• 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.

Next-gen PS5 and next Xbox speculation launch thread |OT8| - The Dark Tower (See Staff Post)

Status
Not open for further replies.
Reader mode
Outrun

Outrun

Member
Oct 30, 2017
5,780
Does it give any one any satisfaction for one machine to be more powerful than the other?

Sony and MS know what they are doing. Heck, they already have the road map for the next iteration.

Insiders, paste bins, github, excel.... The system architects are doing their thing.

It will be glorious for all.
 
CelestialAtom

CelestialAtom

Mambo Number PS5
Member
Oct 26, 2017
6,032
Outrun said:
Does it give any one any satisfaction for one machine to be more powerful than the other?

Sony and MS know what they are doing. Heck, they already have the road map for the next iteration.

Insiders, paste bins, github, excel.... The system architects are doing their thing.

It will be glorious for all.
Click to expand...
Click to shrink...

I just wish more gamers saw it this way. Either way, this next generation will be fucking awesome.
 
Hybrid Rainbow

Hybrid Rainbow

Banned
Jun 23, 2019
6,446
Flutter said:
In the next 3 months we will look back at this moment and think;

We based the speculations since July on terrible backup excel files by an intern.
Click to expand...
Click to shrink...


giphy.gif


I said hot damn!
 
Patent

Patent

Self-requested ban
Banned
Jul 2, 2018
1,621
North Carolina
Outrun said:
Does it give any one any satisfaction for one machine to be more powerful than the other?

Sony and MS know what they are doing. Heck, they already have the road map for the next iteration.

Insiders, paste bins, github, excel.... The system architects are doing their thing.

It will be glorious for all.
Click to expand...
Click to shrink...
Yea i said this the other day but the people who make these systems for both companies are so smart and hardworking, the idea that one of them is gonna fumble the technical aspects is so silly, the systems will make sense for the price, if its 499 or 399 you will be getting your money's worth
 
TheThreadsThatBindUs

TheThreadsThatBindUs

Member
Oct 26, 2017
6,151
United Kingdom
DukeBlueBall said:
powerscalinggpuonlyuljwr.png

From an user test that I saw elsewhere.
The GPU only die power for ~200+W console seems to be in 1800-1950mhz. With a years time that range might be at 2000-2150 mhz.

In other words, 150W for the GPU alone in the PS5 is believable. That gets you 1950mhz in 2019. It might get you 2150mhz in 2020.
Click to expand...
Click to shrink...

It's pretty weird to me that anyone can look at this chart and think Sony will want to put a 36CU 2GHz part in a console—even worse to then argue that such aligns with typical console TDP and power consumption estimates.

I think you have to make the implicit assumption that Sony and AMD's engineers don't know what they're doing.

A larger die, wider shader array and lower clocked part targeting the same raw TF number would make for a significantly cheaper console BOM; regardless of the higher die cost (which itself is debatable, because your 2GHz clock in the former case will throw your yields down the toilet).
 
AegonSnake

AegonSnake

Banned
Oct 25, 2017
9,566
DukeBlueBall said:
powerscalinggpuonlyuljwr.png

From an user test that I saw elsewhere.
The GPU only die power for ~200+W console seems to be in 1800-1950mhz. With a years time that range might be at 2000-2150 mhz.

In other words, 150W for the GPU alone in the PS5 is believable. That gets you 1950mhz in 2019. It might get you 2150mhz in 2020.
Click to expand...
Click to shrink...
right thats the simulating gonzalo graph i love. all i see here is 25% performance gains for 75% power increase when going from 1.6ghz to 2.0 ghz.
 
OT9 Thread Title Poll
  • Index
    • OP
    OP
    Mecha Meister

    Mecha Meister

    Next-Gen Guru
    Member
    Oct 25, 2017
    2,800
    United Kingdom
    Last edited:
  • Index

    • Deleted member 19767

    User requested account closure
    Banned
    Oct 27, 2017
    2,098
    TheThreadsThatBindUs said:
    It's pretty weird to me that anyone can look at this chart and think Sony will want to put a 36CU 2GHz part in a console—even worse to then argue that such aligns with typical console TDP and power consumption estimates.

    I think you have to make the implicit assumption that Sony and AMD's engineers don't know what they're doing.

    A larger die, wider shader array and lower clocked part targeting the same raw TF number would make for a significantly cheaper console BOM; regardless of the higher die cost (which itself is debatable, because your 2GHz clock in the former case will throw your yields down the toilet).
    Click to expand...
    Click to shrink...

    I'm surprised there's anyone thinking Sony are going with high clock 36cu. I mean, it's not impossible - but it really looks unlikely if you assume that they are looking to build the most powerful, cost efficient console they can - same as Microsoft.
     

    Deleted member 1589

    User requested account closure
    Banned
    Oct 25, 2017
    8,576
    How do we know it was an intern?
    He left his resume in the folder.

    Why do you think the excel files were a mess?
    'cs_buffer_load_32bit_c2_cacheswizzle0_half_channel_arden' was apparently the first test done to check on Oberon's BC1 regression test, among other things.

    What is the main question we should ask ourselves?
    Can we tell what are the facts and which are 'facts' based on 6 months of speculation?
     
    Hybrid Rainbow

    Hybrid Rainbow

    Banned
    Jun 23, 2019
    6,446
    TheThreadsThatBindUs said:
    It's pretty weird to me that anyone can look at this chart and think Sony will want to put a 36CU 2GHz part in a console—even worse to then argue that such aligns with typical console TDP and power consumption estimates.

    I think you have to make the implicit assumption that Sony and AMD's engineers don't know what they're doing.

    A larger die, wider shader array and lower clocked part targeting the same raw TF number would make for a significantly cheaper console BOM; regardless of the higher die cost (which itself is debatable, because your 2GHz clock in the former case will throw your yields down the toilet).
    Click to expand...
    Click to shrink...

    So if Sony were going wide and slow, they would have to be pushing at least 56 CUs at 1.5Ghz+ to even get above Stadia. This is not making any sense to me arrrghh.
     
    Dashful

    Dashful

    Community Resettler
    Member
    Oct 25, 2017
    2,399
    Canada
    Outrun said:
    Does it give any one any satisfaction for one machine to be more powerful than the other?

    Sony and MS know what they are doing. Heck, they already have the road map for the next iteration.

    Insiders, paste bins, github, excel.... The system architects are doing their thing.

    It will be glorious for all.
    Click to expand...
    Click to shrink...
    If devs are happy with both systems and don't feel restrained by one or the other. I don't care which system is stronger either. We got impossibly great games on consoles with a jaguar cpu in them this gen. No one is losing next-gen.
     
    DukeBlueBall

    DukeBlueBall

    Banned
    Oct 27, 2017
    9,059
    Seattle, WA
    Flutter said:
    How do we know it was an intern?
    He left his resume in the folder.

    Why do you think the excel files were a mess?
    'cs_buffer_load_32bit_c2_cacheswizzle0_half_channel_arden' was apparently the first test done to check on Oberon's regression test, among other things.

    What is the main question we should ask ourselves?
    Can we tell what are the facts and which are 'facts' based on 6 months of speculation?
    Click to expand...
    Click to shrink...

    The Arden stuff in Oberon results was easy to explain.

    Typo in script config file.
    owbFLzp.png


    If you can read python code, the script hits some test servers and grabs test data and outs them into already populated Xlsm files.

    Spaghetti code.
     
    TheThreadsThatBindUs

    TheThreadsThatBindUs

    Member
    Oct 26, 2017
    6,151
    United Kingdom
    Hybrid Rainbow said:
    So if Sony were going wide and slow, they would have to be pushing at least 56 CUs at 1.5Ghz+ to even get above Stadia. This is not making any sense to me arrrghh.
    Click to expand...
    Click to shrink...

    Then if wider and slower isn't possible, how is anyone thinking MS is reaching 12TFLOPs?

    My personal, position is that 50+CUs reasonably clocked is the way to go, if they want high end performance. <400mmsq APU dies would mean a larger APU cost versus the previous gen., however, less RAM chips (PS4 use day 16) and a completely custom SSD would mean they can claw back some cost, and the lower clocked GPU would keep the cooling and chassis design under control.

    Better yet, HBM would further lower power consumption and PCB complexity, and from the recent details discussed by chris 1515 , it seems HBM2(E) isn't as costly as many want to believe.
     
    Silencerx98

    Silencerx98

    Banned
    Oct 25, 2017
    1,289
    TheThreadsThatBindUs said:
    It's pretty weird to me that anyone can look at this chart and think Sony will want to put a 36CU 2GHz part in a console—even worse to then argue that such aligns with typical console TDP and power consumption estimates.

    I think you have to make the implicit assumption that Sony and AMD's engineers don't know what they're doing.

    A larger die, wider shader array and lower clocked part targeting the same raw TF number would make for a significantly cheaper console BOM; regardless of the higher die cost (which itself is debatable, because your 2GHz clock in the former case will throw your yields down the toilet).
    Click to expand...
    Click to shrink...
    Hmm, what do you think Oberon is, good sir?
     
    DvdGzz

    DvdGzz

    Banned
    Mar 21, 2018
    3,580
    Idc if PS5 is weaker because it will likely be cheaper. It will be my Sony exclusive machine and the cheaper it is, the better. Game Pass already ensures I'll be mostly playing on Xbox so it's nice that I will get the best console version of games.
     
    AegonSnake

    AegonSnake

    Banned
    Oct 25, 2017
    9,566
    TheThreadsThatBindUs said:
    A larger die, wider shader array and lower clocked part targeting the same raw TF number would make for a significantly cheaper console BOM; regardless of the higher die cost (which itself is debatable, because your 2GHz clock in the former case will throw your yields down the toilet).
    Click to expand...
    Click to shrink...
    dont forget that AMD's patent for hardware RT has it go inside the texture mapping units which are part of the CUs along with the shader processors.

    fewer CUs = fewer texture units = worse Ray tracing.

    and like i pointed out, a 40 cu 2.0 ghz part gives you 10.24 tflops for 160w+ according to this graph. for 20 fewer watts you can get a 12 tflops gpu. why would sony go for 20% worse performance for 20% more power?
     
    saintjules

    saintjules

    Member
    Dec 20, 2019
    2,540
    JJD said:
    What prediction? Y'all gon crazy
    Click to expand...
    Click to shrink...
    I forgot to put *sarcasm* in the OG comment. A prediction for Cory to join Microsoft and make a game for Series X.

    Anyway, as an aside this is stemming from a guy having a meltdown about MLB no longer being exclusive and his comments starting having a whirlwind effect. I guess Cory found comments about him on his YouTube page.

     
    Last edited:
    OP
    OP
    Mecha Meister

    Mecha Meister

    Next-Gen Guru
    Member
    Oct 25, 2017
    2,800
    United Kingdom

    Deleted member 1589

    User requested account closure
    Banned
    Oct 25, 2017
    8,576
    Right now it feels like calculus, where you know the result but your method to get to it is based on

    'maybe this is the formula because it gets me to the result on the answer sheet'.
     
    Hybrid Rainbow

    Hybrid Rainbow

    Banned
    Jun 23, 2019
    6,446
    TheThreadsThatBindUs said:
    Then if wider and slower isn't possible, how is anyone thinking MS is reaching 12TFLOPs?

    My personal, position is that 50+CUs reasonably clocked is the way to go, if they want high end performance. <400mmsq APU dies would mean a larger APU cost versus the previous gen., however, less RAM chips (PS4 use day 16) and a completely custom SSD would mean they can claw back some cost, and the lower clocked GPU would keep the cooling and chassis design under control.

    Better yet, HBM would further lower power consumption and PCB complexity, and from the recent details discussed by chris 1515 , it seems HBM2(E) isn't as costly as many want to believe.
    Click to expand...
    Click to shrink...

    But would HBM even be feasible especially since we got word that the next generation of chips from AMD were going to go GDDR6 instead of HBM due to the size? I know someone had posted a link about it earlier today.
     
    Lady Gaia

    Lady Gaia

    Member
    Oct 27, 2017
    2,476
    Seattle
    Mecha Meister said:
    The poll is now live!
    Click to expand...
    Click to shrink...

    I'm sorry to see my favorite and one that many others expressed appreciation for was dropped:

    |OT9| It's. A. Toy.

    Sad, as I committed to supporting it. I guess I can't in good conscience vote for anything else (not that I expect the next thread to amount to much - at this rate it'll be mostly sniping for another 400 pages without much in the way of concrete information.)
     
    AegonSnake

    AegonSnake

    Banned
    Oct 25, 2017
    9,566
    Hybrid Rainbow said:
    But would HBM even be feasible especially since we got word that the next generation of chips from AMD were going to go GDDR6 instead of HBM due to the size? I know someone had posted a link about it earlier today.
    Click to expand...
    Click to shrink...
    some are going to be gddr6, some hbm.

    apparently the biggest one is too big already for hbm to be stacked on top of it. dont ask me how any of this works. but thats what i read.

    the 40 cu navi 12 is going to have hbm for sure according to the same leaks. thats going to be the rtx 5700xt.
     
    Ra

    Ra

    Rap Genius
    Moderator
    Oct 27, 2017
    12,196
    Dark Space
    supraman1120 said:
    A controlled leak here is exactly what they should be doing. A runaway narrative (right or wrong) is damaging, and the fact that they know about the github "leak," they should be doing something IMO.
    Click to expand...
    Click to shrink...
    How is it at all damaging when only a handful of people on a few forums even know about it?

    The real world doesn't know or care about some random teraflop calculations.
     
    Reader mode
    Status
    Not open for further replies.