Support NeoGAF

[Gamer79]

Before the PS5 pro there were some crazy claims being thrown about. So did this statement hold true or is Richard from Digital Foundry caught talking out of his ass?

PS5 Pro-Like PC Would Cost a Fair Bit More, Says Digital Foundry; RTX 4070 Is the Closest GPU

Digital Foundry's Richard Leadbetter compared the PS5 Pro GPU to NVIDIA's GeForce RTX 4070 and said a PC with those specs would cost more.

wccftech.com

 

[LectureMaster]

I like this topic, this is gonna be fun.

 

[Guesclin]

You can't compare PC and CONSOLE optimisation + look at benchmarks on YT

 

[Magic Carpet]

I've already hugged my RTX 4070. Do I need to hug it again?

 

[joeygreco1985]

It looks more like a standard PS5 with a shitty FSR upscaler

 

[HogIsland]

I don't think DF was literally saying Pro is a match for 4070. 4070 crushes Pro in multiple areas. The point is Pro's raster performance paired with Nvidia-like upscaling and Ray tracing enhancements make it most similar to a 4070 in features.

It seems clear a PC with a 4070 will beat a pro in practically every game.

 

[Vick]

I don't think DF was literally saying Pro is a match for 4070. 4070 crushes Pro in multiple areas. The point is Pro's raster performance paired with Nvidia-like upscaling and Ray tracing enhancements make it most similar to a 4070 in features.

It seems clear a PC with a 4070 will beat a pro in practically every game.

Exactly.

When the console doesn't even reach a 40% improvement over base, absolutely not. Even though there are some exceptions exceeding what was promised.

The thing is, with a 4070 you would never risk having developers forcing on you a broken DLSS implementation.

 

[Gaiff]

They were saying it's the closest GPU, not that it was literally on par if I remember. This still rings true.

PSSR despite its maligned reputation in the last few days is still far better than FSR 99% of the time.

The ray tracing performance also seems better than what the 7800 XT can muster.

3070 Ti with its 8GB doesn't cut it.

So, while the 4070 is still faster, it's really the only one the ticks all the boxes to be compared to the Pro.

There will probably be a fairly comparable desktop equivalent soon in the 8700 or 8600 XT. The 5060 or 5060 Ti might also be in the same ballpark.

 

[LectureMaster]

At the end of the day, it is a matter of whether you think it's worth it or not.

I think a lot of folks are happy with their pros, as it's the most beefy console experience, and it's a matter of time before PSSR gets better.

If you chase absolute performance-to-price value, no doubt go with 4070.

 

[gooner4life_uk]

It surpasses the 4070 in some games, in others it doesn't.

 

[Interfectum]

Is the PS5 pro on par with a RTX 4070 like Digital Foundry claimed before launch?​

 

[TintoConCasera]

Nah, it's not at all. Still imo the best option for console bros for going through the rest of the gen.

If I were them tho, I would be lighting up some candles and praying that devs get their shit together when it comes to pro patches. Some good ones out there but also some stinkers that make it look worse than what it really is.

 

[twilo99]

The real question is the PS5 Pro better than the new Intel Arc B580?

If you are building a PS5 Pro "PC" the intel GPU is very tempting at that price point.

 

[TGO]

Not by spec but theoretically it was supposed to give a similar experience be based on what they knew

It looks more like a standard PS5 with a shitty FSR upscaler

But is more like this

 

[Vick]

It surpasses the 4070 in some games, in others it doesn't.

It's likely that where it surpasses it is more due to awful PC ports, like Callisto Protocol, combined with actual love put into Pro patches, or instances where developers intentionally made the Pro version better than PC (Dead Rising, DMC5, GoW Ragnarok), or some of those past Sony PC ports that performed much better on PS5 compared to equivalent-like PC hardware.

I don't think I'm missing many of those others that perform better on Pro, maybe RE Engine games like Resident Evil remakes with RT and Dragon's Dogma 2, but in the latter PSSR implementation isn't perfect anyway.

 

[HeisenbergFX4]

Not by spec but theoretically it was supposed to give a similar experience be based on what they knew

But is more like this

Correct

It’s always why I said real world performance

 

[Fahdis]

It surpasses the 4070.

Nah, it's not at all.

 

[kevboard]

nope. so far it seems to hang around somewhere between a 3060ti and a 4060ti

 

[Euler007]

Found the PCMR safe space.

 

[Corporal.Hicks]

The PS5Pro GPU has 3840 cores, just like the RX6800 on PC, and this card is a bit slower than the RTX4070, but there are examples where it can match it (screenshot below) in AMD optimized games. But keep in mind, the PS5Pro GPU has faster RT compared to the RX6800. In Alan Wake 2 the PS5Pro ray tracing performance is clearly behind the RTX4070, but in Callisto Protocol the PS5Pro might be even faster. According to digital foundry, the PS5Pro verison runs at 1296p-1440p (upscaled to 4K) with locked 60fps, while the RTX4070 has dips below 60fps even at 1440p FSRQ (960p). It's hard to explain the PS5Pro performance in this game. Either they optimized the PS5Pro version much better than the PC version, or the PS5Pro GPU is faster than we thought.

 

[Warnen]

If you mean as in a disappointment like when the 4070 was announced…. Yes

 

[Gaiff]

The PS5Pro GPU has 3840 cores, just like the RX6800 on PC, and this card is a bit slower than the RTX4070, but there are examples where it can match it (screenshot below) in AMD optimized games. But keep in mind, the PS5Pro GPU has faster RT compared to the RX6800. In Alan Wake 2 the PS5Pro ray tracing performance is clearly behind the RTX4070, but in Callisto Protocol the PS5Pro might be even faster. According to digital foundry, the PS5Pro verison runs at 1296p-1440p (upscaled to 4K) with locked 60fps, while the RTX4070 has dips below 60fps even at 1440p FSRQ (960p). It's hard to explain the PS5Pro performance in this game. Either they optimized the PS5Pro version much better than the PC version, or the PS5Pro GPU is faster than we thought.

Or they more than likely use different settings and resolutions because the PC doesn't have DRS as far as I'm aware. They could easily use different RT settings not available on PC as well. You cannot just take a PC benchmark dating back from 2 years ago and compare it to a random snippet on the Pro. You need to match the scenes, settings, and resolution (if possible). Or are we seriously going to believe the Pro performs like a 2080 Ti in one game and a 4070 Ti in another?

Scenes and settings can have an enormous impact on fps and this explains the disparity. Until we have something resembling like-for-like, no point in bringing those up.

Alan Wake 2 is clearly an outlier. The Pro won't get beaten out by a 3070 most of the time.

 

[Bojji]

For most games it's around 3070, 3070ti*, 6800. For some reason might be slower than 6800 in pure raster?

* Of course I'm talking about pure hardware power, 3070 is super limited by tiny vram amount.

For games optimized for PS5/unoptimized on pc (some Sony games) it will be faster than that. Maybe 4070 level or above?

 

[SonGoku]

The Pro is closer to the 3070 Ti than 4070

 

[TheChumpNation]

I bought a 4070 build a few days after the ps5 pro release.

So far I've been blown away. Currently playing cyberpunk with almost every setting at its highest, with dlss quality or balanced and framegen. Playing at above 60 FPS at 4k easily.

Also currently playing RDR2 at near maxed out (a few settings like tree tessellation off), also at 4k (native). Getting 55-65fps.

I don't think the ps5 pro would be giving me anywhere near this level of fidelity, and that's before factoring in how reliant you are on the dev when on console.

I've been very, very happy with my purchase. DLSS is such a cheat code.

 

[SlimySnake]

I dont blame him. If we go by the specs then yes, it shouldve been on par with the 4070. But the 16.7 tflops GPU is only offering 14 tflops worth of performance. Lower than that in some games.

The new RT cores are being held back by whatever the fuck is holding back the GPU. It's definitely not 2x more performance.

PSSR is also very heavy thats eating up precious tlfops.

Honestly, if you looked at that GPU size (60 CUs vs 36), 300 TOPs AI compute power, 2x more RT power, you would say the same thing. On paper, it looked great.

But.

They didnt increase the cache. Didnt add enough vram. Didnt increase the vram bandwidth. Didnt improve the CPU. All of those things result in a console severely compromised by its architecture despite the impressive specs. It's arguably the worst console Sony has produced in my lifetime. Cerny paired up the PS4 GPU with GDDR5 ram that made sure the PS4's GPU advantage didnt get bottlenecked by esram. He created an exceptional i/o block to ensure the GPU didnt get throttled like the XSX GPU did. But this time, he made mistakes that have resulted in a big GPU size that cost them a lot of money without the performance to back it up.

 

[Feel Like I'm On 42]

I'm not a PC gamer but I know enough to know that fuck no the Pro isn't on par with that. I have a Pro and it's overall a big disappointment. Sure, there are a few standout titles and a bunch of games that benefit from "boost mode", however the problem is in almost every really demanding game (sw outlaws, Alan Wake 2, Avatar, SH2) the Pro is performing terribly and it comes down to 2 main reasons and none of them are the "fault of the developers" which is what so many Pro cheerleaders want to believe.....

First off PSSR sucks where it's needed most, in games that have struggled with FSR due to low internal res, the Pro still can't run these games clean. Alan Wake 2 has had 4 patches since the Pro released and the image quality is still a disgrace..the shimmer using PSSR is even more distracting on the Pro than it ever was on PS5. I just got done checking out today's patch and playing through NYC into Caldera Street subway the shimmering terrible and I'm noticing new types of artifacting that's new to PSSR.

Secondly- why didn't Sony make the console more powerful on the GPU side, just enough so that games like AW2 and SH2 could have internal resolutions of 1200p-1440p? I don't think that's unreasonable for a $700-$800 system, is it?

While there's good reason to blame devs for not properly QA testing these patches and not including the original modes there's no reason to think that they're doing something "wrong" when implementing PSSR ...no, PSSR just sucks dealing with low res and certain games rendering. Look closely in games with "good" PSSR implementations and you can find similar artifacting just to a much lower degree.

There may be truth to certain games using older form of PSSR and I really hope that's true but right now there are a lot of bad looking Pro patches and they are all highly demanding games ..ironically i don't see Jedi Survivor being bad in fact that game looks amazing at 30 fps. I guess people really don't play 30 fps much on Pro ...but I judge this console on how well it handles the games that are historical taxing, next gen games and the Pro with its marquee PSSR is failing far more than not in those games...SH2, Avatar, AW2, Jedi (at 60 fps), Black Ops 6, SW Outlaws ...

PS there are issues with IQ in Spiderman 2 as well but I wouldn't classify it as bad per say.

 

[Corporal.Hicks]

Or they more than likely use different settings and resolutions because the PC doesn't have DRS as far as I'm aware. They could easily use different RT settings not available on PC as well. You cannot just take a PC benchmark dating back from 2 years ago and compare it to a random snippet on the Pro. You need to match the scenes, settings, and resolution (if possible). Or are we seriously going to believe the Pro performs like a 2080 Ti in one game and a 4070 Ti in another?

Scenes and settings can have an enormous impact on fps and this explains the disparity. Until we have something resembling like-for-like, no point in bringing those up.

Alan Wake 2 is clearly an outlier. The Pro won't get beaten out by a 3070 most of the time.

Maybe the PS5Pro use lower settings as you say, but I cant tell the difference between my PC running at maxed out settings and screenshots from the PS5Pro version. It seems the PS5pro has all RT features enabled in callisto protocol based on what I saw.

Here's how the RTX4070 runs this game at 1440p FSRQ with RT. That's 960p internally.

 

[kevboard]

but in Callisto Protocol the PS5Pro might be even faster. According to digital foundry, the PS5Pro verison runs at 1296p-1440p (upscaled to 4K) with locked 60fps, while the RTX4070 has dips below 60fps even at 1440p FSRQ (960p). It's hard to explain the PS5Pro performance in this game. Either they optimized the PS5Pro version much better than the PC version, or the PS5Pro GPU is faster than we thought.

the PC version is indeed awful.
but even so, the PS5 Pro version does not run max settings. shadow detail is far below what you can do on PC for example.

I got close to PS5 Pro performance at higher shadow detail settings on a 3060ti

 

[Vick]

Or they more than likely use different settings and resolutions because the PC doesn't have DRS as far as I'm aware. They could easily use different RT settings not available on PC as well. You cannot just take a PC benchmark dating back from 2 years ago and compare it to a random snippet on the Pro. You need to match the scenes, settings, and resolution (if possible). Or are we seriously going to believe the Pro performs like a 2080 Ti in one game and a 4070 Ti in another?

Scenes and settings can have an enormous impact on fps and this explains the disparity. Until we have something resembling like-for-like, no point in bringing those up.

Alan Wake 2 is clearly an outlier. The Pro won't get beaten out by a 3070 most of the time.

This is true. But in the case of Callisto Protocol he might be right about the extra care and optimization on Pro combined with DRS more than settings, as they actually exceed PC in some ways:

Pro:

PC:

And RT at the least in the only comparison we have, is a match to highest PC settings.

PS5 Pro Performance Mode:

RTX 3060ti Native 4K RT Reflections High:

4K FSR2 Performance Mode (1080p internal) RT Reflections High:

4K TAAU 50% (1080p internal) RT Reflections High:

If it offers a rocksolid 60fps with no traversal stutter and has this IQ in its Performance Mode:

It's fair to say it offers a better experience on console even if the RT shadows (which were entirely absent on base PS5 60fps mode) don't match PC resolution. Especially if you also enjoy gimmicks like DualSense features that actually do have an immersion impact in this kind of games.

For games optimized for PS5/unoptimized on pc (some Sony games) it will be faster than that. Maybe 4070 level or above?

If we want to go that route, there are games better on Pro than even a 4090. But has very little to do with hardware.

 

[HogIsland]

The real question is the PS5 Pro better than the new Intel Arc B580?

If you are building a PS5 Pro "PC" the intel GPU is very tempting at that price point.

If Battlemage isn't a compatibility nightmare, it definitely changes everybody's PCPartPicker vs Pro battles from a few months ago. We haven't seen performance like Intel's claiming @ $250 in a long time.

 

[Corporal.Hicks]

If we want to go that route, there are games better on Pro than even a 4090. But has very little to do with harhardware

People usually dont factor optimisation in their comparisons between PC and consoles, but good optimisation can increase fps a lot (even by a factor of 2x). For example, in Crysis 2, the tesselation was calculated for the whole map, even if you were looking at a wall. Moder Maldo optimised the tesselation in this game, so it was only calculated when tesselated objects were really on the screen. Thanks to this optimization my fps more than doubled and I could finally play this game at locked 60fps (I had GTX680 2GB back then).

 

[Bojji]

This is true. But in the case of Callisto Protocol he might be right about the extra care and optimization on Pro combined with DRS more than settings, as they actually exceed PC in some ways:

Pro:

PC:

And RT at the least in the only comparison we have, is a match to highest PC settings.

PS5 Pro Performance Mode:

PC Native 4K RT High:

If we want to go that route, there are games better on Pro than even a 4090. But has very little to do with hardware.

I'm specifically talking about something like the last of us part 1, clear outlier of performance when DF tested many games between PS5 and GPUs.

This game is just made by idiots on PC, if you put "PS5 GPU" (6700) into PC is performs MUCH better, 20% faster than 6800

How can anyone explain this shit? Normally 6800 is ~44% faster than 6700.

 

[Mister Wolf]

 

[Vick]

People usually dont factor optimisation in their comparisons between PC and consoles, but good optimisation can increase fps by a factor of 2x. For example, in Crysis, the tesselation was calculated for the whole map, even if you were looking at a wall. Moder Maldo optimised the tesselation in Crysis 2, so it was only rendering of tesselated objects were really on the screen and thanks to this optimization my fps more than doubled and I could finally play this game at locked 60fps (I had GTX680 2GB back then).

"Now that's a name I haven't heard in a long time" *gif*

Dude was legendary. His Crysis 2 Textures over that pixelated shit Crytek released (that many blamed on the console release) were out of this world.

 

[owandeseis]

To be fair, in different videos they said it was also a 4060ti, a 3070ti with extra VRAM, or a 7700XT, some people just only listened to the 4070 comparison.

Nvidia's GPU just have more and better tech, considering Frame Gen, it's always going to be ahead, and even not using it, it could be same as the Pro or even a bit worse on non RT games like the Sony ones, or it could double Pro's performance or more on games like AW2 or SH2.

The closest comparison right now is probably a 7700xt in non RT games and close to 7800xt on RT ones?

But it'll be easier to compare in a month with the new GPU's, the 8700XT should be pretty similar in all situations, and if Nvidia releases a 5060 12GB version, that might be it too

 

[twilo99]

If Battlemage isn't a compatibility nightmare, it definitely changes everybody's PCPartPicker vs Pro battles from a few months ago. We haven't seen performance like Intel's claiming @ $250 in a long time.

Right.. let's see how the drivers go but I'm guessing things will only get better

 

[hinch7]

The PS5Pro GPU has 3840 cores, just like the RX6800 on PC, and this card is a bit slower than the RTX4070, but there are examples where it can match it (screenshot below) in AMD optimized games. But keep in mind, the PS5Pro GPU has faster RT compared to the RX6800. In Alan Wake 2 the PS5Pro ray tracing performance is clearly behind the RTX4070, but in Callisto Protocol the PS5Pro might be even faster. According to digital foundry, the PS5Pro verison runs at 1296p-1440p (upscaled to 4K) with locked 60fps, while the RTX4070 has dips below 60fps even at 1440p FSRQ (960p). It's hard to explain the PS5Pro performance in this game. Either they optimized the PS5Pro version much better than the PC version, or the PS5Pro GPU is faster than we thought.

IW Engine runs considably faster on AMD hardware. That was the case for MW3 before it.

The 7900XTX for example gave comparative framerates to a 4090 in MW3.

 

[Bojji]

"Now that's a name I haven't heard in a long time" *gif*

Dude was legendary. His Crysis 2 Textures over that pixelated shit Crytek released (that many blamed on the console release) were out of this world.

Absolute legend. Crysis 2 with this had better textures than most 8 gen games.

 

[HogIsland]

Right.. let's see how the drivers go but I'm guessing things will only get better

Drivers have improved for sure, but per GamersNexus, Battlemage also has hardware changes meant to improve compatibility. If that bears out a few months after launch, I will grab one for sure.

 

[Zathalus]

The PS5Pro GPU has 3840 cores, just like the RX6800 on PC, and this card is a bit slower than the RTX4070, but there are examples where it can match it (screenshot below) in AMD optimized games. But keep in mind, the PS5Pro GPU has faster RT compared to the RX6800. In Alan Wake 2 the PS5Pro ray tracing performance is clearly behind the RTX4070, but in Callisto Protocol the PS5Pro might be even faster. According to digital foundry, the PS5Pro verison runs at 1296p-1440p (upscaled to 4K) with locked 60fps, while the RTX4070 has dips below 60fps even at 1440p FSRQ (960p). It's hard to explain the PS5Pro performance in this game. Either they optimized the PS5Pro version much better than the PC version, or the PS5Pro GPU is faster than we thought.

AMD outperforms NVIDIA when RT is enabled in Callisto. At 1440p with RT a 7900 XTX and 4090 are basically on par. Considering it is an AMD sponsored title back when AMD was preventing DLSS to be implemented, I doubt it was optimised for NVIDIA at all. Maybe even deliberately designed to be limiting on NVIDIA to make AMD look good.

 

[Gaiff]

This is true. But in the case of Callisto Protocol he might be right about the extra care and optimization on Pro combined with DRS more than settings, as they actually exceed PC in some ways:

Maybe the PS5Pro use lower settings as you say, but I cant tell the difference between my PC running at maxed out settings and screenshots from the PS5Pro version. It seems the PS5pro has all RT features enabled in callisto protocol based on what I saw.

The problem is this game has absolutely no granularity whatsoever and you cannot even change the settings in real time. You have to exit to the main menu.

Lighting for instance can only be toggled between Low and Standard and has a huge performance impact in some scenes while not so much in others. The difference can vary between less than 10% to over 25%. Consoles could easily be using a custom value between the two and only look a bit worse in some scenes and you'd think they're using the same setting.

Another one is volumetric. If you compare still shots, there really isn't much of a difference. You'd have to find the same place and compare side by side. However, the impact between Medium and High can be as much as 17%.

Spoiler

Once again, consoles could be using a custom value or even mix them up depending on the source. It could be large volumetrics use a lower value to save performance.

Another one is particles that can look very similar between High and Medium, but Medium runs 5-6% faster.

Spoiler

Then there's the ray tracing with RT shadows sometimes barely showing a difference but tanking the fps by 30%.

Here's how the RTX4070 runs this game at 1440p FSRQ with RT. That's 960p internally.

Surely, you have noticed this:

The game is completely single-threaded and this shows a severe CPU bottleneck. The 4070 is asleep at the wheel here drawing 116W.

Also, in this very video, the average for 1440p FSR Q is 64fps, but the average for 4K FSR Q is 80fps.

Spoiler

So going from basically 1440p to 4K increased the fps by a whopping 25%. Of course, they tested different areas, but this harks back to the point I made several posts ago that you cannot just use random benchmarks. You have to do like-for-like because as you can see here, the numbers can vary enormously. You'd think this game is more demanding at 1440p than at 4K with FSR Q using that video, but we all know this isn't the case.

 

[Fess]

I dont blame him. If we go by the specs then yes, it shouldve been on par with the 4070. But the 16.7 tflops GPU is only offering 14 tflops worth of performance. Lower than that in some games.

The new RT cores are being held back by whatever the fuck is holding back the GPU. It's definitely not 2x more performance.

PSSR is also very heavy thats eating up precious tlfops.

Honestly, if you looked at that GPU size (60 CUs vs 36), 300 TOPs AI compute power, 2x more RT power, you would say the same thing. On paper, it looked great.

But.

They didnt increase the cache. Didnt add enough vram. Didnt increase the vram bandwidth. Didnt improve the CPU. All of those things result in a console severely compromised by its architecture despite the impressive specs. It's arguably the worst console Sony has produced in my lifetime. Cerny paired up the PS4 GPU with GDDR5 ram that made sure the PS4's GPU advantage didnt get bottlenecked by esram. He created an exceptional i/o block to ensure the GPU didnt get throttled like the XSX GPU did. But this time, he made mistakes that have resulted in a big GPU size that cost them a lot of money without the performance to back it up.

Cerny said long ago that we can’t look at a TF number and CU count to judge the power of a system.
Is that already forgotten?

This whole versus scenario is odd. Why does it matter? It’s the most powerful PlayStation. If you don’t want to wait on exclusives and want the best experience you’ll need a Pro no matter if it’s like a 2080 or 4080.

 

[hououinkyouma00]

No.

Maybe I'm remembering this wrong, but didn't they say something more along the lines of since AMD doesn't have a more comparable GPU that it has the most "in common" with the 4070?

 

[vkbest]

So, Callisto Protocol is PC bad port, but this game is a good PS5 Port? Can you explain?

 

[Topher]

I like this topic, this is gonna be fun.

It has been a blast the gazillion other times PC and PS5 Pro have been compared so...

 

[sncvsrtoip]

Not even close, its just 30% faster than base ps5 and pssr is big let down.

 

[kevboard]

So, Callisto Protocol is PC bad port, but this game is a good PS5 Port? Can you explain?

at the very least the base PS5 version of AW2 was a very good port.
considering that this game is heavily optimised for mesh shading, while the base PS5 doesn't support mesh shading, it ran astonishingly well.

they clearly took extra care to use primitive shaders on PS5 instead of mesh shaders, which they only way later patched into the PC version to support GTX10 and RDNA1 cards.

so we can say that the base PS5 port of AW2 was given a lot of care. while we know that the PC port of Callisto Protocol is a hot mess

 

[vkbest]

at the very least the base PS5 version of AW2 was a very good port.
considering that this game is heavily optimised for mesh shading, while the base PS5 doesn't support mesh shading, it ran astonishingly well.

they clearly took extra care to use primitive shaders on PS5 instead of mesh shaders, which they only way later patched into the PC version to support GTX10 and RDNA1 cards.

so we can say that the base PS5 port of AW2 was given a lot of care. while we know that the PC port of Callisto Protocol is a hot mess

So basically you are telling me this is not optimized for Pro.

 

[scydrex]

What cpu where used on the PC? A 3600 or a cpu similar to the PS5?