• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.

Rumor: PS5 Pro Codenamed "Trinity" targeting Late 2024 (some alleged specs leaked)

Would you upgrade from your current PS5?

  • For sure

    Votes: 377 41.0%
  • Probably

    Votes: 131 14.2%
  • Maybe

    Votes: 127 13.8%
  • Unlikely

    Votes: 140 15.2%
  • Not a chance

    Votes: 145 15.8%

  • Total voters
    920
Honestly, the specs are very disappointing and as conservative as can be for a pro console. Its looking like almost a 1.5~x increase in compute and possibly bigger improvement which is just baffling...even the ps4 pro which was rather conservative itself and did the easy butterfly design managed a 2x+ boost in compute while the one x was a truly impressive console for a mid gen boost. This console almost comes across as a Nintendo style boost like the switch+ strategy which makes you wonder what is even the point...

If they are going for a premium console why not actually make a capable piece of hardware that can appeal to the segment you are trying to target? Price it at 600$+ if need be and the people who want the enthusiast product will buy it, but making a middling upgrade and still ending up charging 550$+ just comes across as greedy and disappointing.

Where are you getting 1.5x? It's going to be at least 18 Tflops. They also don't want this thing to cost $700+, so people gotta have reasonable expectations for a box that will likely cost $550 max.
 
Last edited:

iorek21

Member
He mentions average looking games. Dont think he cares about 60 fps modes.

Yep, 60fps is good and all, but there’s barely any really worthwhile in the technology department to make this gen shine.

Didn’t even need to be the RT fluff, I’d be satisfied with better physics, simulations, even AI.

This gen still haven’t shown us anything outside better resolutions, improved textures or RT. The tech still isn’t influencing gameplay enough for this gen to feel outdated like the PS4 felt back in 2016.
 
Yep, 60fps is good and all, but there’s barely any really worthwhile in the technology department to make this gen shine.

Didn’t even need to be the RT fluff, I’d be satisfied with better physics, simulations, even AI.

This gen still haven’t shown us anything outside better resolutions, improved textures or RT. The tech still isn’t influencing gameplay enough for this gen to feel outdated like the PS4 felt back in 2016.

I just needed games to play a lot better than the PS4 versions which the PS5 does in spades. I'm way more happy with my PS5 purchase than my PS4/PS4 Pro purchases and that is due to the much better frame rate modes.
 

SABRE220

Member
Blame AMD. Their 7000 series line of cards this thing is likely based on is complete trash from both a performance and thermal perspective. Its no Polaris which had a massive die shrink due to its 16nm design along with a 25% IPC gain. There are no IPC gains here and thermals for their 26 tflops card on 5nm are roughly on par with their 7nm 23 tflops card. It's a complete turd especially if you are Cerny and hoping for a massive power reduction for your console. At current specs, 60 CUs at 2.23 GHz, you are looking at the GPU alone taking 175 watts which is insane considering the other 10 tflops RDNA 2.0 cards like the 6600xt and 6700 non xt top out at 135 watts.

I think this thing will be $599. Bigger than the PS5 with a more expensive cooling solution. Besides, I highly doubt Sony would charge $499 for a premium console after increasing the price of the base version to $549.

Lets just hope they are able to get some power savings and IPC gains next year while they wait for RDNA 4.0. But if its based on RDNA3.0 then yes we are looking at a 72% performance increase which means 1080p games become 1440p games, and 1440p games become 1800p games. Not great, not terrible.
Yeah Amd has been embarrassing with their recent architectures despite them desperately trying to hype it prelaunch. I don't think there is any point of them releasing a console based on rdna3 which is just brutal in terms of efficiency and overall features compared to the competition. There is honestly no rush to release mid gen consoles considering there are barely any current-gen-only releases so waiting for rdna4 should be a core consideration but even then why are they aiming for such a small chip, especially for a midgen console? Even on rdna4 54cus won't pull miracles compute power would still be lacking even if rdna4 brings good efficiency improvements.
 
Last edited:

onQ123

Member
Honestly, the specs are very disappointing and as conservative as can be for a pro console. Its looking like almost a 1.5~x increase in compute and possibly bigger improvement which is just baffling...even the ps4 pro which was rather conservative itself and did the easy butterfly design managed a 2x+ boost in compute while the one x was a truly impressive console for a mid gen boost. This console almost comes across as a Nintendo style boost like the switch+ strategy which makes you wonder what is even the point...

If they are going for a premium console why not actually make a capable piece of hardware that can appeal to the segment you are trying to target? Price it at 600$+ if need be and the people who want the enthusiast product will buy it, but making a middling upgrade and still ending up charging 550$+ just comes across as greedy and disappointing.
So you have learned nothing at all from the 12TF vs 10 situation?
 
Last edited:

SlimySnake

Flashless at the Golden Globes
Yeah Amd has been embarrassing with their recent architectures despite them desperately trying to hype it prelaunch. I don't think there is any point of them releasing a console based on rdna3 which is just brutal in terms of efficiency and overall features compared to the competition. There is honestly no rush to release mid gen consoles considering there are barely any current-gen-only releases so waiting for rdna4 should be a core consideration but even then why are they aiming for such a small chip, especially for a midgen console? Even on rdna4 54cus won't pull miracles compute power would still be lacking even if rdna4 brings good efficiency improvements.
it's 60 CUs, not 54, but I agree, they should wait for full RDNA 4.0 or whatever is next for AMD. But if RDNA 3.5 brings them the IPC gains and thermal gains then thats ok.

As far as I am concerned, they are already a year late than they were last gen when they had the pro out by the end of the 3rd year. So they have waited. I guess MS is waiting an additional year and theirs will be the best mid gen console just like it was last gen. i think sony is willing to live with that and so am i. im already tired of playing these games at low resolutions. A 72% improvement in pixels will be a welcome addition in games like FF16 and dead space.
 

SABRE220

Member
it's 60 CUs, not 54, but I agree, they should wait for full RDNA 4.0 or whatever is next for AMD. But if RDNA 3.5 brings them the IPC gains and thermal gains then thats ok.

As far as I am concerned, they are already a year late than they were last gen when they had the pro out by the end of the 3rd year. So they have waited. I guess MS is waiting an additional year and theirs will be the best mid gen console just like it was last gen. i think sony is willing to live with that and so am i. im already tired of playing these games at low resolutions. A 72% improvement in pixels will be a welcome addition in games like FF16 and dead space.
Its not going to have a full die due to yield.

This gen is a bit of a different situation due to the slow start and massive supply chain issues/covid so being behind a year is not unreasonable. Microsoft are stuck with a tricky situation due to the series S debacle them releasing a significantly stronger Series XX is going to be a branding nightmare and its going to be very difficult to keep the series S in the optimization loop. Sony shouldn't be happy to live with that, Microsoft is in a situation where they might seriously struggle with a midgen console and a very capable ps5 pro would be the perfect branding situation to set ps5 as the premium console choice, I am also tired of the underwhelming releases from the consoles and the low resolutions etc but being happy with the bare minimum has led us to the current shitty state of tech this gen.
 
Last edited:

SlimySnake

Flashless at the Golden Globes
I am also tired of the underwhelming releases from the consoles and the low resolutions etc but being happy with the bare minimum has led us to the current shitty state of tech this gen.
This i agree with. I wanted a 23 tflops 6950xt equivalent which is exactly 100% faster than the 6600xt. That wouldve been our 2x increase in compute power but now I have to decide if some PS5 exclusives are enough for me to plunk down $600 seeing as how my 3080 will likely still outperform this 17.6 tflops AMD GPU in multiplats.

Lets hope Cerny uses his mind to something other than IO this time around and adds some tensor and nvidia RT core equivalent tech. AMD is still very far behind Nvidia in both RT and ML upscaling solutions. but both things will increase the size of the die and thus the thermals and cooling cost.
 
Last edited:

SolidQ

Member
AMD is still very far behind Nvidia in both RT and ML upscaling solutions
AMD will get Traversal in RDNA4. So they will be same with NV
That name Patents for RDNA4
1. BOUNDING VOLUME HIERARCHY HAVING ORIENTED BOUNDING BOXES WITH QUANTIZED ROTATIONS
2. ACCELERATION STRUCTURES WITH DELTA INSTANCES
3. TECHNIQUES FOR INTRODUCING ORIENTED BOUNDING BOXES INTO BOUNDING VOLUME HIERARCHY
4. SPATIAL HASHING FOR WORLD-SPACE SPATIOTEMPORAL RESERVOIR RE-USE FOR RAY TRACING
5. MACHINE-LEARNING BASED COLLISION DETECTION FOR OBJECTS IN VIRTUAL ENVIRONMENTS
6. OVERLAY TREES FOR RAY TRACING
7. GRAPHICS PROCESSING UNIT TRAVERSAL ENGINE
8. VARIABLE WIDTH BOUNDING VOLUME HIERARCHY NODES
9. COMMON CIRCUITRY FOR TRIANGLE INTERSECTION AND INSTANCE TRANSFORMATION FOR RAY TRACING
10. BOUNDING VOLUME HIERARCHY BOX NODE COMPRESSION
11. BVH NODE ORDERING FOR EFFICIENT RAY TRACING
12. FRUSTUM-BOUNDING VOLUME INTERSECTION DETECTION USING HEMISPHERICAL PROJECTION
So you see, no patents about use RT on shader cores
 
Last edited:

hinch7

Member
Blame AMD. Their 7000 series line of cards this thing is likely based on is complete trash from both a performance and thermal perspective. Its no Polaris which had a massive die shrink due to its 16nm design along with a 25% IPC gain. There are no IPC gains here and thermals for their 26 tflops card on 5nm are roughly on par with their 7nm 23 tflops card. It's a complete turd especially if you are Cerny and hoping for a massive power reduction for your console. At current specs, 60 CUs at 2.23 GHz, you are looking at the GPU alone taking 175 watts which is insane considering the other 10 tflops RDNA 2.0 cards like the 6600xt and 6700 non xt top out at 135 watts.

I think this thing will be $599. Bigger than the PS5 with a more expensive cooling solution. Besides, I highly doubt Sony would charge $499 for a premium console after increasing the price of the base version to $549.

Lets just hope they are able to get some power savings and IPC gains next year while they wait for RDNA 4.0. But if its based on RDNA3.0 then yes we are looking at a 72% performance increase which means 1080p games become 1440p games, and 1440p games become 1800p games. Not great, not terrible.
I'd would've said MCM is the culpret for the lackluster efficiency of RDNA 3. But looking across the stack, and the recently released 7600 which is monolithic and 6nm (vs 7nm of top Navi 23).. aaand its all a big F.

Not sure how they claim a 50% improvement gen on gen. There must've been some fuck up in testing(s) to final release because it the numbers just don't add up. Hopefully they get that sorted for next RDNA 4. Otherwise they might as well hand in the towel.
 
Last edited:
This i agree with. I wanted a 23 tflops 6950xt equivalent which is exactly 100% faster than the 6600xt. That wouldve been our 2x increase in compute power but now I have to decide if some PS5 exclusives are enough for me to plunk down $600 seeing as how my 3080 will likely still outperform this 17.6 tflops AMD GPU in multiplats.

Lets hope Cerny uses his mind to something other than IO this time around and adds some tensor and nvidia RT core equivalent tech. AMD is still very far behind Nvidia in both RT and ML upscaling solutions. but both things will increase the size of the die and thus the thermals and cooling cost.

They aren't releasing a Pro unless they managed to hit 2x the theoretical performance of base PS5.

It's Mark Cerny behind the wheel, why are we second guessing him at this point?
 

SlimySnake

Flashless at the Golden Globes
I'd would've said MCM is the culpret for the lackluster efficiency of RDNA 3. But looking across the stack, and the recently released 7600 which is monolithic and 6nm (vs 7nm of top Navi 23).. aaand its all a big F.

Not sure how they claim a 50% improvement gen on gen. There must've been some fuck up in testing(s) to final release because it the numbers just don't add up. Hopefully they get that sorted for next RDNA 4. Otherwise they might as well hand in the towel.
Hmm, but Sony wouldnt need to multi-stack a 60 CU chip right? So 5nm 60 CU chip should get a 40% thermal reduction. That means a 120 watt 6800 (down from 200 watts it is today at 7nm). 6800 is 16 tflops so a 17.5 tflops 60 CU PS5 would have it the same thermal budget as the PS5's GPU. definitely doable.

AMD will get Traversal in RDNA4. So they will be same with NV
That name Patents for RDNA4
1. BOUNDING VOLUME HIERARCHY HAVING ORIENTED BOUNDING BOXES WITH QUANTIZED ROTATIONS
2. ACCELERATION STRUCTURES WITH DELTA INSTANCES
3. TECHNIQUES FOR INTRODUCING ORIENTED BOUNDING BOXES INTO BOUNDING VOLUME HIERARCHY
4. SPATIAL HASHING FOR WORLD-SPACE SPATIOTEMPORAL RESERVOIR RE-USE FOR RAY TRACING
5. MACHINE-LEARNING BASED COLLISION DETECTION FOR OBJECTS IN VIRTUAL ENVIRONMENTS
6. OVERLAY TREES FOR RAY TRACING
7. GRAPHICS PROCESSING UNIT TRAVERSAL ENGINE
8. VARIABLE WIDTH BOUNDING VOLUME HIERARCHY NODES
9. COMMON CIRCUITRY FOR TRIANGLE INTERSECTION AND INSTANCE TRANSFORMATION FOR RAY TRACING
10. BOUNDING VOLUME HIERARCHY BOX NODE COMPRESSION
11. BVH NODE ORDERING FOR EFFICIENT RAY TRACING
12. FRUSTUM-BOUNDING VOLUME INTERSECTION DETECTION USING HEMISPHERICAL PROJECTION
So you see, no patents about use RT on shader cores
About fucking time. Lets hope the Pro gets these improvements. that alone should be a 50% improvement in RT heavy games like cyberpunk.
 
Hmm, but Sony wouldnt need to multi-stack a 60 CU chip right? So 5nm 60 CU chip should get a 40% thermal reduction. That means a 120 watt 6800 (down from 200 watts it is today at 7nm). 6800 is 16 tflops so a 17.5 tflops 60 CU PS5 would have it the same thermal budget as the PS5's GPU. definitely doable.


About fucking time. Lets hope the Pro gets these improvements. that alone should be a 50% improvement in RT heavy games like cyberpunk.

Pretty sure it will be on 4nm for a 2024 release.
 

Knightime_X

Member
So you're extending your generation by additional 3 years? Is that really worth $500?
It'll be worth $1,000 + no launch scalper headaches.
Not to mention that by the time pro is out many games will be much cheaper by then.
 
Last edited:

NEbeast

Banned
Read that shit again about MS saying XSX is their mid gen refresh, S is the standard one. Hell no it isn't. This truly could be their first real big mistake if they truly see it like this. They need to be serious about this, because I think we've all seen cloud isn't taking off anytime soon, go all in with consoles please.


Not gonna be pretty when many games manage to be 60fps on a PS5 Pro and the XSX struggling or not even going past 30fps.

I might really end up getting multiplatform titles starting late next year until next gen on a PS system and I never thought I'd say that.
They are hoping gamepass and getting games for "free" is going to be enough of a carrot to push people to xbox. Atm I just don't see that happening: most people are happy to pay for good games and want them to run the best they can. The SS really put MS in a bad spot to start the generation.
 

chilichote

Member
I wonder if it would make sense for Sony if the PS5 Pro got a big focus on ray tracing - that would be much more of a selling point for the PS6, because what are you going to use to push it when the buzzword "ray tracing" is already "used"?
 

skit_data

Member
I wonder if it would make sense for Sony if the PS5 Pro got a big focus on ray tracing - that would be much more of a selling point for the PS6, because what are you going to use to push it when the buzzword "ray tracing" is already "used"?
Path tracing or something I guess, I barely know what it is but I'll probably get to know in a few years
 

SlimySnake

Flashless at the Golden Globes
My point is don't look at numbers alone
yeah. I got that. And Im telling you that when looking at two consoles with the same exact architecture, the numbers are literally what you have to look at. Be it CUs, higher clocks, tflops, or IPC gains. The numbers in this case tell the whole story. We are NOT talking about PS5 vs XSX anymore.
 

DeepEnigma

Gold Member
yeah. I got that. And Im telling you that when looking at two consoles with the same exact architecture, the numbers are literally what you have to look at. Be it CUs, higher clocks, tflops, or IPC gains. The numbers in this case tell the whole story. We are NOT talking about PS5 vs XSX anymore.
What numbers do we have that can tell the whole story?
 

RoadHazard

Gold Member
You referenced Sony getting away with it thanks to MS. But the industry has been getting away with it back when MS was still everyone's champion, and Nintendo led that charge to great success (DSi accounted for around 80% of DS sales in the years after its release - far more than any Pro / X unit managed to date). What platform holder wouldn't be excited at the promise of increased MSRP *and* more sales in the later part of console's lives?

And it's not like this wasn't attempted before - it's just that GC and PS2 versions of it were both market failures, so not much to talk about there.

There was a GameCube Pro and PS2 Pro? That's news to me.
 

Fafalada

Fafracer forever
There was a GameCube Pro and PS2 Pro? That's news to me.
IMG_0795-scaled.jpeg
Console_psx.jpg
 

SlimySnake

Flashless at the Golden Globes
What numbers do we have that can tell the whole story?
Allow me to rephrase. The numbers in this case (PS5 vs PS5 Pro) WOULD tell the whole story. So far we have CUs and vram bandwidth, but once we have the rest of the specs, they would tell the whole story. This bizarre fascination with downplaying numbers is how we turn into a cult. Lets stick with facts instead of faith.
 

DeepEnigma

Gold Member
Allow me to rephrase. The numbers in this case (PS5 vs PS5 Pro) WOULD tell the whole story. So far we have CUs and vram bandwidth, but once we have the rest of the specs, they would tell the whole story. This bizarre fascination with downplaying numbers is how we turn into a cult. Lets stick with facts instead of faith.
And the supposed leaked CUs with 3 units per CU instead of the current 2 with the same clocks needed for compatibility profiles would be a 30TF machine. That sounds more like a new gen, so I don't think any of this is accurate, but I would like to be pleasantly surprised.
 

SlimySnake

Flashless at the Golden Globes
And the supposed leaked CUs with 3 units per CU instead of the current 2 with the same clocks needed for compatibility profiles would be a 30TF machine. That sounds more like a new gen, so I don't think any of this is accurate, but I would like to be pleasantly surprised.
I wasnt aware of this leak. Sounds awesome.
 

onQ123

Member
yeah. I got that. And Im telling you that when looking at two consoles with the same exact architecture, the numbers are literally what you have to look at. Be it CUs, higher clocks, tflops, or IPC gains. The numbers in this case tell the whole story. We are NOT talking about PS5 vs XSX anymore.
Actually you're still wrong

And nothing that has been leaked tell us how this console will perform .
 

ChiefDada

Member
Allow me to rephrase. The numbers in this case (PS5 vs PS5 Pro) WOULD tell the whole story. So far we have CUs and vram bandwidth, but once we have the rest of the specs, they would tell the whole story. This bizarre fascination with downplaying numbers is how we turn into a cult. Lets stick with facts instead of faith.

No it wouldn't if the rumors/likelihood of RDNA 4 utilizing dedicated traversal engines instead of shader cores for RT traversal compute. Different architecture and fixed function hardware ideally yields improved efficiency.
 

DeepEnigma

Gold Member
I wasnt aware of this leak. Sounds awesome.
I'm just going by the numbers people are saying in here based on the WGP and current clock speeds. I am speculating on the clock speeds being the same (or more, but not less) based on not breaking compatibility within profiles on the base machine. The clocks can't be less than the base system on such low extraction layers for their games without running into issues I would think, going by the PS4Pro and the Cerny talks.

But again, that was based on 30WGP for the TF range based on this.


However, it makes more sense that not all 60 would be active (for yields and more per wafer), and only 54 possibly with another 18 added on to the already 36 (18+18+18=54, which would be less than 30TF, probably in the 26-27TF range at those clocks) but supposedly with 3 shader engines per CU instead of 2 on the current system. I wonder if the extra shader engine plays into the RT acceleration modifications that are also heavily rumored.

Then again, what if all this is wrong and it's still 36, but with 3 units per CU, higher clocks, and other custom modifications. Then the TF would be much lower, but the gains would be made up for the RT focus. Who knows.
 
Last edited:

StereoVsn

Gold Member
Blame AMD. Their 7000 series line of cards this thing is likely based on is complete trash from both a performance and thermal perspective. Its no Polaris which had a massive die shrink due to its 16nm design along with a 25% IPC gain. There are no IPC gains here and thermals for their 26 tflops card on 5nm are roughly on par with their 7nm 23 tflops card. It's a complete turd especially if you are Cerny and hoping for a massive power reduction for your console. At current specs, 60 CUs at 2.23 GHz, you are looking at the GPU alone taking 175 watts which is insane considering the other 10 tflops RDNA 2.0 cards like the 6600xt and 6700 non xt top out at 135 watts.

I think this thing will be $599. Bigger than the PS5 with a more expensive cooling solution. Besides, I highly doubt Sony would charge $499 for a premium console after increasing the price of the base version to $549.

Lets just hope they are able to get some power savings and IPC gains next year while they wait for RDNA 4.0. But if its based on RDNA3.0 then yes we are looking at a 72% performance increase which means 1080p games become 1440p games, and 1440p games become 1800p games. Not great, not terrible.
It will be $599 and without disk drive, IMO, if the Slim gets released. That way they can save on space/price in Pro and use same disk SKU as the Slim.
 

Loxus

Member
Why would Sony disable 6 compute units if the total is 60? Both the PS5 and the series consoles have 4 out of their total CUs disabled and 2 in the previous gen.
The more CUs, the lower yields you'll get.

Also, it plays into backwards compatibility.
54 ÷ 3 = 18CUs
18 CUs for PS4 BC
18 + 18 = 36CUs for PS5 compatibility.
18 + 18 + 18 = 54CUs for PS5 Pro compatibility.

Both PS5 and the Serises consoles follow the same concept of disabling one WGP per shader engine.

Both PS5 and the Serises consoles only have 2 shader engines.
1 WGP = 2CUs
Disabling 1WGP for each shader engine = 4CUs.

As for the PS4 and XboxOne gen.
One CU for each shader engine is disabled because it doesn't carry the WGP concept, where you can't disable 1 CU within a WGP.

So in a way, PS4 and XboxOne gen is still following the same concept of disabling 1 WGP per shader engine, but since the CUs aren't pair together, 1 CU is disabled.
 
Last edited:

StereoVsn

Gold Member
Yeah Amd has been embarrassing with their recent architectures despite them desperately trying to hype it prelaunch. I don't think there is any point of them releasing a console based on rdna3 which is just brutal in terms of efficiency and overall features compared to the competition. There is honestly no rush to release mid gen consoles considering there are barely any current-gen-only releases so waiting for rdna4 should be a core consideration but even then why are they aiming for such a small chip, especially for a midgen console? Even on rdna4 54cus won't pull miracles compute power would still be lacking even if rdna4 brings good efficiency improvements.
RDNA 4 on say a 4-5 nm node, at sufficiently low prices for a console is going to be too close to potential PS6 release, IMO. Sony wouldn't want their enthusiast crowd buying into Pro and then releasing PS6 couple years later.
 

SABRE220

Member
RDNA 4 on say a 4-5 nm node, at sufficiently low prices for a console is going to be too close to potential PS6 release, IMO. Sony wouldn't want their enthusiast crowd buying into Pro and then releasing PS6 couple years later.
A 2x boost in power/performance over the ps5 is hardly going to keep the enthusiast market from buying a new generation console come on man. This is being seriously optimistic considering amds pathetic tech advancements as of late. In terms of actual cu/diesize the compute boost is slightly above 1.5x and hopefully rdna4 can make up for that.

Not to mention this generation is going to go on for a while considering the complications suffered at launch, developers are not willing to even move from their lastgen pipelines four years in, and they aren't going to be happy or willing developing for a new generation any time soon.
 
Last edited:

SABRE220

Member
So you have learned nothing at all from the 12TF vs 10 situation?
That must have sounded smarter in your head...we are talking about the same console manufacturer,using the same development tools/api etc based on leaks getting only a 1.5x increase in compute power and 25%boost in bandwidth, being extremely optimistic you can hope that rdna3.5 etc will bring up signifigant ipc improvements to bring the overall performance improvement up close to 2x which is already conservative considering the ps4pro. In your head what magical shit will suddenly make the ps5pros gpu suddenly perform much faster than its compute potential with only 54cus? Have you seen amds pathetic gains in efficiency in rdna3?
 
Last edited:

SABRE220

Member
Where are you getting 1.5x? It's going to be at least 18 Tflops. They also don't want this thing to cost $700+, so people gotta have reasonable expectations for a box that will likely cost $550 max.
Ps5 is 10.2 tflops at 36 cus, assuming they keep the same clocks due to console cooling etc it would produce 15.8 tflops at 56 cus(60cus based on the current rumor and due to yield issues atleast 4 cus will be disabled). Dont fall for rdna3's inflated tfop numbers, this thing is only a 50 percent increase in the number of compute processors which is underwhelming compared to even the conservative ps4pro which managed a 2.2x boost in compute.

Hopefully, the rumors are halfbaked and it uses a bigger/better rdna4 based gpu but being optimistic about amd has only led to disappointment.
 
Last edited:
Top Bottom