 / 21.67
Avg = 15 / 15.28
Min = 11.67 / 11.67

Max showed almost 1ms more, avg showed .28 ms increase in input lag

Compare the same way to Gsync.

Gsync only adds .06 ms to input lag.
It also shows less min and the same max.

That shows gsync has less lag 'where it matters' as you say.

But like I said, either way is imperceptible.

Lastly, I'm very familiar with overwatch and how it works, the ingame fps limiter is much better for input lag than an external limiter.
Nor does the 300fps max cause any increase in input lag, which battlenonsense verifies multiple times in his analysis videos.

 

You lived without lots of things for 20 years, is anything innovative?

 

If all you played for 20 years was a mine sweeper, no, you don't need it and never will.

 

He talked about problems he faced. Sadly he did not show frametime graph in video. It is good idea to look at it, I tell you. I am not taking anyone's word for their statement about frametime w/o rtss graph in video.
Secondly, he took very few samples, that's why you think nV side does not change in delay. It is just not measured.
At 14:08 check 60Hz 60fps GSync OFF/ON... GSync increased average fps by ~1ms, but at same time he measured that it decreased minimum and maximum delay by ~1ms.
Then who can blame him, 1200fps camera is not enough. It captures frame every 0,833ms => 0,41667ms average error (0,833ms max error)... Considering he is measuring as small time frame as 9ms, it is huge error.
So, result like: Max = 20.83 / 21.67 is actually just 1 frame which either falls within 1 frame window where camera does not capture change or it hits the edge and small time difference is measured as high.

I did ask people around about Overwatch and them posting RTSS frametime graph before, not much success on our little forums. People rather do not want to know.

 

3dMark Timespy score for RTX 2080
https://videocardz.com/77763/nvidia-geforce-rtx-2080-3dmark-timespy-result-leaks-out

 

Looks kinda fake to me. Could be a custom overclock on a non 2080 card.

 

Looks perfectly in line with expected performance. What would constitute a 'non-fake' score to you?

 

Official review scores by Hilbert