Playstation Returnal Launched ! Resogun creators (Housemarques) PlayStation 5 Exclusive

[Necrokiller]

Woh criteria frame rate check karnay ke liye hai, after frame times have already been examined

And since Returnal has stuttering on PS5, and CerebralTiger is telling us that its "minor", so a 58fps 5th percentile on PC will also prove that the stuttering is minor.

 

[CerebralTiger]

PS5 se tou kaafi zyada instances of stutter houn ge when you factor in shader compilation

 

[Necrokiller]

PS5 se tou kaafi zyada instances of stutter houn ge when you factor in shader compilation

Agar "kaafi zyaada" hue to 5th percentile wouldn't be 58fps or higher. We already know what's "acceptable" is an ever sliding scale for CerebralTiger. Stay toooooooned I guess

 

[CerebralTiger]

Woh criteria frame rate check karnay ke liye hai, after frame times have already been examined

 

[Necrokiller]

Can't have high FPS percentiles with "PS5 se bhi zyaada" frame time spikes. CerebralTiger ko bas shauk hai ICU bharrne ka

 

[CerebralTiger]

Frame times ka ICU criteria different hai. As Tom suggested in his Stray analysis, even random infrequent stutters can be annoying

Returnal's stutters on PS5 are only during area loads, and, therefore, not random and certainly not during combat (just like Sifu's stutters on PC)

 

[Necrokiller]

CerebralTiger still dodging the fact that frame times spikes affect fps percentiles too. Afterall according to CerebralTiger game drops to 0fps.

As long as its above 5th percentile 58fps, fiiiine by meee

 

[CerebralTiger]

CerebralTiger still dodging the fact that frame times spikes affect fps percentiles too.

But not equivalently. Often times, a minor drop of 1-5 fps is accompanied by a massive 'stutterous' frame time spike

 

[Necrokiller]

But not equivalently. Often times, a minor drop of 1-5 fps is accompanied by a massive 'stutterous' frame time spike

lol @ often times. Didn't anybody tell CerebralTiger that all percentiles are calculated using the frame times?

But where are dem 0 fps read outs during these "stutterous" stalls? Can CerebralTiger give us an example?

 

[CerebralTiger]

lol @ often times. Didn't anybody tell CerebralTiger that all percentiles are calculated using the frame times?

Technically, yes, because frame rate is a derivative of frame time to begin with. But frame rate readings on their own don't show the whole picture, as seen in the screenshots above

But where are dem 0 fps read outs during these "stutterous" stalls? Can CerebralTiger give us an example?

They're effectively drops to 0fps, but you won't see them on the frame rate counter (unless it's similar to those macro-pauses in Elden Ring )

 

[Necrokiller]

Technically, yes, because frame rate is a derivative of frame time to begin with. But frame rate readings on their own don't show the whole picture, as seen in the screenshots above

Jab percentiles calculate hi frame times se ho rahe hain tou complete picture hi mile gi. It disregards the fps counter in the screenshot above, not the frame times.

They're effectively drops to 0fps, but you won't see them on the frame rate counter (unless it's similar to those macro-pauses in Elden Ring )

Lmao @ effectively

 

[CerebralTiger]

Jab percentiles calculate hi frame times se ho rahe hain tou complete picture hi mile gi. It disregards the fps counter in the screenshot above, not the frame times.

1st and 5th percentile frame rate are calcooolated using, well, frame rate readings. It doesn't account for those effective 0fps drops

 

[Necrokiller]

1st and 5th percentile frame rate are calcooolated using, well, frame rate readings. It doesn't account for those effective 0fps drops

Yiiiikessss

No they are not. Unless CerebralTiger just exposed VGTech

Ab CerebralTiger gaya research karne lmao

 

[CerebralTiger]

Like I said, frame rates are derived from frame times in the first place, so saying that all percentiles are frame time based isn't wrong. However, the percentile frame rates don't take frame time spikes into account. Frame rate is limited information to begin with

 

[Necrokiller]

However, the percentile frame rates don't take frame time spikes into account.

They do
And if we know for sure that certain testers aren't doing it, then their methodology is misleading. The whole point of percentiles and lows is granularity from frame times that simple fps readings cannot provide.

 

[CerebralTiger]

They do

Nope. Frame rate only tells you how many frames are delivered in a second. It tells you nothing about how evenly they are delivered

Unless CerebralTiger just exposed VGTech

Curious to see where and how VG Tech refutes the above statement

 

[Necrokiller]

Nope. Frame rate only tells you how many frames are delivered in a second. It tells you nothing about how evenly they are delivered

I'm talking about percentiles which are calculated from frame times. Tools convert the readings into FPS which is a way of displaying the end result in a different way albeit coming from the same source i.e. frame times

 

[CerebralTiger]

I mean, that's how all frame rate overlays ought to be converting it, much like the one used by Digital Foundry. But as we can see with the example of that 100ms+ frame time spike in Stray at 55fps, the two readings don't always vary consistently due other bottlenecks (e.g. shaders weren't ready in time).

We know:

8.3ms => 120fps
16.6ms => 60fps
25ms => 40fps
33ms => 30fps
EternalBlizzard => kthxbai

 

[Necrokiller]

I mean, that's how all frame rate overlays ought to be converting it, much like the one used by Digital Foundry. But as we can see with the example of that 100ms+ frame time spike in Stray at 55fps, the two readings don't always vary consistently due other bottlenecks (e.g. shaders weren't ready in time).

I didn't say anything about the FPS counter though, which is measuring frames delivered per second.

I've been talking about the percentiles which the tools calculate based on frame times. The end result can be represented in FPS (and it is when it says "percentile frame rate") but the data is derived from frame times exactly because the fps counter does not have that level of granularity

So yeah, when I share 5th percentile fps from Returnal, you can safely assume the data is calculated using frame times

 

[CerebralTiger]

So yeah, when I share 5th percentile fps from Returnal, you can safely assume the data is calculated using frame times

While that's good to know, it's a meaningless statistic for a game that stutters. Stray, for example, would have a lower 1st and 5th percentile frame rate targeting 60fps on PS5 than it would targeting a higher arbitrary frame rate on PC, despite being a worse overall experience on the latter, as correctly highlighted by Tom and Alex

A better way to gauge performance in such games is to look at the target frame time and compare it with the number of frames that weren't delivered accordingly. Then, exclude the ones that are within the range of twice of the target frame time (e.g. 33.3ms for the target frame time of 16.6ms) on account of judder. The ones that remain are stutters