Digital Foundry Article Technical Discussion [2021]

Status
Not open for further replies.
Not sure what you imply here but simple looking at comparable hardware desktop counterpart Oddworld is one of worst implementation of ps5 version (second dynamic cb in performance mode Avengers). Whats the reason I don't know (imho small not most capable team but maybe some ps5 limitation).

PS5 should be able to do better in that game, on the other hand its kinda hard to compare to a full fledged (comparable) pc CPU aswell, which isnt restricted by thermals, power delivery, clocks, cache etc etc.
 
DF Crysis update @ https://www.eurogamer.net/articles/...ysis-remastered-next-gen-console-patch-tested

Update 11/4/21 3:00pm: We've spent the last few days testing the Crysis Remastered upgrade for PlayStation 5 and can confirm that overall, we're looking at the same results for the Sony platform as we are for Xbox Series X, with a couple of changes brought about by the fact that PS5 retains the resolution settings for each mode as found on PlayStation 4 Pro - a common situation on so-called 'back-compat plus' titles.

We can confirm that the resolution and performance targets mentioned in our original piece hold true: the performance and ray tracing modes target 60fps at 1080p resolution, while the quality mode targets a maximum 1800p instead with the same 60fps aspirations. However, dynamic resolution scaling is in full effect on this title and results in actual play are quite variable. As you'll see in the embedded video below, it can make ascertaining differences between the various modes somewhat challenging. All told, for those interested in Series X comparisons, PlayStation 5 generally runs a touch smoother than Series X, likely by virtue of its lower resolution targets. This is especially evident in the ray tracing mode. Meanwhile, Series X has a resolution advantage.

However, delivering an actually locked 60 frames per second experience seems to be beyond both versions and the reasons behind this are puzzling. In the embedded below, we share some theories about this - I recommend checking it out.

 

I was gonna give this a pass until watching this - Linneman is very enthusiastic.
I've also found that DF does a much better job at marketing than most other forms of marketing lol; at least for us. People come in for the performance metrics, they end up being lulled by gameplay, artistic direction and design.
 
this game runs like crap and doesn't even look very good by today's standards, a mess.
with VRR/Freesync, there is a bit of save here for the few fortunate enough for this setup. But without, I would agree; it's a bad remake. No new features, nothing really. They had a chance to do something like Demons Souls remake (given the age). But went the path of a remaster.
 
DF Crysis update @ https://www.eurogamer.net/articles/...ysis-remastered-next-gen-console-patch-tested

Update 11/4/21 3:00pm: We've spent the last few days testing the Crysis Remastered upgrade for PlayStation 5 and can confirm that overall, we're looking at the same results for the Sony platform as we are for Xbox Series X, with a couple of changes brought about by the fact that PS5 retains the resolution settings for each mode as found on PlayStation 4 Pro - a common situation on so-called 'back-compat plus' titles.

We can confirm that the resolution and performance targets mentioned in our original piece hold true: the performance and ray tracing modes target 60fps at 1080p resolution, while the quality mode targets a maximum 1800p instead with the same 60fps aspirations. However, dynamic resolution scaling is in full effect on this title and results in actual play are quite variable. As you'll see in the embedded video below, it can make ascertaining differences between the various modes somewhat challenging. All told, for those interested in Series X comparisons, PlayStation 5 generally runs a touch smoother than Series X, likely by virtue of its lower resolution targets. This is especially evident in the ray tracing mode. Meanwhile, Series X has a resolution advantage.

However, delivering an actually locked 60 frames per second experience seems to be beyond both versions and the reasons behind this are puzzling. In the embedded below, we share some theories about this - I recommend checking it out.


Finally a face-off where all versions lost.
 
I've never played Crysis, but I'd love to. However, knowing that things like granular destruction of flora are pared down, I'm just going to wait until a better made remaster/remake comes along. Or until I have a "can if play Crisis" PC.
 
I've never played Crysis, but I'd love to. However, knowing that things like granular destruction of flora are pared down, I'm just going to wait until a better made remaster/remake comes along. Or until I have a "can if play Crisis" PC.

I played the original on PS3 and really liked it for the same reason I liked the original Far Cry. A nice big open world sandbox, a nonsense pulp story and lots of enemies to obliterate in whatever creative way you can imagine. :yes:
 
However, delivering an actually locked 60 frames per second experience seems to be beyond both versions and the reasons behind this are puzzling. In the embedded below, we share some theories about this - I recommend checking it out.

Haven't watched the video yet, but I'm guessing one of the reasons is that Crysis isn't taking advantage of any of the systems RDNA 2 IPC improvements, seeing as both are still straddled to their predecessors' code and settings. "BC" has its limitations, especially on PS5 where Sony seems inept at improving (at the moment anyhow).
 
Haven't watched the video yet, but I'm guessing one of the reasons is that Crysis isn't taking advantage of any of the systems RDNA 2 IPC improvements, seeing as both are still straddled to their predecessors' code and settings. "BC" has its limitations, especially on PS5 where Sony seems inept at improving (at the moment anyhow).
If there are specific PS5 and XSX binaries, they would take advantage of IPC differences between RDNA and GCN. ie: they should be submitting (should be allowed to submit rather) instructions every cycle vs every 4 cycles and should take full advantage of bandwidth and the available ALU; though that doesn't mean they can't be stuck with an older rendering pipeline however (or still using older shaders that are optimized for GCN cycle submission for instance as an avenue for investigation)
 
Last edited:
DF Crysis update @ https://www.eurogamer.net/articles/...ysis-remastered-next-gen-console-patch-tested

Update 11/4/21 3:00pm: We've spent the last few days testing the Crysis Remastered upgrade for PlayStation 5 and can confirm that overall, we're looking at the same results for the Sony platform as we are for Xbox Series X, with a couple of changes brought about by the fact that PS5 retains the resolution settings for each mode as found on PlayStation 4 Pro - a common situation on so-called 'back-compat plus' titles.

We can confirm that the resolution and performance targets mentioned in our original piece hold true: the performance and ray tracing modes target 60fps at 1080p resolution, while the quality mode targets a maximum 1800p instead with the same 60fps aspirations. However, dynamic resolution scaling is in full effect on this title and results in actual play are quite variable. As you'll see in the embedded video below, it can make ascertaining differences between the various modes somewhat challenging. All told, for those interested in Series X comparisons, PlayStation 5 generally runs a touch smoother than Series X, likely by virtue of its lower resolution targets. This is especially evident in the ray tracing mode. Meanwhile, Series X has a resolution advantage.

However, delivering an actually locked 60 frames per second experience seems to be beyond both versions and the reasons behind this are puzzling. In the embedded below, we share some theories about this - I recommend checking it out.

Well, stable 60 would really be a surprise. It does not even run that stable on PC at 60. It is still a game that is very single-thread CPU limited. The only thing I can imagine is, that the Playstation has the advantage, that the game engine had to be "rewritten" to work with the Playstation API instead of DirectX. So it might get optimized while the engine was ported.
From a GPU perspective, I really don't see why a such old game shouldn't run at full 4k. The only thing why it might be the case is that it almost only uses "old" features and not newer much more efficient features. After all they didn't rewrite the whole game, they just updated the existing one.
 
this game runs like crap and doesn't even look very good by today's standards, a mess.

I was very excited to get this “remaster” on release day on pc. I noticed that my gfx was “recommended “ shocker as my 1660ti was almost never recommended for anything. I was thinking finally I will experience this classic at at least highish settings. But before I purchased I decided to have a look at reddit forum and was shocked how poorly it ran and people with much much more powerful PCs had a lot of FPS issues. To be honest I am surprised that it runs so well on all consoles while not really running well. It’s a mess always been always will be
 
Well, stable 60 would really be a surprise. It does not even run that stable on PC at 60. It is still a game that is very single-thread CPU limited.

Even for a single threaded design 13 years of cpu+gpu development should have smoothed this over. That it doesn't is quite puzzling.
 
Even for a single threaded design 13 years of cpu+gpu development should have smoothed this over. That it doesn't is quite puzzling.
To be honest, I never understood this either. If memory serves, the absolutely fastest CPUs from 2007-2008 were 3.2Ghz Core 2 duos/quads and maybe 2.6Ghz Phenoms from AMD. If single threaded clock rate is the limiting factor, intel went backwards with the i7 (launched at 2.6Ghz) I think but AMD pushed forward and eventually had Bulldozer variants at 5Ghz. Basically double the clock rate of the earlier Phenoms but I don't think they ran Crysis completely smooth either, and I think intel still beat them at a lower clock speed.
 
Even for a single threaded design 13 years of cpu+gpu development should have smoothed this over. That it doesn't is quite puzzling.
you can always find some algorythms that work well with few "objects" but scale badly when it is used for more. So just to use the physics engine of the game the work is doubled by the framerate. But at the same time resolution and visual range had been increased. Both parts can also lead to increased workload for the phyiscs engine. Because smaller parts might now be visible and can no longer get excluded and things that just weren't in sight before must now also behave correctly (like trees in the distant that move a bit in the winds). All this and the older engine (that was just not made for this kind of resolution and range) can lead to higher and higher workloads, that if single-threaded will lead to problems quite fast.
Also not to forget, that todays engines are much better in handling stuff, that just is not on the screen and therefore does not get that much attention. Tools got better which lead to better and more efficient engines. E.g. Crysis 2 and 3 were designed with consoles in mind. So they optimized the engine and levels for exactly that case, that a console could run the game. Crysis was was designed for high end PC of that time. This game was part of the reason why I bought a GTX 460 at that time. And it didn't even run really good at that card. But having arround 30fps at that time, was good enough for that time with such graphics.
 
you can always find some algorythms that work well with few "objects" but scale badly when it is used for more.
Yup, then you get cases like some physics interactions where if you are simulating a bunch of slow-moving objects or objects not interacting then wider parallel processing works well but once objects start interacting with each other then physics need to be calculated in the order of cause-and-effect, e.g. ball A hits ball B which hits ball C which hits ball D. Then you can't simulate the physics of ball A and D at the same time, because you need to introduce incorporate the physics of balls B and C.

The same can be true of AI. Sometimes it's fine to have them independent, other times you want this dude to quickly react to that dead being shot dead.
 
In the latest installment of Digital Foundry Direct Weekly, the team tackle a bunch of new patches, discuss the exciting implications of AMD's GPU chiplet patent, enthuse about the Pac-Man 99 announcement and ponder the Castlevania Resurrection Dreamcast prototype. The Patreon Supporter Q+A throws up some interesting questions this week like, how do we cope with 'spiteful criticisms and supposed debunks'...?

00:00:00 Introduction
00:02:00 Mortal Shell Patch
00:07:40 Outriders Patch
00:11:33 Cyberpunk 2077 Patch
00:19:35 AMD GPU Chiplet Patent
00:24:15 Pac-Man 99 announcement

Retro Corner
00:26:13 Castlevania Resurrection Dreamcast Prototype Discovered
00:27:47 Tom Got A PVM

DF Content Discussion
00:39:15 Ryzen 3900X to 5950X: Alex's PC Upgrade
00:51:12 PS5 SSD Load Times Project
00:55:27 Crysis Remastered Next-Gen Patch
00:57:33 Oddworld Soulstorm
01:00:19 Patreon Exclusive Videos
01:01:16 Patreon Supporter Q+A
 
Status
Not open for further replies.
Back
Top