PS5 Pro codenamed ‘PROJECT TRINITY’ details and release date

Alabtrosmyster

Veteran
26 Jun 2022
3,399
2,944
So that means MS mid gen refrish gets beat by Sony regular console? 🤡

The Series S was competing directly against the PS5 all along and did't managed to sell 1/4 the units despite costing as far as $250 less? Turned out to also be completely outclassed in every possible way by the PS5 DE as well? 🤡🤡🤡

If this is actually going to be MS take and they don't have something to compete with the PS5 Pro we are in for a unprecedented beating for the next of this gen and I doubt the Xbox brand can survive that.
Xbox made its name on having the most "powerful" console in its first generation, and even to some degree in the 360 gen... Many of their older fans are still high on the og xbox performance for its time.

Sadly, some people argued that the XB SS would really compete with, if not beat, the PS5 in most games because of some buzz word salad.
 
  • haha
Reactions: Umar and Nhomnhom

Nhomnhom

Banned
25 Mar 2023
8,414
11,560
Xbox made its name on having the most "powerful" console in its first generation, and even to some degree in the 360 gen... Many of their older fans are still high on the og xbox performance for its time.

Sadly, some people argued that the XB SS would really compete with, if not beat, the PS5 in most games because of some buzz word salad.
Now they have the worst console by any metric, congratulations to them.
 

ethomaz

Rebolation!
21 Jun 2022
12,086
9,803
Brasil 🇧🇷
PSN ID
ethomaz
I disagree that its the only reason is to make money. If it were not for pro models last gen would have left behind by PC. And you have gotten PS5 much sooner which would have been basically the PS4 pro.
You would have ended up spending the same on a new console cycle $499 regardless.
The last game I played on base PS4 let nothing in desire to PC 🤷‍♂️

But if you want some examples look at GoW or TLOU2.
 
  • Like
Reactions: Alabtrosmyster

Hezekiah

Veteran
23 Jul 2022
1,403
1,380
IMO releasing a Pro console gives MS the opportunity to release their next gen console a year or more before the PS6, if MS skips a pro console themselves. Given the state of things, that’s not a wise decision by Sony.
That would just mean an extra year of cross-gen games for the next Xbox, and a second chance for Sony to crush Xbox in the 'specs war' - with PS5 Pro, and PS6.
 
Last edited:

Hezekiah

Veteran
23 Jul 2022
1,403
1,380
IMO PS5 Pro will be worst upgrade than PS4 Pro that already was a bad choice.

I played all games on PS4 with the base model.

Mid-gen refresh exists only to make some users to buy two hardware in the same generation increasing revenue in the middle of
The gen.

You are better served buying 6-7 games than a mid-gen refresh imo.
Nobody is made to do anything.

You can upgrade as you would your phone, or not keep your current hardware. Users who buy a Pro have the chance to play games at a higher framerate, and perhaps increased resolution.

It's funny that people make out it's anything but a choice. And that the console is being written off by some well before it's even released.
 
24 Jun 2022
3,980
6,950
So a mid-gen refresh worst than base PS5 😂

@Welfare that is why we tell you are a Xbox fan.

He has no understanding of RDNA3 architecture going by this. For starters he (nor any of us) know if it's 60 CUs enabled or disabled. Going with RDNA3 architecture, it's likely 1 GCD x 2 SE x 2 SA x 5 WGP x 8 SIMD x 24 ALU. 4 CUs being disabled would be 3,584 shader cores, but that would only be possible by reducing the WGP count to 4 and increasing the ALU count to 28. Which is possible, but again we don't know.

His problem is that he is obviously speculating to insinuate the PS5 Pro is of no significant performance increase over the Series X, when the Series X has struggled to establish any lead over the PS5 matching the paper specs difference in just a small handful of areas. He doesn't understand the architectural changes and efficiency gains with just RDNA3 alone over RDNA2, or what the PS5 Pro will likely be (RDNA 3.5 or possibly RDNA4, both cases with custom features included).

So whatever TF perf it'd land at, it would perform cleanly over a base PS5 and Series X due to new GPU rendering features and improvements alone. He's basically falling for another "TF is everything" trap, but this time going by CU count 😄.
 

anonpuffs

Veteran
Icon Extra
29 Nov 2022
10,496
11,937
He has no understanding of RDNA3 architecture going by this. For starters he (nor any of us) know if it's 60 CUs enabled or disabled. Going with RDNA3 architecture, it's likely 1 GCD x 2 SE x 2 SA x 5 WGP x 8 SIMD x 24 ALU. 4 CUs being disabled would be 3,584 shader cores, but that would only be possible by reducing the WGP count to 4 and increasing the ALU count to 28. Which is possible, but again we don't know.

His problem is that he is obviously speculating to insinuate the PS5 Pro is of no significant performance increase over the Series X, when the Series X has struggled to establish any lead over the PS5 matching the paper specs difference in just a small handful of areas. He doesn't understand the architectural changes and efficiency gains with just RDNA3 alone over RDNA2, or what the PS5 Pro will likely be (RDNA 3.5 or possibly RDNA4, both cases with custom features included).

So whatever TF perf it'd land at, it would perform cleanly over a base PS5 and Series X due to new GPU rendering features and improvements alone. He's basically falling for another "TF is everything" trap, but this time going by CU count 😄.
I highly doubt the ps5 pro is using rdna3. There is too significant an architectural gap and the performance increase too insignificant. For one rdna3 double pumps fp32 calcs, that's a ton of transistor count dedicated to that and real world numbers give a paltry uplift in exchange. If anything it would be the ps5 architecture ported down to 4nm silicon and doubled in CU count like the ps4 pro was.
 

Shmunter

Veteran
22 Jul 2022
3,047
3,533
I highly doubt the ps5 pro is using rdna3. There is too significant an architectural gap and the performance increase too insignificant. For one rdna3 double pumps fp32 calcs, that's a ton of transistor count dedicated to that and real world numbers give a paltry uplift in exchange. If anything it would be the ps5 architecture ported down to 4nm silicon and doubled in CU count like the ps4 pro was.
They are talking additional RT capabilities tho. Could be Sony customisation.
 

historia

Veteran
29 Jun 2023
2,818
2,720
I highly doubt the ps5 pro is using rdna3. There is too significant an architectural gap and the performance increase too insignificant. For one rdna3 double pumps fp32 calcs, that's a ton of transistor count dedicated to that and real world numbers give a paltry uplift in exchange. If anything it would be the ps5 architecture ported down to 4nm silicon and doubled in CU count like the ps4 pro was.
RDNA 3 didn't increase transitors count. We know this because even those they decrease node and increase CUs, the die size decreased.

For how they tried to improve performance, I and @thicc_girls_are_teh_best already discuss above. Essential more SIMD and shader counts per CUs.

Double CUs count is a poor approach because there is a limit to how many CU can execute a specific instructions, and those perfomance gains will only get applied to at higher fidelity, lower fidelity will not get improvement in framerates. Besr course is pure IPC improvement which I hope RDNA 4 can achieve.
 

historia

Veteran
29 Jun 2023
2,818
2,720
I find it funny that MS shoot themselves in the feet by not only created the 52 CUs Series X and win marketing points, which was handcapped by 16 CUs Series S ,which ended up losing them shit ton of money and marketshare even got larger to Sony.

Navi literally an architecture designed to be operated at higher clocked compared to GCN, and improving cooling is cheaper than bigger die size on APU, which the more CUs or CUDA cores you go will, the more diminishing returns will occur.
 

anonpuffs

Veteran
Icon Extra
29 Nov 2022
10,496
11,937
RDNA 3 didn't increase transitors count. We know this because even those they decrease node and increase CUs, the die size decreased.

For how they tried to improve performance, I and @thicc_girls_are_teh_best already discuss above. Essential more SIMD and shader counts per CUs.

Double CUs count is a poor approach because there is a limit to how many CU can execute a specific instructions, and those perfomance gains will only get applied to at higher fidelity, lower fidelity will not get improvement in framerates. Besr course is pure IPC improvement which I hope RDNA 4 can achieve.
???

7900 XT (RDNA 3)
84 CUs (96 with 12 disabled) = 57.7 billion transistors

6950XT (RDNA 2)
80 CUs = 26.8 billion transistors

what exactly are you talking about?
 

anonpuffs

Veteran
Icon Extra
29 Nov 2022
10,496
11,937
Ok I admit I am wrong, IPC doesn't increase put transitors count increase. Moore laws is dead.
rdna 3 sucks, this isn't a problem with moore's law (which has been dead for like 15ish years), the ipc increase sucks because the architecture is bad. they put the L3 cache on chiplets and went to dual-issue fp32 shaders which increased power draw and latency and something is wrong with those dual issue shaders because they aren't giving twice the performance increase
 
  • they're_right_you_know
Reactions: ethomaz

historia

Veteran
29 Jun 2023
2,818
2,720
rdna 3 sucks, this isn't a problem with moore's law (which has been dead for like 15ish years), the ipc increase sucks because the architecture is bad. they put the L3 cache on chiplets and went to dual-issue fp32 shaders which increased power draw and latency and something is wrong with those dual issue shaders because they aren't giving twice the performance increase
I mean like monolithic design is better most of the time. I even improve rasitization alone. Their MCM touch up is just pathetic. What's next, relocate HVA traversal tree?
I think Moore's law is fine, given if there is enough bandwidth to work with. Diminishing return happens when you don't get enough headroom to improve, it basically happens at any field. CPUs performance does improve, just GPUs are handcapped by a lot of factors

Either way, more CUs will definitely not be enough for double performance uplift. There gonna be some drastic change in RDNA 4, which I doubt, unless Sony can come up with some proprietary solution of their own at API levels.
 

Gediminas

Boy...
Founder
21 Jun 2022
7,569
9,284
He has no understanding of RDNA3 architecture going by this. For starters he (nor any of us) know if it's 60 CUs enabled or disabled. Going with RDNA3 architecture, it's likely 1 GCD x 2 SE x 2 SA x 5 WGP x 8 SIMD x 24 ALU. 4 CUs being disabled would be 3,584 shader cores, but that would only be possible by reducing the WGP count to 4 and increasing the ALU count to 28. Which is possible, but again we don't know.

His problem is that he is obviously speculating to insinuate the PS5 Pro is of no significant performance increase over the Series X, when the Series X has struggled to establish any lead over the PS5 matching the paper specs difference in just a small handful of areas. He doesn't understand the architectural changes and efficiency gains with just RDNA3 alone over RDNA2, or what the PS5 Pro will likely be (RDNA 3.5 or possibly RDNA4, both cases with custom features included).

So whatever TF perf it'd land at, it would perform cleanly over a base PS5 and Series X due to new GPU rendering features and improvements alone. He's basically falling for another "TF is everything" trap, but this time going by CU count 😄.
He doesn't falling for any trap, he perfectly knows what he is doing.

Just xbots are dumb enough thinking that they can get away with this idiotic FUD takes.