gamegpuEverybodys Gone to the Rapture Benchmarks

[csbin]

http://gamegpu.com/rpg/р&#108...1077;/everybodys-gone-to-the-rapture-test-gpu


Nvidia GeForce/ION Driver Release 364.72

AMD Radeon Crimson Edition 16.4.1

 

[Silverforce11]

For a walking simulator, the performance is pretty bad overall.

 

[StrangerGuy]

For a walking simulator, the performance is pretty bad overall.

Nah, it's OK as long as we can tax a 980 Ti to oblivion in mere 1080p, it doesn't matter how ho-hum the graphics are. We got our money's worth because that is obviously progress.

 

[PhonakV30]

this game is ugly.

 

[ThatBuzzkiller]

DX12 and console optimizations can't come fast enough ...

 

[AtenRa]

wow that GTX980Ti performance, on CryEngine ??

 

[Bacon1]

Wow performance is terrible all around, takes a 980 just to hit 30 min FPS @ 1080p :O

Gotta be a worse release than even Quantum Break, and people can't try to blame UWP for this monstrosity.

 

[Pinstripe]

The trick is not to play everything on Ultra. Adjust some settings a notch or two down, and enjoy smooth framerate while maintaining virtually the same graphics fidelity.

But I know I know, our ego won't allow that.

 

[dark zero]

NO card is playable at 4K, no SLI or CF support and no Fury performance..... holy crap... this is the Crysis of this year.

 

[Arachnotronic]

Poorly written junk

 

[StrangerGuy]

NO card is playable at 4K, no SLI or CF support and no Fury performance..... holy crap... this is the Crysis of this year.

Eh, no that's insulting Crysis, where it's vast IQ increases over than existing games justifies the similar increases in requirements on hardware. Now, we have games that look and somehow runs even worse than older ones while developers give pathetic excuses about DX12 as a red herring to steer attention away from how terrible they are at their jobs.

 

[Ma_Deuce]

Am I missing something on their graphs? It seems like whenever it's close, they put nvidia on the top. Even if the numbers for AMD look better such as the 960 vs 380x in 1080p

 

[myocardia]

missed something

 

[MrTeal]

missed something

What did he miss? They're ranked by min framerate, but in the case of the 295X2 and 780/780Ti, when they tied at 30 for min framerate the 295X2 was on top with 47 average. For the 960 and 380 at the same 24 min though, the 380X has a higher average but is placed below the 960. Same with the 280X and 680, and the 690 and 950.

 

[Hitman928]

What did he miss? They're ranked by min framerate, but in the case of the 295X2 and 780/780Ti, when they tied at 30 for min framerate the 295X2 was on top with 47 average. For the 960 and 380 at the same 24 min though, the 380X has a higher average but is placed below the 960. Same with the 280X and 680, and the 690 and 950.

I believe their ranked by minimum frame rate including fractions of an fps, even though what you see is the rounded value.

For example:

Card A = 45 avg, 35.4 min (35 is shown on the graph)
Card B = 50 avg, 35.3 min (35 is shown on the graph)

Card A would be ranked higher than card B in the graph. I think it's a stupid way to do it, but that's how I understand the ranking.

 

[kraatus77]

I believe their ranked by minimum frame rate including fractions of an fps, even though what you see is the rounded value.

For example:

Card A = 45 avg, 35.4 min (35 is shown on the graph)
Card B = 50 avg, 35.3 min (35 is shown on the graph)

Card A would be ranked higher than card B in the graph. I think it's a stupid way to do it, but that's how I understand the ranking.

you can't get minimum/maximum with x.y value lol. only the average. there's no 35.4 minimum, only 35. l2math

 

[Hitman928]

you can't get minimum/maximum with x.y value lol. only the average. there's no 35.4 minimum, only 35. l2math

You can if you do multiple tests and take the average of the minimums. . .

If they're only doing one run for each card on an unscripted benchmark run, then there's more problems than the ranking system.

 

[Erenhardt]

minimum FPS is actually the lowest average in the probing interval. Frames per second are not measured in 1 second intervals, most of the time.

For example:
Game runs at around 30 fps, each frame is delivered every 33ms.
Probing fps with 15 ms intervals would results in:
0ms = 0/3 frames - start point
20ms = 0/3 frames - 0 new frames avg fps = 0
40ms = 1/3 frame - 1 new frame, avg fps = 50 fps
60ms = 1/3 frame - 0 new frames, avg fps = 0
80ms = 2/3 frame - 1 new frame, avg fps = 50 fps
100ms = 3/3 frames - 1 new frame, avg fps = 50 fps

overall:
time 0,1s
frames = 3
avg fps = 30 fps
min fps = 0 fps (at 20ms and 60ms marks)

Ofc noone is doing measurements that often (it was just to show the point that min fps is an average), but 100ms is a default setting in Rivatuner statistics/Afterburner IRC

 

[antihelten]

minimum FPS is actually the lowest average in the probing interval. Frames per second are not measured in 1 second intervals, most of the time.

For example:
Game runs at around 30 fps, each frame is delivered every 33ms.
Probing fps with 15 ms intervals would results in:
0ms = 0/3 frames - start point
20ms = 0/3 frames - 0 new frames avg fps = 0
40ms = 1/3 frame - 1 new frame, avg fps = 50 fps
60ms = 1/3 frame - 0 new frames, avg fps = 0
80ms = 2/3 frame - 1 new frame, avg fps = 50 fps
100ms = 3/3 frames - 1 new frame, avg fps = 50 fps

overall:
time 0,1s
frames = 3
avg fps = 30 fps
min fps = 0 fps (at 20ms and 60ms marks)

Ofc noone is doing measurements that often (it was just to show the point that min fps is an average), but 100ms is a default setting in Rivatuner statistics/Afterburner IRC

This is one way of doing it, but I'm fairly certain that for the majority of games the minimum FPS is simply calculated from the singular frame that had the highest rendering time (this is the way FRAPS works at least).

you can't get minimum/maximum with x.y value lol. only the average. there's no 35.4 minimum, only 35. l2math

Of course you can have min/max FPS with decimal places. For instance imagine that the slowest frame takes 38 milliseconds to render, that would then be equal to 1000/38 = 26.315789... FPS

The monitor refreshes at fixed intervals obviously (unless you have a GSYNC/FreeSync monitor), but this will only affect the minimum FPS if you're running with VSYNC on (which they obviously weren't in this case).

 

[airfathaaaaa]

The trick is not to play everything on Ultra. Adjust some settings a notch or two down, and enjoy smooth framerate while maintaining virtually the same graphics fidelity.

But I know I know, our ego won't allow that.

if a 980ti cant max out a freaking game on 1080p then i dont want to play that game its simple

 

[kraatus77]

This is one way of doing it, but I'm fairly certain that for the majority of games the minimum FPS is simply calculated from the singular frame that had the highest rendering time (this is the way FRAPS works at least).




Of course you can have min/max FPS with decimal places. For instance imagine that the slowest frame takes 38 milliseconds to render, that would then be equal to 1000/38 = 26.315789... FPS

The monitor refreshes at fixed intervals obviously (unless you have a GSYNC/FreeSync monitor), but this will only affect the minimum FPS if you're running with VSYNC on (which they obviously weren't in this case).

fps, not render time. there's no frame in decimals. because at any given second you can't have x.y frames on screen only x frames. it's a freakin single image not x.y images. but the calculation we do makes those differences.

in that situation it will be 1000/38.461ms to render 26fps not the other way around.


when you take a pic, it's a single image not one and a half. don't know if i can explain it any better, im limited by my crappy English

 

[Noctifer616]

fps, not render time. there's no frame in decimals. because at any given second you can't have x.y frames on screen only x frames. it's a freakin single image not x.y images. but the calculation we do makes those differences.

in that situation it will be 1000/38.461ms to render 26fps not the other way around.


when you take a pic, it's a single image not one and a half. don't know if i can explain it any better, im limited by my crappy English

Actually you are wrong. Pixels on your screen don't all refresh at the same time, so yea, you can actually have half of the current frame and half of the last frame, or even parts of three frames at once on screen. Just look up screen tearing.

 

[antihelten]

fps, not render time. there's no frame in decimals. because at any given second you can't have x.y frames on screen only x frames. it's a freakin single image not x.y images. but the calculation we do makes those differences.

in that situation it will be 1000/38.461ms to render 26fps not the other way around.


when you take a pic, it's a single image not one and a half. don't know if i can explain it any better, im limited by my crappy English

I'm sorry but you're just plain wrong. FPS is calculated from frame times (since it's just the inverse), and as Noctifer pointed out above you can absolutely have fractional frames on screen (screen tearing).

Even if you use a GSYNC/FreeSync monitor and thus avoid tearing, you can still have a fractional FPS value, since such a monitor does not necessarily refresh at an interval that gives an integer value when divided over a second. For instance if said monitor refreshes once every 9 ms, then that is equal to 111.111... FPS.

 

[sm625]

if a 980ti cant max out a freaking game on 1080p then i dont want to play that game its simple

Why? It would seem to me that a game should run around 30 min fps on the very max settings on the top end card. Otherwise you are kind of making the same claim that some people make about speedometers.... which is that they should only read 0-85mph max because that is the range most people ever use.

 

[cytg111]

Game has some nice screens outthere. I can see how some of the forrest scenes may be taxing, think ill look up some actual gameplay later.