Digital Foundry Article Technical Discussion Archive [2014]

Status
Not open for further replies.
All I know is that if the difference is like going from my old MacBook to my new Pro with SSD, which is literally like night and day - I love using Literally completely wrong like that - then there should be quite a difference on PS4 too. Literally.
 
You can't compare scenes that are completely different. Digital foundry even said the performance was slightly better on PS4. Because when you compare identical scenes like the cutscene from the beginning...
Frames during cutscenes are not really relevant for gameplay. Also a 1 frame difference may be some inaccuracy.
Dips while driving would interfere with the gameplay. Yes it is hard to equaly reproduce such situations, but these are those that count. Just wait for the final analysis from DF.
 
All I know is that if the difference is like going from my old MacBook to my new Pro with SSD, which is literally like night and day - I love using Literally completely wrong like that - then there should be quite a difference on PS4 too. Literally.

I remember inadvertantly watching an interview with some kid from One Direction and he said that he was "literally gutted" about something. I laughed solidly for a good thirty seconds, until I wished he actually had been.

I'd like to hear some more views on this game from those that have actually bought it. I have it on 360, but I'm not sure whether it's worth the upgrade. I'm more tempted with Little Big Planet for the moment.
 
Article says it has judder

The article says "gameplay is mostly locked at 30fps with just minor fluctuations beneath its target. Curiously, despite a consistent update, judder can manifest a little during gameplay"

You made it sound worse than this hence my asking.
 
GTA 5's next-gen grass compared

However, the more locations we visited, the more it became clear that PlayStation 4 does possess the advantage in terms of detail level - it's just that the extent of the boost varies on a per-location level. In our first set of clips, we took the road north and explored the mountainside, finding no appreciable difference. However, spawning a chopper and flying onto the Vinewood hills, we discover plenty of areas where the results of the screenshot comparisons are borne out.

Moving on again into the Paleto Forest - an area where we would expect to see a rich array of greenery, our sample clips show that the compromise is limited to plants and grass. More imposing environmental detail, like tree density for example, seems to be untouched on Xbox One. The PS4 advantage here is much more subtle - a few bonus plants and small grassy areas. Overall, this is the kind of effect where the PS4's larger GPU and higher fill-rate come to the fore, and while it shouldn't impact any purchasing decision (Xbox One is hardly barren and featureless in the affected locations), it's nice to see that Rockstar has utilised the additional graphics power on tap in the Sony console.
 
http://assassinscreed.ubi.com/en-US....aspx?c=tcm:152-185420-16&ct=tcm:148-76770-32

It's interesting that the crowd size in AC Unity is not what is causing the framerate problems. Many of us speculated that they could have just reduced the size of the crowds to get the framerate under control.

Thanks for that link. The speculation about NPCs is also kind of Ubisoft's fault

"Technically we're CPU-bound," he said. "The GPUs are really powerful, obviously the graphics look pretty good, but it's the CPU [that] has to process the AI, the number of NPCs we have on screen, all these systems running in parallel.

anyhoo from the new post from Ubisoft.

We can tell you that we have detected a distinct discrepancy between what we observed in the pre-launch versus post-launch environment. In spite of our testing, it looks like the instruction queue is becoming overloaded and impacting performance

Does the instruction queue refer to draw calls or some other queue. Lighting assets are taking up supposedly nearly half of the BD so the management of such a basic thing applied to every frame is obviously a challenge. Whether or not the CPU is getting a bit into the weeds telling the GPU what to do with the lighting or there is memory bus contention a bit of both ( or something else ) is TBD. Hopefully Ubisoft will take this opportunity to bolster the GPU "compute" facilities on the next iteration of their gaming engine. Maybe we will bet at least a slideshow on the subject when things settle down :)

Despite the framerate issues I just love looking at this game ( online, I don't have it so I can't rightly complain about gameplay ) and I sure hope this means companies will push forward in terms of lighting as opposed to backing off. All games of this type should look as good, it's what we deserve after waiting far too long for a new console :p
 
How does something change from pre-release to post-release on a closed system?
 
Even though DF said offline mode didn't impact the frame rate Ubisoft seems to suggest that the "environment" of the retail game isn't the same as the in-house one leading one to suspect something about being online somehow throwing a spanner in the works or it's just a delaying tactic until a new post is offered up.
 
We also don't know what frame rates Ubisoft was seeing in-house so while there maybe a discrepancy it's magnitude may be another question.
 
I've had the odd stutter, often at big intersections where there are 40+ cars, but have observed very little pop-in even when travelling at speed. I'm using a 7200rpm drive and I wonder if this is making a difference. I'd be interested to hear from anybody with a pure (not hybrid) SSD.

I'm running a 500 GB 840 Evo pre-firmware update. Give me a particular scenario you want to test and I can try to upload that to YouTube. We can compare that to a video of you doing the same thing as see if there's a difference.
 
How does something change from pre-release to post-release on a closed system?

Perhaps the dev systems are not exactly identical to retail units. Perhaps the dev systems don't have to content with an OS doing things like muddying instruction and data cache one of the four-core clusters. They're said so little in such a vague way it's hard to tell what they're talking about! o_O
 
I'm running a 500 GB 840 Evo pre-firmware update. Give me a particular scenario you want to test and I can try to upload that to YouTube. We can compare that to a video of you doing the same thing as see if there's a difference.

I'm struggling to think of something that reproducible. I'm also a little stuck on things to test because, as I said, I'm not getting the pop-in others are.

Are you? My thumb is absolutely knackered, I just did the third triathlon - and won! 30 minutes of hitting X. Apologies in advance for the language, but Jesus fucking Christ, Rockstar! :mad:
 
http://assassinscreed.ubi.com/en-US....aspx?c=tcm:152-185420-16&ct=tcm:148-76770-32

It's interesting that the crowd size in AC Unity is not what is causing the framerate problems. Many of us speculated that they could have just reduced the size of the crowds to get the framerate under control.

Depends on which frame rate issues they were talking about: specific instances of the game plummeting to < 20 fps i.e. online mode, or the game just generally chugging along all the time.

Fixing the worst dips would seem to be the priority and the kind of thing that might have caught them out in testing.
 
I'm struggling to think of something that reproducible. I'm also a little stuck on things to test because, as I said, I'm not getting the pop-in others are.

Are you?

It's been pretty smooth for me so far, but I'm only a few hours in. A nice change of pace from the digital PS3 version I originally played. I think I'll grab the fastest car I can and try going full throttle through downtown and see how things perform.

My thumb is absolutely knackered, I just did the third triathlon - and won! 30 minutes of hitting X. Apologies in advance for the language, but Jesus fucking Christ, Rockstar! :mad:

I remember that 3rd triathlon being brutal, but I also remember feeling pretty good when I finally completed it. Reminds me of the microwave room in MGS (MGS2, I think it was). Not exactly looking forward to doing that all over again on the PS4 version but I'm sure I will anyway.
 
Perhaps the dev systems don't have to content with an OS doing things like muddying instruction and data cache one of the four-core clusters.

If something like this was occurring maybe the more segmented nature of the XB1 virtual machine setup might be helpful under such circumstances.
 
Status
Not open for further replies.
Back
Top