• 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.

When will the first 'next gen' console be revealed?

  • First half of 2019

    Votes: 593 15.6%
  • Second half of 2019(let's say post E3)

    Votes: 1,361 35.9%
  • First half of 2020

    Votes: 1,675 44.2%
  • 2021 :^)

    Votes: 161 4.2%

  • Total voters
    3,790
  • Poll closed .
Status
Not open for further replies.

Deleted member 12635

User requested account closure
Banned
Oct 27, 2017
6,198
Germany
Price wise 16GB of ram would be cheaper than it was in the PS4 for total cost if system is $500.
So it would be intresting to see where the spend the extra money or it for the SSD.
I hoping for 24 or at 16 with another set for the OS .
The SOC will cost more even they achieve it to not increase its size. So some of the cost savings on RAM are already eaten there. 16gb also gives you less complexity on your PCB while you can achieve high enough bandwidth for a 10tf gpu with 8 memory chips.
 

Deleted member 12635

User requested account closure
Banned
Oct 27, 2017
6,198
Germany
The conversation started in trying to extrapolate what GPU will be likely based on the rumoured Navi cards power outputs.
I don't really want something more then 200w.
It was more of a joke. Thought the emoji would tell this.

But seriously don't expect power draws beyond 150W unless a platform holder wants to spend extra money on cooling.
 

M3rcy

Member
Oct 27, 2017
702
You know, that "leak" of a 8GB HBM2/16GB DDR4 memory setup would perfectly fit an early dev kit that was running a Vega 56/64 in a PC.
 

tapedeck

Member
Oct 28, 2017
8,005
For those that know a lot more about console SOCs than myself..what's generally considered the max 'acceptable' wattage for a home console? Does it change based on the cooling method?
 

anexanhume

Member
Oct 25, 2017
12,918
Maryland
That one post at reddit by a throwaway account that PS5 uses HBM + DDR4 after the Wired article is either true or made by a madman with shit ton of time on his hands.

As someone stated here earlier, Sony has patented a heatsink last year to work on a stacked die on top of an APU, which is how HBM chips would be at. The 1.6 HBM chips could very well be similar to the low cost HBM chips Samsung has been developing and revealed in 2016 that a GPU made for laptops could be used;



Samsung-Low-Cost-HBM.png


It's not like AMD doesn't have experience using HBM2 chips on their GPUs either.

Although, the heatsink could be from a PS5 prototype that won't see the light of day, as well as Cerny has been championing unified RAM so...

The reddit post actually went up shortly before the Wired article.

Another interesting point to make regarding this rumor is that TSMC actually consulted with Sony (among others) during the development of InFO_MS, which would make Sony familiar with it and probably in the back of their mind when developing consoles. I'd also like to point out that Sony is not a stranger to being inventive with memory. The Vita had a special packaging method in use to get very high bandwidth to the Vita's GPU.

Also, let's assume for a second there was a typo in the original rumor, because it doesn't quite make sense as it is. It says:

allowed them to go below ~50 GFLOPs per GB/sec. bandwidth but still keep above 40 GFLOPs per GB/sec.

Which doesn't make sense. Typically when we talk about GPUs, we talk about GB/s per GF/TF, not the other way around. However, 40GB/s per GF would be a stupidly low number, so let's now assume they meant TF. We need a second part of the comment for future information

InFO_MS allows them to drive their 1.6 Gbps chips @ 1.7 Gbps (435 GB/sec.) without having to increase the voltage above 1.2v

1.6Gbps per second would result in 409.6GB/s in a 2-stack config. 1.7Gbps bumps it to 435.2GB/s, which is competitive with 256-bit GDDR6 solutions. If we're now assuming the 40-50 GB/s per TF is valid, this gives us a range of 8.7TF to 10.875TF for PS5.

Finally, regarding my skepticism around HBM2 supply, some key things have happened since Samsung's comments about low capacity. SK Hynix and Micron have both entered the market in full force (after the latter abandoned HMC development). And the crypto market crashed. With DRAM and NAND markets easing up, that capacity has to shift somewhere.

Regarding HBM pricing, it's hard to know much it has eased over the past few years (we do have some Vega VII rumored costs for reference), but I think it may be possible to get it down to less than 50% more than GDDR6 per GB, perhaps even just 35% to 40% higher. When you consider that they just need 8GB instead of 16GB of GDDR6, their solution is extremely cost competitive. At that point, it becomes a lot more attractive. HBM is also done on contract pricing (i.e. not floating with market costs), so a big order from Sony locks that factor in and sets up a mutually beneficial relationship with that partner to help them build up their own capacity.

The only rumor around this giving me pause is the digitimes rumor that stated ASE will do the packaging. Other than that, a lot of this rumor makes sense the more I dig into it.

Also, I imagine if that PS4 rev mentioned is coming, it's definitely this Fall. Since it's a console rev, it may not get cracked open to confirm the 7nm EUV from Samsung part, but the timing makes so much sense with MS pushing costs down with the SAD model and the rumored E-revision of the device internals.

Finally, here's the rumor in its entirety for posterity:

PS4 refresh
  • sometime between september and november
  • 199
  • fabbed on samsung 7nm EUV
  • best wafer pricing in the industry
  • die size 110mm²
  • no PRO refresh, financially not viable yet
  • too close to PS5 as well
PS5 memory and storage systems
  • 24 GB RAM in total (20 GB usable by games)
  • 8 GB in form of 2 * 4-Hi stacks HBM2
  • Sony got "amazing" deal for HBM
  • in part due to them buying up bad chips from other customers which can't run higher then 1.6 Gbps while keeping 1.2v.
  • HBM is expected to scale down in price a lot more than GDDR6 over the console lifetime
  • Samsung, Micron and SK Hynix already shifting part of their capacity towards HBM due to falling NAND prices
  • Sony will be one of the first high volume customers of TSMCs InFO_MS when mass production starts later this year (normal InFo already used by Apple in their iPhone)
  • InFO_MS brings down the cost compared to traditional silicon interposers - has thermal and performance advantage as well
  • InFO_MS allows them to drive their 1.6 Gbps chips @ 1.7 Gbps (435 GB/sec.) without having to increase the voltage above 1.2v
  • HBM is more power efficient compared to GDDR6 - the savings were invested into more GPU power
  • additional 16 GB in form of DDR4 @ 256 bit for 102.4 GB/sec.
  • 4 GB reserved for OS, the remaining 12 GB usable by games
  • memory automatically managed by HBCC and appears as 20 GB to the developers
  • HBCC manages streaming of game data from storage as well
  • developers can use the API to take control if they choose and manage the memory and storage streaming themselves
  • memory solution alleviates problems found in PS4
  • namely that CPU bandwidth reduces GPU bandwidth disproportionately
  • 2 stacks of HBM have 512 banks (more banks = fewer conflicts and higher utilization)
  • GDDR6 better than GDDR5 and GDDR5x in that regard but still less banks than HBM
  • at the same time trying to keep CPU memory access to slower DDR4
  • very satisfied with decision to use two kinds of memory for price to performance reasons
  • allowed them to go below ~50 GFLOPs per GB/sec. bandwidth but still keep above 40 GFLOPs per GB/sec.
 

M3rcy

Member
Oct 27, 2017
702
The HBCC has been around since Vega but for current-gen games 8GB VRAM hasn't generally been a limitation and the feature really hasn't had much of an impact.

In a console, it can give more effective memory capacity, though, since the HBCC is also flexible enough to work with virtual memory on a storage device at the expense of overall available bandwidth (since you will be using some of that bandwidth to shuffle data in and out of your "fast" memory).

This is, I think, what will power the "software stack" that Cerny refers to in the Wired interview.

When I was advocating for not only an SSD in the next-gen consoles, but also one that was as fast as possible, this is why. Hell, I first brought it up back in January.
 

Pheonix

Banned
Dec 14, 2018
5,990
St Kitts
Indeed.

Funny you say that because guess who featured SSDs in his console predictions? For some components there is a rationale, for others there isn't!
Don't get me wrong, I personally see them going with 16GB of GDDR6 coupled with 4GB of DDR4 for the OS. I don't even think they need more than 16GB of RAM exclusively for the games especially if they are using a gen 4 NVMe Drive.
 

anexanhume

Member
Oct 25, 2017
12,918
Maryland
Providing not only the specs, but the root causes for them (and for those to make sense) is compelling.
Agreed. I only wish the ASE rumor didn't conflict. Digitimes could be wrong there, but that would invalidate the whole thing IMO.

Edit: for whatever my opinion is worth, only OTs are restricted to the 400 page limit. I say we keep this thread alive until Computex.
 
Last edited:

Pheonix

Banned
Dec 14, 2018
5,990
St Kitts
The reddit post actually went up shortly before the Wired article.

Another interesting point to make regarding this rumor is that TSMC actually consulted with Sony (among others) during the development of InFO_MS, which would make Sony familiar with it and probably in the back of their mind when developing consoles. I'd also like to point out that Sony is not a stranger to being inventive with memory. The Vita had a special packaging method in use to get very high bandwidth to the Vita's GPU.

Also, let's assume for a second there was a typo in the original rumor, because it doesn't quite make sense as it is. It says:



Which doesn't make sense. Typically when we talk about GPUs, we talk about GB/s per GF/TF, not the other way around. However, 40GB/s per GF would be a stupidly low number, so let's now assume they meant TF. We need a second part of the comment for future information



1.6Gbps per second would result in 409.6GB/s in a 2-stack config. 1.7Gbps bumps it to 435.2GB/s, which is competitive with 256-bit GDDR6 solutions. If we're now assuming the 40-50 GB/s per TF is valid, this gives us a range of 8.7TF to 10.875TF for PS5.

Finally, regarding my skepticism around HBM2 supply, some key things have happened since Samsung's comments about low capacity. SK Hynix and Micron have both entered the market in full force (after the latter abandoned HMC development). And the crypto market crashed. With DRAM and NAND markets easing up, that capacity has to shift somewhere.

Regarding HBM pricing, it's hard to know much it has eased over the past few years (we do have some Vega VII rumored costs for reference), but I think it may be possible to get it down to less than 50% more than GDDR6 per GB, perhaps even just 35% to 40% higher. When you consider that they just need 8GB instead of 16GB of GDDR6, their solution is extremely cost competitive. At that point, it becomes a lot more attractive. HBM is also done on contract pricing (i.e. not floating with market costs), so a big order from Sony locks that factor in and sets up a mutually beneficial relationship with that partner to help them build up their own capacity.

The only rumor around this giving me pause is the digitimes rumor that stated ASE will do the packaging. Other than that, a lot of this rumor makes sense the more I dig into it.

Also, I imagine if that PS4 rev mentioned is coming, it's definitely this Fall. Since it's a console rev, it may not get cracked open to confirm the 7nm EUV from Samsung part, but the timing makes so much sense with MS pushing costs down with the SAD model and the rumored E-revision of the device internals.

Finally, here's the rumor in its entirety for posterity:
I am really loving the whole HBM thing.

Just don't wanna get my hopes too high. But this rumor seems very very very specific.
 

Hudsoniscool

Banned
Jun 5, 2018
1,495
But people don't like to buy stuff that feels "obsolete".

Even if they still own a 1080 TV, they would rather buy a 4K console for when they upgrade the TV, even if they don't end up upgrading it anytime soon.

Specially 2 years from now, when the talk will shift to 8K, how are they gonna market a 1080 sku??

I still don't believe it will be 4TF tbh
im sorry but this post won't age well. The vast majority of people don't care about 4K. Parents buying their 10 year old and such a Christmas present don't care if it does 4K or not.

And In 2020 the talk will not switch over to 8k. 8k tv adoption won't even hit 10% for another 10 years.
 

anexanhume

Member
Oct 25, 2017
12,918
Maryland
Anex,can you explain for the rest of us this ASE thing,please?

Sure.

ASE is a company that is known for their packaging services. For a long time, they were the dominant provider.

So after you fabricate your die, you have to package it somehow so that you can put it on an actual PCB. Up until the past decade, packaging was very boring and it was essentially just another PCB in between the die and actual PCB.

Then, people started putting memory on top of processors, embedding passive components in the package, and even building up the package around the wafer rather than a discrete package mated after the fact.

InFO (integrated fan-out) burst onto the scene when Apple popularized it. It offered thermal and performance advantages in the total solution. This was a TSMC invention.

Since then, TSMC has spawned many iterations of InFO. InFO_MS specifically targets packages with memory on them such as HBM.

TSMC is now regarded as a leader in the packaging world, and ASE has no equivalent to InFO_MS. If it is true that ASE is doing the packaging, this rumor is DOA, IMO.

I am really loving the whole HBM thing.

Just don't wanna get my hopes too high. But this rumor seems very very very specific.

It only got more specific once I looked into it and extrapolated a bit on the info I had.
 

VX1

Member
Oct 28, 2017
7,007
Europe
Sure.

ASE is a company that is known for their packaging services. For a long time, they were the dominant provider.

So after you fabricate your die, you have to package it somehow so that you can put it on an actual PCB. Up until the past decade, packaging was very boring and it was essentially just another PCB in between the die and actual PCB.

Then, people started putting memory on top of processors, embedding passive components in the package, and even building up the package around the wafer rather than a discrete package mated after the fact.

InFO (integrated fan-out) burst onto the scene when Apple popularized it. It offered thermal and performance advantages in the total solution. This was a TSMC invention.

Since then, TSMC has spawned many iterations of InFO. InFO_MS specifically targets packages with memory on them such as HBM.

TSMC is now regarded as a leader in the packaging world, and ASE has no equivalent to InFO_MS. If it is true that ASE is doing the packaging, this rumor is DOA, IMO.



It only got more specific once I looked into it and extrapolated a bit on the info I had.

Thanks man,appreciate it!
 

gofreak

Member
Oct 26, 2017
7,819
Worth a double post. New DF hotness.



It got interesting when he pointed to the PS4 chips that seemingly popped up in a similar vein.

One thing I'd suggest is that maybe these are chips appearing in dev kits...and not necessarily retail product. (Even if, obviously, a final retail chip might be quite similar).
 

Lagspike_exe

Banned
Dec 15, 2017
1,974
Sure.

ASE is a company that is known for their packaging services. For a long time, they were the dominant provider.

So after you fabricate your die, you have to package it somehow so that you can put it on an actual PCB. Up until the past decade, packaging was very boring and it was essentially just another PCB in between the die and actual PCB.

Then, people started putting memory on top of processors, embedding passive components in the package, and even building up the package around the wafer rather than a discrete package mated after the fact.

InFO (integrated fan-out) burst onto the scene when Apple popularized it. It offered thermal and performance advantages in the total solution. This was a TSMC invention.

Since then, TSMC has spawned many iterations of InFO. InFO_MS specifically targets packages with memory on them such as HBM.

TSMC is now regarded as a leader in the packaging world, and ASE has no equivalent to InFO_MS. If it is true that ASE is doing the packaging, this rumor is DOA, IMO.



It only got more specific once I looked into it and extrapolated a bit on the info I had.

Awesome stuff, thanks!
 

M3rcy

Member
Oct 27, 2017
702
It only got more specific once I looked into it and extrapolated a bit on the info I had.

Even the TFlop number being slightly lower than what might be expected makes sense if you have to fit two memory interfaces onto the chip instead of just one since you'd have to remove something (some CUs in this case) to keep the die size in the target range.
 

Deleted member 22750

Oct 28, 2017
13,267
Are devkits final?

What if there are problems that make development a problem or overheating issues / bad design
 
Last edited:

anexanhume

Member
Oct 25, 2017
12,918
Maryland
Even the TFlop number being slightly lower than what might be expected makes sense if you have to fit two memory interfaces onto the chip instead of just one since you'd have to remove something (some CUs in this case) to keep the die size in the target range.
Not sure I follow you. The video I posted a week or so ago regarding HBM2 vs. GDDR6 actually suggested that HBM2 controllers were smaller than those used for GDDR6 for comparable bandwidths. I can understand needing to cut down die size for yield/cost targets, but a higher CU count was never realistic in that case anyway (unless of course they were going to do an IO die all along).
 

foamdino

Banned
Oct 28, 2017
491
Just a thought - perhaps MS are going completely discless with Lockhart & Anaconda - which would give them some more $$ to play with, hence their confidence in being more powerful
 

Deleted member 1589

User requested account closure
Banned
Oct 25, 2017
8,576
So basically if that HBM rumor is correct (which is doubtful since Digitimes is a much, much better source of news than a one off reddit account) we might look at a 10 tflops system?
 

M3rcy

Member
Oct 27, 2017
702
Not sure I follow you. The video I posted a week or so ago regarding HBM2 vs. GDDR6 actually suggested that HBM2 controllers were smaller than those used for GDDR6 for comparable bandwidths. I can understand needing to cut down die size for yield/cost targets, but a higher CU count was never realistic in that case anyway (unless of course they were going to do an IO die all along).

I'm talking about having only GDDR6 interfaces vs. having both HBM2 and DDR4 interfaces on the chip.
 

BitsandBytes

Member
Dec 16, 2017
4,576
Nice to see Rich firmly in the 10-12TF camp now.

Anyone know what date the PS4/Pro 3D Mark scores where found or posted about by APISAK?
 

Pheonix

Banned
Dec 14, 2018
5,990
St Kitts
It only got more specific once I looked into it and extrapolated a bit on the info I had.

Lol...makes sense now....
It got interesting when he pointed to the PS4 chips that seemingly popped up in a similar vein.

One thing I'd suggest is that maybe these are chips appearing in dev kits...and not necessarily retail product. (Even if, obviously, a final retail chip might be quite similar).
I strongly doubt that engineering o qualification sample chips are what makes it into early dev kits. Early dev kits are basically PCs.

It didn't just get interesting, it got flat out real. Only thing now is that we don't know if thats actually a PS5 APU, it could be for the next Xbox. And we don't know how many CUs its being clocked against.
 

nib95

Contains No Misinformation on Philly Cheesesteaks
Banned
Oct 28, 2017
18,498
Just a thought - perhaps MS are going completely discless with Lockhart & Anaconda - which would give them some more $$ to play with, hence their confidence in being more powerful

I very much doubt that. They'd be shooting themselves in the foot just to save what, $40?, If that given the console wouldn't launch for another year to two.
 
Status
Not open for further replies.