Digital Foundry Article Technical Discussion [2021]

Unfortunately, DF ignore suggestions about measuring power consumption. And for some reason they don't even answer why.

 

Obviously in the pockets of big oil!

 

what are we trying to measure, i have a kill a watt meter and a xbox series x

 

Power consumption on ps5 and xsx in multiplatform games in the same scenes.
Or typical power consumption in multiplatform games and exclusives if you own only one of the consoles.

Forza Horizon 5 (perf and quality mod)
Gears 5
The Matrix Awakens
Metro Exodus Enhanced
Watch Dogs Legion/Control
AC Valhalla/CoD Vanguard/Cold War/Far Cry 6

Thanks in advance.

 

I remember the video DF made around the Series consoles launch and Gears 5 drawing the most power out of the games tested. Gears 5, at that point, you could argue, was the only game taking advantage of what the Series was bringing to the table. With Matrix Awakens making what appears to be full use of UE5, it would have been nice to get a glimpse into its power consumption compared to games like Gears 5, FH5, Flight Simulator, Doom Eternal...with Halo Infinite thrown in.

They could also look into the power consumption for games on the PS5, Ratchet and Clank, Demons Souls, Miles Morales, whatever the best-looking UE4 game on the PS5 is right now compared to the Matrix Awakens.

I was hoping that it would be part of the Matrix video, though. I'm not sure how interesting the content is on its own.

 

I'm curious why this is important. I mean, I get the academic side of it. But what information would we really learn by doing power consumption shootouts between the 2 consoles. Isn't what they output to the screen and speakers what really matters?

 

Also, power consumption [game] testing isn't always reliable. A single piece of unoptimized code (even in the simplest of scenes) can drive up clocks, which of course affects wattage usage/consumption.

 

Maybe I misunderstood you (English is not my native language), but I am really interested in this topic and would be interested to know why DF does not want to review it.
Is it too hard to do? Do they just not have the time? Or does it make no sense at all?

This could give an insight into how heavily loaded consoles are in games coming out now.

If, for example, one of the consoles in one of the games consumes power on average significantly less than its maximum and noticeably less than the other console, then it was obvious that the game was simply poorly optimized for this console.
And understanding this would immediately stop many senseless disputes.

 

Pencil and paper

 

Maybe. But there are scenarios where particular software is unoptimized and instead of drawing more power, it causes stalls in the processing time that actually would cause lower power draw. While alternately, things could be coded in a way that are simply less efficient, and require more computational power to produce similar output on different hardware (or even the same hardware with different APIs). In the first case, the power draw would be lower but also the performance. In the second, the power draw would be higher but the performance lower. Power draw doesn't really tell you anything without knowing what the code is doing.

 

It seems to me that ideally it is necessary to record power consumption measurements on the camera and synchronize them with the video sequence of the game.

Yes, this is not an ideal method for all cases, but it can be quite interesting on average.

 

Weirdly - given the emphasis on environment impact these days - this seems fairly niche equipment. When I worked in areas where power usage was important, we used Fluke equipment and a quick Google search shows they are still around and do provide this equipment.

 

I assume the budget power logging market isn't big enough to warrant anybody developing a cheap solution. This type of equipment is aimed at people designing and selling equipment that needs to be power efficient, rather than people with a vague interest in testing the power consumption of devices.

This is why these devices are self-contained units with inbuilt computers that log the data and present the data, for later transfer as opposed to a devices that monitor usage and output the data to another device - requiring external software to interpret it.

That's likely the issue. It's not a complex problem but there is no market for it. The smaller the market, the bigger the price to develop the solution. ¯\_(ツ)_/¯

 

I know there are APC/UPSs that track and keep historical data on how much power (per receptacle) is being used by a device. Whenever I'm rendering large film projects, I sometimes check out the UPS web-interface just to see how much power is being consumed.

 

It's also not a massively complex problem for a pass-through power device to send realtime power draw to an app. My house is already full of smart plug adaptors that allow you to automatically turn on/off non-smart devices like the lava lamp. It wouldn't be massively complicated to monitor and transit the power draw as well as the status over Bluetooth instead of Zigbee.

An interesting project for somebody.

 

00:00:00 Introduction
00:00:43 Ray traced emissive lighting details
00:04:41 SSD and streaming tests
00:08:36 PS5 and Xbox Series X resolution counts
00:11:08 PS5 vs PC high resolution press shots
00:12:36 Xbox Series S vs Series X Visuals
00:15:14 Matrix Awakens Demo vs 4K Blu-ray