Digital Foundry Article Technical Discussion [2021]

I started watching this on my PC, had to go outside to get my kids off the bus so I picked it up on my phone. On a large screen, it looks fine. Like a good looking game. On my phone, wow. The small screen and better contrast (phone is OLED, PC is LCD) this video looks pretty close to reality.

 

Why would PS5 be above 200w?

 

because the PS5 always tries to run as fast as possible if the power budget allows it. The peak power budget is at around 220W.

 

The power consumption on PS5 actually varies a lot depending of the game. From the various benchmarks around it's about 100-150W in BC games, usually 175-195W in the vast majority of PS5 native games (gameplay). 200W reached in many cutscenes and I think >220W only ever reached in Control photo-mode.

 

this game has 60 fps cape

 

Except it doesn't, how fast it runs is governed by processor load and a games frame rate cap just like every other CPU/GPU.

 

Sure. But I would still guess that at least one of the systems is maxed out (more or less).
Yes it has a 60fps cap and that is why a power consumption test would make sense. Because you have a game that is behaving more or less the same on both systems.
A power consumption test could at least indicate if the system has headroom or not.

 

doubt it, to have 60 fps with so little drops system is probably capable to run above 70fps on average (for example ratchet 40fps mode on 120hz screen without downgrade comparing to 30fps)

 

Wouldn't really work on PS5 because of boost clocks and how they scale depending on load, temperature and power.

Games with an unlocked frame rate either during gameplay or photo mode are realistically the only way to reliably test PS5's 'head room'

Even though it will only test head room in that one specific scene, in that one specific game, running that specific code.

 

it's still an upgraded last gen game, they wanted to hit that perfect 60fps and were fine with it, they are not a big dev so certainly no time to optimize more for each specific system.

 

Not dGPU's as far as im aware of. In modern laptops the CPU/GPU can load balance, but even that isnt exactly like the PS5's since Sony modified it for their games system.

 

In normal operation the PS5's boost clocks operate without regard for temperature, as it's boost clocks are based on some kind of uniform set of rules that are the same for all PS5's.

This is good because it allows for consistent experiences across units and environments. It's the same as the XSX (and all other consoles afaik) in this sense.

 

Any GPU that has boost balances based on temps, power and frequency.

And while they don't work the same it was my way of saying that with so many variables now governing GPU's it's difficult to compare power use by just laying a game as so much is in play.

We still don't know what PS5's lowest clock is the for GPU or how many clock straps the GPU even has.

 

In day to day no but there will be situations (however rare) where the temp would affect clock rate.

 

Yes but, for example a 2080Ti delivers its said 13.5TF's of compute (and according clocks then), this is the base clock, it clocks upwards from there if temps, load etc allow for that. It doesnt either share any load between the CPU. That is considering you got matching PSU, airflow, and other such factors ofcourse.
However it isnt the same system as Sony has for its PS5 since that's a reverse engineered smartshift (laptop) solution.

How much the PS5/gpu draws in like for like games/settings vs xbox, thats another DF video i hope

We probably will never know. The closest thing to compare it to seems ryzen equipped laptops where gpu/cpu share power loads, but even there it seems quite different.

Maybe when thermal SoC limits are reached, but then the thing would shutdown i imagine.

 

As I said, in the day to day gaming temperature will not be a problem, but there will be situations (however rare) where temperature could cause a drop.

 

I have always thought that Control in RT+ Photomode is causing downclocking on PS5, I can't even test mine as my watt meter broke...(God damn Fermi )

 

Who knows, could be engine related aswell. And even if we can measure power draw at the wall, its still unknown whatever the clocks are doing at that point. Seeing general PS5 performance right now, it doesnt seem to downclock all that much or at all in multiplatform games (seeing how close it performs to xsx). Maybe it does in more power hungry games, but these probably are the likes of forbidden west, rift apart etc which we cant really compare since their only on the PS5. Maybe when the PC versions appear more can be tested, but thats awhile left i think.