Current Generation Games Analysis Technical Discussion [2020-2021] [XBSX|S, PS5, PC]

[BRiT]

Hard to understand what's going on from that write-up -- are they saying the ps5 is doing checkerboard reconstruction, but the xsx is running native 4k? That seems like a gigantic gap in pixel count being pushed thats kinda being brushed over?

Yes, if PS5 is doing checkerboarding then it's native horizontal resolution is cut in half. If you run the numbers it's around 60%-70% more pixels on Series X in performance (depending on ranges).

 

[snc]

PS5 has some headroom as it mostly renders at max res:

While XSX mostly renders sub-native:

So, yes, CBR was probably not the best way on PS5, particularly considering the possible CBR artifacts. But at least the framerate is the most stable on PS5 I guess.

I think they just use ps4pro settings in performance mode on ps5, definitly over agressive as in high resolution mode is almost same as on xsx

PS5 in quality mode uses a dynamic resolution with the highest resolution found being 3840x2160 and the lowest resolution found being 3200x1800. Xbox Series X in quality mode uses a dynamic resolution with the highest resolution found being 3840x2160 and the lowest resolution found being approximately 3413x1920.

thats like 13.7% advantage for xsx for minimum res

 

[BRiT]

I think they just use ps4pro settings in performance mode on ps5, definitly over agressive as in high resolution mode is almost same as on xsx

I'm not sure why they opted to fall back to the CB 4Pro path when it should be obvious they have enough resources to use DRS with higher image quality than that.

 

[Proelite]

I think they just use ps4pro settings in performance mode on ps5, definitly over agressive as in high resolution mode is almost same as on xsx

thats like 13.7% advantage for xsx for minimum res

Interestingly it's a 62% advantage in 60 fps mode for minimum res. I would have assumed it to be the same.

 

[snc]

Interestingly it's a 62% advantage in 60 fps mode for minimum res. I would have assumed it to be the same.

Alex asked about it

as I thought, they just used ps4pro settings for 60fps mode not that was the hardware limitation/performance reason comparing to xsx (imho bad decision as from my experience dynamic resolution looks better than cb)

 

[Proelite]

Alex asked about it

as I thought, they just used ps4pro settings for 60fps mode not that was the hardware limitation/performance reason comparing to xsx (imho bad decision as from my experience dynamic resolution looks better than cb)

Doesn't explain the minimum res. Should be 1442 x 1622, 1662p checkerboard, to be 11.5% below the X's res ala the same gap as in the 4k quality mode minimum res. Right now it's ~38% below the X.

I guess something like a 50-100% power gap won't be "dramatic" in developer's eyes. That describes Switch to Xbox One gap.

I also think they should have gone with dynamic resolution instead of the checkerboard.

 

[iroboto]

as I thought, they just used ps4pro settings for 60fps

Doesn't explain the minimum res.

These quotes combined make sense

There isn't much to glean here. Yes they used ps4pro settings for 60fps. But it's not explained why they had to. They did have an alternative render path that they had with quality mode. So it's hard to say. I would say, it's a good time to look at CPU usage there as that is a possible avenue to explore for PS5, heavy CPU game @ 60fps taxing away GPU power, but at 30fps, a non-issue.

 

[snc]

Doesn't explain the minimum res. Should be 1442 x 1622, 1662p checkerboard, to be 11.5% below the X's res ala the same gap as in the 4k quality mode minimum res. Right now it's ~38% below the X.

I guess something like a 50-100% power gap won't be "dramatic" in developer's eyes. That describes Switch to Xbox One gap.

I also think they should have gone with dynamic resolution instead of the checkerboard.

question how this ps4pro dynamic cb settings works, maybe it has some discrete options of resolution and it's not as continues as dynamic res

 

[Proelite]

question how this ps4pro dynamic cb settings works, maybe it has some discrete options of resolution and it's not as continues as dynamic res

PS4 Pro had a continuous DRS + checkerboard.

Expect both machines to scale between 50 to 60 per cent of 4K all the way up to full resolution

https://www.eurogamer.net/amp/digitalfoundry-2020-marvels-avengers-beta-test-analysis

Also VG tech mentions resolutions like 1260p, 1320p, 1440p, 1620p across their videos regarding the Pro.

So the steps seem fairly granular. Seems to be 60p per step.

If the PS5 is doing 1440p CB in Avengers it means it's not capable of doing 1500p CB.

 

[snc]

PS4 Pro had a continuous DRS + checkerboard.



https://www.eurogamer.net/amp/digitalfoundry-2020-marvels-avengers-beta-test-analysis

Also VG tech mentions resolutions like 1260p, 1320p, 1440p, 1620p across their videos regarding the Pro.

So the steps seem fairly granular. Seems to be 60p per step.

If the PS5 is doing 1440p CB in Avengers it means it's not capable of doing 1500p CB.

So why can go head to head with xsx when using same dynamic resolution in 4k ?:d You statement ps5 is not capable of doing 1500p cb is just your narrative

 

[BRiT]

So why can go head to head with xsx when using same dynamic resolution in 4k ?

Maybe a difference in CPU overhead between the different modes? *shrug* It is a mystery to me.

 

[snc]

Maybe a difference in CPU overhead between the different modes? *shrug* It is a mystery to me.

we know now from Alex that they don't use cb because of power difference but because they just chose ps4pro settings for performance mode, but sure, we can have in this one title in one mode mysterious lack of cpu power on ps5

 

[BRiT]

we know now from Alex that they don't use cb because of power difference but because they just chose ps4pro settings fro performance mode

But WHY did they do that?????? It makes no sense.

 

[iroboto]

we know now from Alex that they don't use cb because of power difference but because they just chose ps4pro settings for performance mode, but sure, we can have in this one title in one mode mysterious lack of cpu power on ps5

Avengers and Hitman 3 are both considered heavy CPU based console games in which we see larger resolution drops at 60fps from it's 30fps counterpart.

Alex mentioned that chose CB because it best suited the task for them. Alex didn't say they chose it because they weren't paid enough.

2 points is not a trend, I'm just being frank. But it's worth investigating to see if there are more points like this in the future. Maybe they can measure power draw for these titles, I'm not sure. We have no way of knowing, but because the data points are absent doesn't mean the problem cannot possibly exist. PS5 is power draw limited, that's how they designed the console to have the same variable clocks across all PS5s.

XSX is not. They had to choose a lower fixed clock rate but be able to supply an enormous amount of power in case power draw rises.

Inefficient compared to PS5, yes, but CPU and GPU do not share.

 

[snc]

But WHY did they do that?????? It makes no sense.

I will not write it lazy but..... ;d its wast just simple for them to base on what they have on ps4pro (in the end its just free update)

 

[snc]

Avengers and Hitman 3 are both considered heavy CPU based console games.
Alex mentioned that chose CB because it best suited the task for them. Alex didn't say they chose it because they weren't paid enough.

haha sure, that why in hitman 3 its rock steady 60fps in mendoza and drops to 40 on xsx ? /s common, I think control is more cpu heavy and fps are more stable on ps5

 

[iroboto]

haha sure, that why in hitman 3 its rock steady 60fps in mendoza and drops to 40 on xsx ? /s common, I think control is more cpu heavy and fps are more stable on ps5

its running significantly lower resolution though relative to itself.

XSX has it's own issues, but clearly CPU loads will not impact GPU ones on XSX.
That's can't be said as being true for PS5.

PS5 operates nothing like the fixed clock setup like XSX.
Feel free to debate anything you want around the data; but at the end of the day they don't the same challenges for optimization.

Don't point at XSX saying this or that to make a point about how games should run on PS5. XSX got it's challenges and we need to compare XSX to XSX to find them.

PS5 challenges are different. You need to compare PS5 to PS5.

 

[snc]

its running significantly lower resolution though relative to itself.

XSX has it's own issues, but clearly CPU loads will not impact GPU ones on XSX.
That's can't be said as being true for PS5.

PS5 operates nothing like the fixed clock setup like XSX.
Feel free to debate anything you want around the data; but at the end of the day they don't the same challenges for optimization.

Don't point at XSX saying this or that to make a point about how games should run on PS5. XSX got it's challenges and we need to compare XSX to XSX to find them.

PS5 challenges are different. You need to compare PS5 to PS5.

sure its forum to interpret and create theories but for me it looks simple as used in performance mode settings accordingly from ps4pro/xsx and not that really need to lower pixel count by 70% to match results (especialy after Alex comment and seeing similar results in 4k mode)

 

[Proelite]

You statement ps5 is not capable of doing 1500p cb is just your narrative

I mean if it's doing 1440p CB in a scene, then it's not doing 1500p CB in that same scene. Likewise if the X is doing 1296p in a scene, it's incapable of 1350p.



The developers would need to do more optimization, change graphical settings, and/or lower the fps target.

 

[iroboto]

sure its forum to interpret and create theories but for me it looks simple as used in performance mode settings accordingly from ps4pro/xsx and not that really need to lower pixel count by 70% to match results (especialy after Alex comment and seeing similar results in 4k mode)

yea that's fine. I'm not saying it can't just be a labour problem. Just looking back at heavier CPU titles here:

The trend for me looking back.
a) AC Valhalla 60fps, 1440p
b) Control 1440p at 60fps, and the dips are GPU related (hmm)
c) Hitman 1800p ?
- does avengers belong here-?

CPU Lite Titles
Dirt 5 120fps 1440p
4Kish 60fps for CODs (1440p-1200p @ 120fps)
etc

I'm just wondering if this is a particular pattern that will develop with more titles. Not to say that these can't be GPU limited, but GPU limited on PS5 means a variety of things, because the cap on GPU changes depending on the load put on the CPU. It's hard (impossible) for us to separate if the CPU is drawing power back (or not smart shifting power over to GPU).

The only way we'd even get a clue, is if the delta between 30 and 60fps mode is significantly larger than the delta of 30 and 60fps mode for XSX; as theoretically, increasing frame rate while decreasing resolution is in favour of PS5 architecture dramatically: CU scaling (you need less) and clock speed differential (more fill rate)