• Light/Dark
  • Ever wanted an RSS feed of all your favorite gaming news sites? Go check out our new Gaming Headlines feed! Read more about it here.
  • We have made minor adjustments to how the search bar works on ResetEra. You can read about the changes here.

PS5 Speculation |OT12| - Aw hell, Transistor's running this again?

How hyped are you?

  • A little hyped

    Votes: 98 15.7%

  • A lot of hyped

    Votes: 50 8.0%

  • WALNUUUUUTSSSSS

    Votes: 222 35.5%

  • Hyped enough to eat this whole bag of walnuts

    Votes: 63 10.1%

  • Hyped enough to bite this moose

    Votes: 37 5.9%

  • Hyped enough to scramble a dozen eggs

    Votes: 39 6.2%

  • Hyped enough to be even more hyped, like, cyclical or something

    Votes: 116 18.6%
  • Total voters
    625
Status
Not open for further replies.
Reader mode
Trup1aya

Trup1aya

Literally a train safety expert
Member
Oct 25, 2017
21,542
MykhellMikado said:
Where are you getting these improvement claims? I looked at every single one of those links posted and absolutely non of them compare the Xbox One to the Xbox X and certainly aren't listing the improvements you are claiming.
Click to expand...
Click to shrink...

Gamesradar kept track... all sourced from the developers/publishers

For example here's Bethesda's announcement for how their games support X1X. Here's an interview about For Honor You can find similar articles about each title if you're interested.
 
Last edited:
Ashen one

Ashen one

Member
Feb 20, 2020
29
So could we come closer if we add the wattage needed for the different configurations? I would need help with that though.

After getting the comment that you can now disable SEs "left and right" (for BC) the 3SE-theory seems weak (but not totally dissmissed bc reasons). I read the whitepaper from AMD and agree it looks alot more scalabe now. My limited understanding is that there are no chips with more then 2SEs today and also the usual number of CUs per SA is 10 or 12 (5-6 Dual compute units). It also looks like the middle ground of disabled CUs for yield is 4, however there might be interesting to disable more for lower power usage (and perhaps cheaper chips?).

I also think they are clocked higher then MS. Im thinking 1,85Ghz (can also be tuned).

I also personally think Sony really wants the total wattage down because of Japan and Im guessing they dont want much more then 175w-190w total (max), (PS4 is 190w and PS3 was 200). If we can get an overview it would be easier to finetune.

I need help calculating and filling in the blanks (would settle for a wattage-calculation-formula and I can fill it in myself), but to me the most interesting options would then be:

Variant 1:
Max CU: 48
CU count after disabling 4: 44
Clock speed: 1,85Ghz
TF: 10,4
Watt usage for GPU:
Watt usage for total APU:

Variant 2:
Max CU: 60
CU count after disabling 6: 54
Clock speed: 1,85Ghz
TF: 12,7
Watt usage for GPU:
Watt usage for total APU:

Variant 3:
Max CU: 60
CU count after disabling 4: 56
Clock speed: 1,85Ghz
TF: 13,2
Watt usage for GPU:
Watt usage for total APU:
 

Deleted member 61469

Attempted to circumvent ban with alt account
Banned
Nov 17, 2019
1,587
People are really running wild because some unknown youtube channel twitter guy who has practically no viewers is looking for attention any way he can.
 
Brees2Thomas

Brees2Thomas

Member
Dec 27, 2019
1,525
SlipperyFishes said:
It's time to stop talking RDNA3 in this thread. It's not happening. AMD fucking mentioned RDNA2 a few days ago for both consoles.

this is just like when people thought the Wii and X1 might have some kind of secret sauce
Click to expand...
Click to shrink...
Why are you so certain it couldn't be RDNA2 with a RDNA3 feature or two? Like modiz posted earlier, it would be no different than the advanced RPM tech that PS4 Pro surprisingly added.
 
Doctor Avatar

Doctor Avatar

Member
Jan 10, 2019
2,648
anexanhume said:
Not saying this means anything, but from Playstation OT they pointed out that GI usually does a cover reveal first week of the month and that hasn't happened yet.
Click to expand...
Click to shrink...

Supposed to be a deep dive on PS5 after the Feb reveal media event.

Feb reveal cancelled due to coronavirus, they are waiting for Sony to put out their State of Play replacement.

Please.
 

Deleted member 5028

User requested account closure
Banned
Oct 25, 2017
9,724
Coda said:
People are really running wild because some unknown youtube channel twitter guy who has practically no viewers is looking for attention any way he can.
Click to expand...
Click to shrink...
Exactly. There's a great hope that Sony are going to pull out the most advanced system ever made with features that either don't exist or have never been discussed in the hope they'll one-up the XSX.
BlacknGoldBlood said:
Why are you so certain it couldn't be RDNA2 with a RDNA3 feature or two? Like modiz posted earlier, it would be no different than the advanced RPM tech that PS4 Pro surprisingly added.
Click to expand...
Click to shrink...
Because you're just hoping now based on a random person who posted to neogaf. If you're talking about it including Polaris tech, that was due for 2016 so sure, it made sense. RDNA 3 has absolutely not started production until at least 2021-22
 
III-V

III-V

Member
Oct 25, 2017
18,827
When we do not know all features of RDNA2, any discussion of RDNA3 features seem a bit premature.
 

Deleted member 62280

User requested account closure
Banned
Dec 18, 2019
497
BlacknGoldBlood said:
Why are you so certain it couldn't be RDNA2 with a RDNA3 feature or two? Like modiz posted earlier, it would be no different than the advanced RPM tech that PS4 Pro surprisingly added.
Click to expand...
Click to shrink...
I'd like to point out this is what some were arguing just last week RDNA 1 with RDNA 2 features now people are saying RDNA2 with RDNA3 features. People need to take a step back. AMD themselves are saying both are RDNA 2 that's it. Both could have RDNA 3 features but that's highly unlikely given they're not even set to be launched till 2021 or later.
 
Doctor Avatar

Doctor Avatar

Member
Jan 10, 2019
2,648
PS5 is obviously not going to be RDNA3, people need to catch themselves on.

There is absolutely no discernment when it comes to these rumours, the credible ones are given as much credence as patiently ridiculous ones. Aside from the people making hilarious snide remarks about RDNA3, you guys rule.

Both will be RDNA2.
 
kingdali

kingdali

Member
Feb 13, 2019
136
Both Consoles will have RDNA2 but the XSX will have more customized RDNA2 features and thats why they said exclusive.
PS5 will have a "standard" RDNA2.
 
Kyoufu

Kyoufu

Member
Oct 26, 2017
16,582
Doctor Avatar said:
PS5 is obviously not going to be RDNA3, people need to catch themselves on.

There is absolutely no discernment when it comes to these rumours, the credible ones are given as much credence as patiently ridiculous ones. Aside from the people making hilarious snide remarks about RDNA3, you guys rule.

Both will be RDNA2.
Click to expand...
Click to shrink...

The topic isn't about PS5 being on the RDNA3 architecture. It's about it using features from it, like how Pro's Polaris GPU used a Vega feature.

Not sure why people are lacking reading comprehension wrt this. We've been speculating about this possibility since the first OT.

But again, when we don't know the full RDNA2 featureset then it's pointless to speculate about RDNA3 features.
 

Deleted member 5028

User requested account closure
Banned
Oct 25, 2017
9,724
Kyoufu said:
The topic isn't about PS5 being on the RDNA3 architecture. It's about it using features from it, like how Pro's Polaris GPU used a Vega feature.

Not sure why people are lacking reading comprehension wrt this. We've been speculating about this possibility since the first OT.
Click to expand...
Click to shrink...
Because the fact is this feels exactly like when the One X was released and we had a whole bunch of people shouting FP16 means Pro is more powerful and doing all kinds of math to prove their point.

if you want to go down this route then perhaps both platforms have elements of RDNA 3. But they don't. Otherwise they would have said on the AMD stream that elements are used in next gen consolesto their investors. But they didn't.
 
Kyoufu

Kyoufu

Member
Oct 26, 2017
16,582
SlipperyFishes said:
Because the fact is this feels exactly like when the One X was released and we had a whole bunch of people shouting FP16 means Pro is more powerful and doing all kinds of math to prove their point.
Click to expand...
Click to shrink...

I don't recall that being a thing.

SlipperyFishes said:
if you want to go down this route then perhaps both platforms have elements of RDNA 3. But they don't. Otherwise they would have said on the AMD stream that elements are used in next gen consolesto their investors. But they didn't.
Click to expand...
Click to shrink...

They would be classed as "special sauce" or whatever that Lisa Su called it in an interview not too long ago, so they'd be things the platform holders would talk about in their reveals, not AMD. After all, it was Mark Cerny who confirmed RPM for PS4 Pro, not AMD.

You have to keep in mind that we don't know the full GPU features of either console. Why rule this kind of thing out, especially when there's precedent?
 
anexanhume

anexanhume

Member
Oct 25, 2017
12,918
Maryland
Couple of random thoughts -

III-V is right that RDNA 3 discussion is pointless in light of how little we know about RDNA 2. Here's what we actually learned about RDNA 2 this week:
  • It's coming this year (this was already assumed).
  • Big Navi is RDNA 2. There is no RDNA 1 part above 40 CU (this was also heavily assumed).
  • It has hardware ray-tracing. This was known.
  • It has VRS. This was a reasonable assumption.
  • It is an unknown 7nm process, despite being 7nm+ previously in the roadmap.
  • It has 50% performance-per-watt compared to RDNA 1.
Here's what we don't know:
  • Is it still on 7nm+, or N7P? Has it been designed and validated to both, since the libraries are incompatible?
  • Contingent on the previous point, does it have implications for consoles that use RDNA 2?
  • How is that 50% achieved? Is there any component of process, or is it solely logic and design enhancements coupled with IPC gains?
  • Of the IPC gains over RDNA 1, how much of that is VRS, which can have huge performance gains on its own. AMD only stated the 50% improvement was "according to internal estimates."
  • Further on the 50% PPW gains, we have no idea what variables are controlled in this test. If RDNA 2 is more efficient and can clock higher, pushing RDNA 1 to the same high clocks would make it wildly inefficient and distort the true improvements of RDNA 2 versus RDNA 1.
  • Is the rumor of a 505 mm^2 GPU with 80 CUs accurate? If so, it represents a large increase of area when you control for I/O and even an increased 384-bit GDDR6 bus.
Given how little we know of RDNA 2, there's no logical basis upon which to speculate about RDNA 3. The 50% PPW improvement of RDNA 2 with respect to RDNA 1 is not understood at all. Therefore, we can't speculate about what nebulous architecture enhancements might lie in RDNA 3 vs. RDNA 2 (and we know for certain a console won't be getting any process improvements - its node has already been decided).
 
Brees2Thomas

Brees2Thomas

Member
Dec 27, 2019
1,525
anexanhume said:
Couple of random thoughts -

III-V is right that RDNA 3 discussion is pointless in light of how little we know about RDNA 2. Here's what we actually learned about RDNA 2 this week:
  • It's coming this year (this was already assumed).
  • Big Navi is RDNA 2. There is no RDNA 1 part above 40 CU (this was also heavily assumed).
  • It has hardware ray-tracing. This was known.
  • It has VRS. This was a reasonable assumption.
  • It is an unknown 7nm process, despite being 7nm+ previously in the roadmap.
  • It has 50% performance-per-watt compared to RDNA 1.
Here's what we don't know:
  • Is it still on 7nm+, or N7P? Has it been designed and validated to both, since the libraries are incompatible?
  • Contingent on the previous point, does it have implications for consoles that use RDNA 2?
  • How is that 50% achieved? Is there any component of process, or is it solely logic and design enhancements coupled with IPC gains?
  • Of the IPC gains over RDNA 1, how much of that is VRS, which can have huge performance gains on its own. AMD only stated the 50% improvement was "according to internal estimates."
  • Further on the 50% PPW gains, we have no idea what variables are controlled in this test. If RDNA 2 is more efficient and can clock higher, pushing RDNA 1 to the same high clocks would make it wildly inefficient and distort the true improvements of RDNA 2 versus RDNA 1.
  • Is the rumor of a 505 mm^2 GPU with 80 CUs accurate? If so, it represents a large increase of area when you control for I/O and even an increased 384-bit GDDR6 bus.
Given how little we know of RDNA 2, there's no logical basis upon which to speculate about RDNA 3. The 50% PPW improvement of RDNA 2 with respect to RDNA 1 is not understood at all. Therefore, we can't speculate about what nebulous architecture enhancements might lie in RDNA 3 vs. RDNA 2 (and we know for certain a console won't be getting any process improvements - its node has already been decided).
Click to expand...
Click to shrink...
Bravo. I say we bookmark this post Transistor
 
Transistor

Transistor

Outer Wilds Ventures Test Pilot
Administrator
Oct 25, 2017
37,340
Washington, D.C.
Official Staff Communication
There have been a number of posts lately that have been less speculation and more trolling, hostility, and generalizing. People are allowed to speculate that the GitHub leak might be the final specs. People are allowed to speculate that they're not final. Calling out other members who believe what you don't is unacceptable and will be moderated as such. Cut the hostilities and keep the thread fun and informative without attacking each other.
 
III-V

III-V

Member
Oct 25, 2017
18,827
anexanhume said:
Couple of random thoughts -

III-V is right that RDNA 3 discussion is pointless in light of how little we know about RDNA 2. Here's what we actually learned about RDNA 2 this week:
  • It's coming this year (this was already assumed).
  • Big Navi is RDNA 2. There is no RDNA 1 part above 40 CU (this was also heavily assumed).
  • It has hardware ray-tracing. This was known.
  • It has VRS. This was a reasonable assumption.
  • It is an unknown 7nm process, despite being 7nm+ previously in the roadmap.
  • It has 50% performance-per-watt compared to RDNA 1.
Here's what we don't know:
  • Is it still on 7nm+, or N7P? Has it been designed and validated to both, since the libraries are incompatible?
  • Contingent on the previous point, does it have implications for consoles that use RDNA 2?
  • How is that 50% achieved? Is there any component of process, or is it solely logic and design enhancements coupled with IPC gains?
  • Of the IPC gains over RDNA 1, how much of that is VRS, which can have huge performance gains on its own. AMD only stated the 50% improvement was "according to internal estimates."
  • Further on the 50% PPW gains, we have no idea what variables are controlled in this test. If RDNA 2 is more efficient and can clock higher, pushing RDNA 1 to the same high clocks would make it wildly inefficient and distort the true improvements of RDNA 2 versus RDNA 1.
  • Is the rumor of a 505 mm^2 GPU with 80 CUs accurate? If so, it represents a large increase of area when you control for I/O and even an increased 384-bit GDDR6 bus.
Given how little we know of RDNA 2, there's no logical basis upon which to speculate about RDNA 3. The 50% PPW improvement of RDNA 2 with respect to RDNA 1 is not understood at all. Therefore, we can't speculate about what nebulous architecture enhancements might lie in RDNA 3 vs. RDNA 2 (and we know for certain a console won't be getting any process improvements - its node has already been decided).
Click to expand...
Click to shrink...
Good points. I am now thinking of the slide of RDNA2 50% perf, and I do not recall process improvements attributed to a portion like we saw with the GCN-> RDNA move. Could some of the process improvements be attributed to VRS? I should hope that is not how they consider the gains. That would be a little shady. dad joke but damn it now I see I was beat by Kyoufu
 
Jonnax

Jonnax

Member
Oct 26, 2017
4,973
Y'all realise that RDNA 2, 3 is branding right?

These consoles are under AMD's semi custom service. The console makers have input as to what features go into the GPU.

The likelyhood is that it will have features not available in RDNA 2 but also lack other features of RDNA 2.

GPU manufacturers rebadge their GPUs all the time.

Where an old gen card has a new gen part number.

Read this:
graphicscardhub.com

Rebrand, Rebadge vs Refresh of Graphics Cards Explained

The difference between Rebrand, Rebadge, and Refresh in relation to Graphics Cards and GPU. These terms are closely related to each other and there is a lot of confusion about what they mean in the context of graphics cards. If you are a graphics card user or are in process of buying a new...
graphicscardhub.com graphicscardhub.com

AMD isn't going to brand their console GPUs as RDNA 1.5 or whatever even if that was the reality.
 
anexanhume

anexanhume

Member
Oct 25, 2017
12,918
Maryland
Miniature Kaiju said:
Hardware TSS acceleration is RDNA3 and PS5 will have that.

I hope.

Kidding aside, since Sony seems to be still invested in VR, they'd definitely benefit a lot from TSS.
Click to expand...
Click to shrink...
You mean like this?

Texel shading in texture space - Advanced Micro Devices, Inc.

<div p-id="p-0001">A graphics processing unit is configured to map pixels of a first frame of a video stream to texels, select a subset of the texels for shading based on previously cached texels that
www.freepatentsonline.com www.freepatentsonline.com

Kyoufu said:
I'd find it weird and kinda shady if they were using VRS for their IPC estimates lol.
Click to expand...
Click to shrink...
iseegujyi.png
 
Last edited:
Miniature Kaiju

Miniature Kaiju

Member
Oct 27, 2017
7,166
Somewhere South
anexanhume said:
You mean like this?

Texel shading in texture space - Advanced Micro Devices, Inc.

<div p-id="p-0001">A graphics processing unit is configured to map pixels of a first frame of a video stream to texels, select a subset of the texels for shading based on previously cached texels that
www.freepatentsonline.com www.freepatentsonline.com
Click to expand...
Click to shrink...

Yup. TSS can enable them to render a second viewport almost for free (can reuse most shaded samples between render targets), and since it can vary shading rate temporally, get consistent 120Hz by dropping the rate when under load. Will make VR games look much, much better.
 
DavidDesu

DavidDesu

Banned
Oct 29, 2017
5,718
Glasgow, Scotland
Miniature Kaiju said:
Yup. TSS can enable them to render a second viewport almost for free (can reuse most shaded samples between render targets), and since it can vary shading rate temporally, get consistent 120Hz by dropping the rate when under load. Will make VR games look much, much better.
Click to expand...
Click to shrink...
For real? Wow. Still hoping for a really advanced headset with eye tracking and foveated rendering, which alone would boost VR visuals immensely. This as well... ooft.
 
Reader mode
Status
Not open for further replies.