Oh i agree,but what would MS do if they find out that Sony will definitely launch late 2019? ( as i expect)
Launch Next Xbox just 2 years after XB1X ?
-
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 and next Xbox launch speculation - timing and specification
- Thread starter zomgbbqftw
- Start date
- Status
Oh i agree,but what would MS do if they find out that Sony will definitely launch late 2019? ( as i expect)
Launch Next Xbox just 2 years after XB1X ?
HBM debuted in 2015 on Fury X. It was expensive then. It was expensive when this article debuted. It's expensive now. LCHBM was proposed as a direct response to the fact it's still expensive. What makes you think things are going to change in the next two years? Do you see a bunch of new interposer-based products sprouting up all over?
HBM needs a momentous shift to become viable for consoles. Organic interposers would be one such thing, but there's no guarantee they'll reach the density needed any time soon.
All three major vendors are players in the HBM market:
https://www.prnewswire.com/news-rel...-amd-are-dominating-the-market-300612168.html
Probably the same as what sony would do if they found out MS were launching nextbox in fall 2019 and try and match the date.
Also regarding RAM, if one went with HBM and the other went with GDDR6.
The HBM option could have less overall RAM to mitigate the higher cost of HBM.
They could have two options in there desired price bracket
20gb HBM @1000gbps
Or
28gb GDDR6 @600gbps
Im sure each option would have its pros and cons.
Last edited:
I see,thanks.Anyway,HBM would be a huge surprise in any future console!
Yes,but whole new console after only 2 years?That would be unprecedented,i think.
What happened to low cost HBM that was announced to be in development by SK Hynix and Samsung? https://www.tweaktown.com/news/5353....it&utm_medium=twitter&utm_campaign=tweaktown
I think its a preferable situation for MS to have a competing console then to not have one, for MS and its consumers.
And sony would be launching a console 3yrs later, is 2yrs significant and 3yr is not?
Also remember both the the PRO and X are mid gen upgrades both part of PS4 and Xbox one families.
Like I said, it won't be in scope until holiday 2018.
This is basically the same argument people were making about gddr5 in the Ps4 and how it would never happen because of price and yield.
Then it was 4 gigs max because price and yield.
I'm completely understanding of your side of this theory believe me. I'm just deducing based on the HBM based cards coming out end of this year, AMD pushing HBM tech, and the timing of the new consoles end of 2019 or 20.
I think it's going to be a very similar surprise to the 8 gigs of gddr5 this gen.
If this happens, The arguments about power will be based on that 50% bandwidth advantage, and possibly speed advantages of gddr6.
Exciting stuff.
The primary detractor on GDDR5 was cost.
This situation is not comparable. There was no reasonable alternative to GDDR5 for that kind of bandwidth at the time. HBM2/3 have a completely viable alternative in GDDR6.
No one has demonstrated the volume or cost structures to suggest HBM will be viable for consoles any time soon. Waving your hands and stating vague things like "AMD is pushing HBM" are merely hopeful sentiments.
HBM has been featured in consumer cards with very low volume and research focused cards with very high mark-ups.
Microsoft explicitly stated they looked at HBM for X and gave reasons for rejecting it that weren't just yield and cost (testability and granularity). It's not a panacea technology by any means. Their own job posting focused on GDDR6, not HBM in their engineer search.
It should speak volumes that the vast majority of consumer GPU cards coming down the pipe are expected to be GDDR6, not HBM. These are cards that retail up to $600 in lower volume than consoles and they're still sticking with GDDR6.
Last edited:
I mean, it's the same argument. People were arguing price and yield, absolutely. Even after Sony revealed it was gddr5.
There absolutely was a viable alternative to gddr5, their competitor used it. DDR3. Cheaper, better yield, people expected it.
I'm saying HBM tech is going to be as viable as gddr6 tech by the end of 2019 or 20. Maybe not as inexpensive, but definitely viable.
I think you're being a bit close minded about the subject tbh.
5 years? We can barely hit 30fps at 1080p and AI is still terrible. 2020 would be perfect
I've already outlined all of the challenges HBM faces, many of which you've ignored.
DDR3 was not a viable alternative in terms of bandwidth. GDDR6 is compared to all but the most aggressive (and expensive) HBM implementations.
You're being dismissive and not willing to engage with the realities of the situation. I don't know what else to tell you.
Would it be worth to have less ram at HBM rather than more at gddr6? You would still lose that ram
DDR3 wasnt a viable replacement for GDDR5, MS had to add some ESRAM and even then it wasn't a match.
I'm not ignoring the challenges you've outlined, far from it. I've been following HBM sine 2015. I know the challenges quite well. I just think 5 years after its debut is going to be a situation much different than you paint here.
Much of the most recent information on those challenges was discussed mid 2017. By holiday 2018, there's going to be a more realistic view. This is what I'm saying.
I'm not saying it was a viable replacement or even as good, I'm saying it was a viable alternative. Viable enough that Sony's largest competitor used it in their launch consoles.
Gddr5 was a much better bet than ddr3 with esram. Yes.
Only two, Samsung and SK Hynix.
Micron doesn't produce HBM, they are stupidly holding onto their HMC specification which nobody uses and for whatever reason they don't establish an HBM production line.
There is an japanese articel from pc.watch.impress which covers the situation.
Essence of it, DRAM prices went up, most costumers are high performance clients and the priority for 2-Hi stacks and a low-cost HBM specification went out the window, focusing more on high margin products, with more capacity and speed.
https://translate.google.de/transla...jp/docs/column/kaigai/1112390.html&edit-text=
https://pc.watch.impress.co.jp/docs/column/kaigai/1112390.html
Thanks for the correction. The PR piece I linked unfortunately didn't distinguish between the two upon re-read.
8/23/2016
https://arstechnica.com/gadgets/2016/08/hbm3-details-price-bandwidth/
It would not completely surprise me if one or both of the main console makers used HBM3.
https://arstechnica.com/gadgets/2016/08/hbm3-details-price-bandwidth/
It would not completely surprise me if one or both of the main console makers used HBM3.
I honestly think that HBCC and a reasonably sized NAND cache obviates the need for HBM or even the larger bus and higher GDDR6 quantities (e.g. 24GB).
With the HBCC functionality, they could probably get away with a pool of only 16GB GDDR6. And as long as they can clock it high enough for a reasonable amount of bandwidth (400 - 500+ GB/s), with the lower RAM capacity they're at least not going to have too much issues with having sufficient memory bandwidth to fill it.
A 256 bit bus and sub-16Gb/s GDDR6 chips will likely mean the system runs a little tight on bandwidth for gaming GPU workloads, but it won't be the end of the world, since it very much depends on the size and performance of the GPU and what new bandwidth-saving GPU features AMD includes in the Navi architecture.
With the HBCC functionality, they could probably get away with a pool of only 16GB GDDR6. And as long as they can clock it high enough for a reasonable amount of bandwidth (400 - 500+ GB/s), with the lower RAM capacity they're at least not going to have too much issues with having sufficient memory bandwidth to fill it.
A 256 bit bus and sub-16Gb/s GDDR6 chips will likely mean the system runs a little tight on bandwidth for gaming GPU workloads, but it won't be the end of the world, since it very much depends on the size and performance of the GPU and what new bandwidth-saving GPU features AMD includes in the Navi architecture.
I would argue that Sony willing to commit to sticking HBM in ~100 million consoles could be enough to bring about that momentous shift.
Well our only hope for hbm is ps5 as microsoft is straight up hiring for a lead on their gddr6 team
GDDR6 is only one of the RAM technologies they're looking to hire for. That job listing was basically covering everything. They can use HBM if they really want to, if it's actually viable from a cost standpoint.
HBM isn't realistic for either company anyway.
I was thinking that too, especially if MS jumps in too.
See, not really.
You keep talking about standards, and specs.
I'm talking about the entire TV manufacturing industry pivoting away from supporting 3D media in the home. And the unlikelyhood that they will pivot back. And without displays, there's no market for content.
I bought one of the few 4k passive 3D TVs that can display full 1080 per eye. That spec increase, and improvement in user experience wasn't enough to keep 3D support going. In fact most 4k 3D tvs were still using active shutter glasses.
This is reasonable, but I'm curious what Microsoft might be thinking.
The accepted wisdom seems to be that a) they will launch at the same time as Sony and b) they will aim to out-power PS5, if possible, which, sans a significant price differential, seem mutually exclusive, unless they can produce a superior design for the same price. This is possible for either company, but ultimately it comes down to chance which is better, all else being broadly equal.
So if the current wisdom is if Sony hit 2019, so will MS, or if Sony hit 2020, so will MS (to avoid a debilitating difference in install base), how can MS be sure what Sony will do? My point here is that MS may be able to be fairly sure of Sony's launch plan, but if they aren't, and if they want to avoid being a year late, surely they have to aim for 2019?
This might actually be a better strategy for them - if Sony hit 2019 then MS are day and date. If not, they have 6-12 months in the NG market alone. Certainly this might give Sony a later power advantage, but that's not certain and may not be all that significant. I don't think we should discount that MS may attempt to be disruptive and chase the success they had with the 360 approach, and I'm not convinced a new Xbox 2 years after the X1X is all that damaging if the attraction of next generation games is sufficient to pull in the customers. Ultimately the X1X is a souped-up Xbox One and is unlikely to ever play NG games, so if you want NG games, you need NG Xbox. It's a PR problem, but certainly not insurmountable.
If MS did get out ahead of PS5, they get all the excitement of NG and a user-base of millions before Sony are out of the gate. The down-side is a potential power differential, but that isn't necessarily a huge problem. I don't think this is a likely scenario, but it is possible.
I just think we can't assume that MS can know when Sony plan to launch (or vice-versa), or that they are utterly wedded to we-must-outperform-the-PS5-come hell-or-high-water. Doing the disruptive, unexpected thing might be what they need to do.
At present I expect both PS5 and Xbox NG in 2020, or both in 2019, but perhaps we shouldn't be so convinced that either company can be certain what their competitor is planning, or that alternate approaches are untenable.
Sony I believe, will have no issues with MS releasing 6 month or so before them. This is because I believe the critical thing is to have the release day exclusives ready. That the best chance for either side is not to release first, but to have the first party games blowing the socks off people. Moving the release of the console forwards has the detrimental effect of shortening the development of any first party games they got planned.
My opinion is that the next gen would be the battle of 1st party exclusives. (Unless Sony screwed up and didn't have backwards compatibility for PS4 games.)
I am also assuming for the first year of release, that third parties would just make current gen games with mid-gen graphic upgrades for the new consoles, until the ecosystem of the new gen get large enough.
Don't agree really. 1 or 2 exclusives for launch plus the next gen version of all the 3rd party games will be plenty for either system. All the service games will probably get a port.
They need announced games for the first year alright.
Well my point is that both sides would try to do this. But it isn't like anyone TRY to make bad games; bad games just happen, just like no one want to make bad movies. Clearly, 2 blockbuster 1st party exclusives would be enough to sell a console, Nintendo proved that enough for the Switch. But to make sure the game actually delivers, my argument is that getting the game finished on time might be a higher priority than getting one's new console out before the competitor does.
That having your console out early just isn't worth the costs, if it means half-baked first party games.
Why would they be half baked? You're going to have cheap easy to make cross gen games, that alone is going to make the transition easier.
You'll have a couple of 3rd party and 1st party exclusives. At least a few will be good. Would it be great to have more? Sure. But they won't need more for a successful launch. The PS5 at least will be sold out for months all the while more games are releasing.
There's been plenty of people arguing in this thread on why it's better to hold the big guns off until there's more consoles under tellys.
Nobody will care about Pro after PS5 launch... first party games should be made only for PS5.
I don't see why not. Even if it's just 8gb of slower ram for system functions leaving the full 16 of the faster available just for games.
Well if they go with 24gb of fast RAM and then reserve 8 for system functions isn't that the same thing?
It's 4gb in a PS4 pro I believe. 3.5 in the base model. Don't know the split on Xbox.
What were the 360 and PS3 rocking? Nothing even remotely close anyway. 8 doesn't seem too far off.
I don't know if they'll need as much as 8GB, but OS needs is one of the reasons I'm confident we're not going to end up with only 16GB like a few people keep saying.
They will want to add a bunch of features and improve many existing ones, and they'll need a lot of memory to have it all running and available all the time.
I'm going off memory here, so I'm likely to be wrong: Xbox One was around 3GB at launch. Xbox 360 was 32MB, I think for most of its lifetime? And the PS3 went down to I think 96MB after some updates, don't know if it used less as time went on.
Try searching Netflix for 3D movies, you will find some made in China 3D movies that are extremely well done. Slow pans and great content.
Key here is your 4K TV can only support HD 3D because the cost and power needed to support UHD 3D, at the present time, requires memory like GDDR6, a more advanced HEVC codec that takes more processing power and lots of GPGPU for the depthmap so they skipped supporting UHD 3D until the hardware industry can support it. 3D is on hold till 8K TVs are released in 2020. This is also when many are speculating a cheaper more powerful PS5 could ship. CES 2018 had a panel discussing VR/360 publishing of UHD media standards which means the media is coming 1-2 years later.
HEVC Profiles for 8 bit and 10 bit which consoles that have a UHD Blu-ray (XB1 S and X) or do not have a HEVC hardware codec (PS4) (they use a GPU or CPU and software firmware updated AFTER the console is released) can use the HEVC codec free of cost. Two years after the release of HEVC for UHD Blu-ray, we have HEVC multi-view plus depthmap. Not only can this codec and UHD standard support lossless 3D but it can support 3D views from multiple angles. HD frame packed 3D required converting the two 1080P images to what is essentially the UHD standard ( first frame and depth map) and then generate the multiple views but in this case it's not lossless. Costs for the use of the HEVC codec depend on the Profile used, the most expensive being UHD Multi-view plus depthmap.
UHD 3D is part of every UHD delivery standard except UHD Blu-ray but the drive speed for UHD Blu-ray can support UHD 3D by design. UHD 3D/8K Blu-ray is coming so 3D is not dead. In addition VR goggles can support UHD 3D and I expect all 8K TVs to support some form of glassless 3D. Free use of the HEVC Multi-view plus Depthmap codec will require a UHD/8K Blu-ray player in the game console and then UHD 3D media will be supported for VR and TV via a Blu-ray player or IPTV.
Last edited:
Most OS features require memory in the MBs. The largest consumers are app multitasking and of course the high resolution game streaming and recording functions. Recording game footage at 4k60fps will require a bigger OS footprint than the current PS4 OS, yes, but not THAT much bigger, since they won't be recording the footage at the highest possible bitrate (since you'd run out of storage very very quickly).
Thinking about it more, doesn't the XB1X already allow for this? And if so I'm pretty such the OS footprint there is nowhere near 8GB.
There's also the option of adding in some additional cheap DDR4 or LPDDR4 RAM like in the Pro, for the OS. The OS gains nothing by using the expensive, fast GDDR6 RAM, so rather than waste a precious resource like that, it's better to engineer a separate RAM pool for the OS functionality.... it's not like there isn't a precedent for it (both PS4 versions had an additional RAM pool for the OS to differing extents).
ATSC 3.0 will be supported by both Microsoft and Sony consoles and Emergency Alert support is required. That requires Always on Always Connected internet and instant on power support. This, for power reasons, requires a ARM subsystem with it's own low power RAM. Just increase the size of the cheap Low Power ram that can also be used for background game recording or DVR but maybe not both at the same time?
This. 16GB of really fast GDDR6 plus 8GB of LPDDR4. The OS/streaming storage/media apps/possible future features can use the LPDDR4 for the most part, with 1GB of GDDR6 being used for the parts of the OS that really need to be on the main RAM. End result, we go from 5GB of RAM for games on PS4 to 15 GB of RAM for games on PS5, and that's before we take into account the greater RAM efficiency from HBCC. It would probably work fine, though 24GB on a larger bus would still be even better.
- Status