Digital Foundry Article Technical Discussion Archive [2015]

Status
Not open for further replies.
Great work at DiCE to get to 900/60 but I have to wonder if that means that the 64 player Conquest game mode of BF4 is just too much for the engine? Oh well roll on BF5 if they could get that to a steady 900/60 w/64 players I'll be laughing (or even mostly steady)
 
That is impressive. Top shelf visuals without the need to resort to variable image quality to maintain performance consistency. This gives me hope that dynamic resolution won't be the go to solution to achieve 60fps on consoles.
 
The DDR3 in X1 has ~10% lower latency than the GDDR5 in ps4.

Though generally you are hitting the l1 & l2 cache which is where the 1.75 ghz higher clockspeed comes into play. The perfomrance difference between X1 & ps4 seems to be ~+10% in X1 favor. So its likely cpu related performance differences.

I tried to cross-reference presentations and the SDK leak concerning latency, and a potentially larger contribution is in the interconnect.
https://forum.beyond3d.com/posts/1816112/

AMD's APUs have a habit of worsening latency the more integrated the GPU gets in the hierarchy. Microsoft also said they did some work on the interconnect, and the on-die bandwidth is higher relative to Orbis.

It may take 50-70 cycles more to start talking to DRAM on the PS4.
The in-module cache hierarchy and DRAM access appear to be rather close in the latency contribution they have, so whatever that goes on between a miss leaving one module's cache and the access being started to DRAM seems to be showing up.
 
Wow, that's some eastman level hyperbole there. I've sunk more than 60 hours into Fallout 4 on PS4 and yes, there are some issues that effect visuals in a few areas.

Agreed. The performance seem to have little correlation to combat or on screen visuals in my ecperience. Just randoms frame dips and mostly the game is at 30fps. That being said the game looks like a 1080P port of FO3 and has a ton of glitched and bugs. I think I made four glitch videos yesterday alone. I think their engine is just a mess of crappy code ported forward over the years. They need to scrap it and start fresh, higher some good people too. Using this game to measure console performance is about the worst idea possible.

 
Or PC hardware performance really, i rarely get 100% GPU/CPU usage on PC and i get frame drops all over the place after i reached diamond city. Area around the Trinity tower is a mess for me. Stuttering all over the place. If the next TES is using the same engine i won't play it, period.
 
Agreed. The performance seem to have little correlation to combat or on screen visuals in my ecperience. Just randoms frame dips and mostly the game is at 30fps. That being said the game looks like a 1080P port of FO3 and has a ton of glitched and bugs.
What, you don't like like companions or enemies dropping out of the sky randomly?

They should add the track It's raining men! Because sometimes it literally is! Men and dogs and ghouls and robots! :LOL:
 
I have yet to see one companion that I send to the settlement ever show back up. I haven't seen my dog or the PI guy since I sent them back. I also lost the Brotherhood guy, he just vanished when I fast traveled. I need more rain...
 
X1 Battlefront perf

Looks less stable than Ps4 but still pretty good, most of the time it stays above 50.
 
Indeed...

40 does seem to be rather arbitrary - maybe they had some internal metrics as to what they could push comfortably. How does BF4 perform with fewer players?
 
FWIW, when DF played the PS4 version, it was at a DICE event where they had 30 PS4s setup so they were limited by the number of players/consoles. We'll see how it holds up with the full 40. But as Clukos said, the beta ran smoother on PS4 too. I expect similar performance on both. And regardless, it's a huge improvement over the BF games - both visually and performance wise.
 
Status
Not open for further replies.
Back
Top