Digital Foundry Article Technical Discussion [2024]

[homerdog]

lol that 7800X3D result there is funny. As RANDOMLY, for NO REASON, I have recorded similar behaviour across processors in that game. It is random, has nothing to do with what is happening on screen. DD2 sucks IMO still.

I can see this with the naked eye in Dragon's Dogma 2 even with framegen on. Sometimes the game will start stuttering really hard for no apparent reason and then stop for no apparent reason. It's not reproducible and I have no idea what causes is. At least it's pretty rare for that to happen.

 

[raytracingfan]

I just looked at LG Displays financials for 2023, and if I'm reading it right they spent $1 billion USD on R&D in 2023.

Unfortunately, LG Display's latest quarterly report lumps revenue from panels for monitors, laptops, and tablets together at 40.5% of the company's total revenue, so it isn't possible to get an exact number for monitors alone. But I'll bet that panels for monitors make up no more than around half of that 40.5%, and the vast majority of those monitors are just regular office monitors. Gaming monitors are a niche within the overall monitor market, and premium gaming monitors with the highest possible Hz and lowest possible pixel response times and blur would be a niche within that niche. One would expect that the share of the R&D budget that goes towards panels for premium gaming monitors would be proportional to their small share of revenue.

 

[davis.anthony]

@Dictator you had one job for that 9800X3D review video and you failed.

Shame on you sir.

 

[pjbliverpool]

The game on PS5 already performs close to a 3070 Ti. With the Pro’s better GPU, we should be getting into 4070 Ti territory indeed.

It's probably not quite that high actually. If the Pro performance mode is running at PS5 performance mode settings (as there is apparently a separate slower PS5 Pro quality mode with the same settings at PS5 Quality mode) then I guess it's around the PC Medium preset. Also those Techspot benchmarks were in one of the most GPU taxing areas of the game they could find so it's likely the recorded 85-100 fps range in the DF PS5 Pro video would be lower in Tech Spots scene.

The 4070Ti comes in at 109 average and 98 low at 1440p TAA Medium and there were definitely lows into the low 80's in Johns PS5 Pro video so perhaps 4070-3080 performance bracket might be a bit closer to the mark. Still much higher than the 3070Ti level we should normally expect though. And lets not even talk about the atrocious AMD performance in this game (Pro is probably performing more like a 7900GRE than a 6800).

 

[Gamer09]

The PS5 Pro value dilemma - is a release from FOMO all you're getting here?

I don't know who needs to hear this, but I'll let you into a secret: it's okay to spend your money on whatever you want…

www.eurogamer.net

 

[Below2D]

It's probably not quite that high actually. If the Pro performance mode is running at PS5 performance mode settings (as there is apparently a separate slower PS5 Pro quality mode with the same settings at PS5 Quality mode) then I guess it's around the PC Medium preset. Also those Techspot benchmarks were in one of the most GPU taxing areas of the game they could find so it's likely the recorded 85-100 fps range in the DF PS5 Pro video would be lower in Tech Spots scene.

The 4070Ti comes in at 109 average and 98 low at 1440p TAA Medium and there were definitely lows into the low 80's in Johns PS5 Pro video so perhaps 4070-3080 performance bracket might be a bit closer to the mark. Still much higher than the 3070Ti level we should normally expect though. And lets not even talk about the atrocious AMD performance in this game (Pro is probably performing more like a 7900GRE than a 6800).

NxGamer tested with a 5600X+6800 and found the PS5 performs 22% better. In this game where AMD massively underperforms, that puts it in line with a 3070 Ti.

Of course, he's not exactly known to be objective, but that's the only data point I got where someone compares both games side-to-side with the same settings, in the same areas, and doing the same run.

The area he and HU tested is Svartalfheim. NxGamer averaged 67fps in his run, and HU averaged 61fps, so they were really close. Performance Mode on PS5 also has a few settings that are higher than the medium preset.

As you can see here, the 3070 is 18% faster than the 6800. That puts the PS5's performance slightly behind the 3070 Ti or on par with the 7800 XT where AMD underperforms by as much as 25% compared to equivalent NVIDIA cards.

I think @Dictator wanted to test the game against PS5, but was waiting for patches, and then the DAV came along with the Pro a bit later, so I doubt there's any time left for him to do side-by-sides.

 

[Dampf]

It's interesting that nearly all PS5 ports for PC run noticeably worse on equivalent GPUs and nearly equivalent settings. I think it shows Sony games were tailored specifically for PS5's UMA architecture and fast hardware accelerated decompression.

 

[pjbliverpool]

It's interesting that nearly all PS5 ports for PC run noticeably worse on equivalent GPUs and nearly equivalent settings. I think it shows Sony games were tailored specifically for PS5's UMA architecture and fast hardware accelerated decompression.

I dont see why the hardware decompression would factor in at all on GPU comparisons.

The games being developed from the ground up for the architecture and fast paths of the PS5 GPU and API and then being ported with likely no or minimal re-architecting for DirectX and the PC architecture is almost certainly the culprit. And yes UMA would play a part there.

 

[techuse]

It’s interesting to wonder whether the PC ports of Sony titles are unoptimized relative to the field or if Sony titles on PS5 are super optimized. I lean towards the latter but we have yet to see a PS5 title built with current gen in mind.

 

[Below2D]

It’s interesting to wonder whether the PC ports of Sony titles are unoptimized relative to the field or if Sony titles on PS5 are super optimized. I lean towards the latter but we have yet to see a PS5 title built with current gen in mind.

Spider-Man 2 and Rift Apart weren't built with the PS5 in mind?

I'd also say it's both optimized for PlayStation and not that well optimized for PC. This was ported by a tiny team.

 

[Remij]

Obviously the first party games for PS5 are extremely well optimized for the hardware. They have entire teams of people who's goals are to make the game as efficient as possible for the specific hardware. PC is not ever going to be as efficient. There may be some games which are able to take advantage of certain capabilities of the PC architecture more easily than others.. while others are impacted by it.

Let's also keep in mind that these ports are being handled by much smaller crews of people, with much less time and budgets.. they often have to make concessions that they wouldn't if PC was the lead platform from the beginning.

 

[techuse]

Spider-Man 2 and Rift Apart weren't built with the PS5 in mind?

I'd also say it's both optimized for PlayStation and not that well optimized for PC. This was ported by a tiny team.

The realities of development didn't allow for either to have tech pushed much beyond PS4 games. Well made games but I expect big improvements over both by generations end.

 

[Cappuccino]

I dont see why the hardware decompression would factor in at all on GPU comparisons.

The games being developed from the ground up for the architecture and fast paths of the PS5 GPU and API and then being ported with likely no or minimal re-architecting for DirectX and the PC architecture is almost certainly the culprit. And yes UMA would play a part there.

My thing with this is PS5 is just running a PC architecture, and yet when ported to PC these games not only run worse overall but run worse specifically on RDNA cards. You’d think if it was hyper-optimized for an RDNA console then AMD cards would get a free bump over Nvidia, but this doesn’t happen (except for COD, the actual lone exception where AMD outperforms Nvidia).

 

[iroboto]

My thing with this is PS5 is just running a PC architecture, and yet when ported to PC these games not only run worse overall but run worse specifically on RDNA cards. You’d think if it was hyper-optimized for an RDNA console then AMD cards would get a free bump over Nvidia, but this doesn’t happen (except for COD, the actual lone exception where AMD outperforms Nvidia).

I would say it's absolutely optimized for console, otherwise these devices wouldn't be getting the output that they do compared to the PC equivalent parts. They can fully take advantage of hardware, code paths, low level access and design paradigms that are super ideal for processing but perhaps inefficient on PC due to the difference in architecture and goals.

 

[Cappuccino]

I would say it's absolutely optimized for console, otherwise these devices wouldn't be getting the output that they do compared to the PC equivalent parts. They can fully take advantage of hardware, code paths, low level access and design paradigms that are super ideal for processing but perhaps inefficient on PC due to the difference in architecture and goals.

Right but what difference in architecture?? This isn’t Cell, it’s two desktop level architectures (Zen and RDNA), those optimizations should carry over to PC, at least somewhat (and certainly relative to Nvidia!)

 

[techuse]

Right but what difference in architecture?? This isn’t Cell, it’s two desktop level architectures (Zen and RDNA), those optimizations should carry over to PC, at least somewhat (and certainly relative to Nvidia!)

The API on PS5 is completely different and seemingly far superior to DX. PS5 exclusives will take full advantage of all the fast paths and hardware functionality that isnt exposed on PC. Multiplatform titles are unlikely to go this route as it complicates development for the other versions of their game. In those scenarios, optimizations are more likely to carry over. In the case of Sony ports, minimal budgeting seems to be allocated so efforts are probably focused on getting the game to function properly and whatever is left they dedicate to Nvidia for marketshare reasons.

Developers have been asking for better tools on PC for years. Unfortunately, neither AMD, Nvidia, Intel or Microsoft seem to care. Understandably so since it would be more likely to hurt their profits rather than help.

 

[raytracingfan]

Right but what difference in architecture??

As others have stated, consoles have UMA - they share the same memory pool. PCs have separate system memory and VRAM. That makes a difference for titles that were designed for a specific console first, wringing out every drop of performance from that platform, and are ported to PC later.

 

[Cappuccino]

The API on PS5 is completely different and seemingly far superior to DX. PS5 exclusives will take full advantage of all the fast paths and hardware functionality that isnt exposed on PC. Multiplatform titles are unlikely to go this route as it complicates development for the other versions of their game. In those scenarios, optimizations are more likely to carry over. In the case of Sony ports, minimal budgeting seems to be allocated so efforts are probably focused on getting the game to function properly and whatever is left they dedicate to Nvidia for marketshare reasons.

Developers have been asking for better tools on PC for years. Unfortunately, neither AMD, Nvidia, Intel or Microsoft seem to care. Understandably so since it would be more likely to hurt their profits rather than help.

Yeah this all makes sense to me, more of a software/hardware interface issue than a hardware architecture situation. Explains how PS5 does well vs Xbox despite the hardware differential.

As others have stated, consoles have UMA - they share the same memory pool. PCs have separate system memory and VRAM. That makes a difference for titles that were designed for a specific console first, wringing out every drop of performance from that platform, and are ported to PC later.

Wouldn’t unified memory be a performance drag? I can’t think of many scenarios where two unshared pools of memory are worse than a single pool, unless the former are comically slower than the latter.

 

[techuse]

Yeah this all makes sense to me, more of a software/hardware interface issue than a hardware architecture situation. Explains how PS5 does well vs Xbox despite the hardware differential.


Wouldn’t unified memory be a performance drag? I can’t think of many scenarios where two unshared pools of memory are worse than a single pool, unless the former are comically slower than the latter.

There was an interview, done by DF I believe, where a developer stated that one of the code paths used in a Sony title couldn’t be replicated on PC due to DX limitations. The workaround was a lot slower.

Unified memory has the benefit of faster data sharing. This can allow for a lot of efficient algorithms.

 

[pjbliverpool]

Wouldn’t unified memory be a performance drag? I can’t think of many scenarios where two unshared pools of memory are worse than a single pool, unless the former are comically slower than the latter.

There was an interview, done by DF I believe, where a developer stated that one of the code paths used in a Sony title couldn’t be replicated on PC due to DX limitations. The workaround was a lot slower.

Unified memory has the benefit of faster data sharing. This can allow for a lot of efficient algorithms.

I would image the unified memory usage is as much (or more) an ease of development thing than a performance thing. Absolutely having unified memory is easier from a developer standpoint and so devs designing a game as a console exclusive would make full use of it.

Directly porting that over to PC is then going to have performance issues if communication over PCIe is excessive. It may be that the game could be radically re-designed to avoid most of that PCie traffic at the cost of additional developer headaches, while also achieving equal or better performance on the PC side to PC equivalent hardware in the consoles. But the development/porting effort involved with that would likely be far to high for Sony's taste. Afterall they have a vested interest in their ports overperforming on their own hardware vs PC so why spend excessive amounts to bring the PC side performance up?