Next Generation Hardware Speculation with a Technical Spin [post E3 2019, pre GDC 2020] [XBSX, PS5]
- Thread starter DavidGraham
- Start date
-
- Tags
- john sell
- Status
The center fluid volumn is 1.5 cats.
Microsoft rewrote the rules... something... something... something...
Rewrite rules, and i thought Sony was pro at hyping.
Then look at mobile that went to the complete extreme with software, when any asking price was too much, but players will pay hundreds of dollars for in-game consumables. People, aye?
Silent_Buddha
Legend
If they go as bold as MS in terms of size, they might as well make it an 8" cube.
I did like how the NeXT cube looked back in the 80's.
Then we can have Cube versus Tower wars.
Regards,
SB
These kits tend to be much larger than normal; at least based on the history of PS kits. I’m not sure if it’s (anywhere close) reflective of the final product.
But then again; I’ve been wrong on everything so far so perhaps this is it!
I do trust Sony on the design of the PS5.
They always kinda push the launch with a premium not too plasticky design.
I mean they do great jobs with some of their mobiles and tv designs.
Design and size is also important for the Japanese market as some people in crowded cities live in tiny apartments. So it's imperative for them to make a piece of hardware not too bulky.
The flat design was nice you could either of the console under something or between shelves.
Or slide it vertically in between devices like in a bookshelf.
As for the noise i hope they will find a solution as to make it as a quiet as possible.
I still have all my original playstation, fat Ps2, fat ps3, ps3 is the only one which i had to buy a slim model because of the ylod.
And please Sony allow us to switch storage capacity and not make it too proprietary lol.
Or at least like the ps4 pro let us use an external hdd.
Maybe they will put a mini oled touchscreen on the front with on/off switch. I personally would have loved to see it lokking something like that. But only in my dreams.
They always kinda push the launch with a premium not too plasticky design.
I mean they do great jobs with some of their mobiles and tv designs.
Design and size is also important for the Japanese market as some people in crowded cities live in tiny apartments. So it's imperative for them to make a piece of hardware not too bulky.
The flat design was nice you could either of the console under something or between shelves.
Or slide it vertically in between devices like in a bookshelf.
As for the noise i hope they will find a solution as to make it as a quiet as possible.
I still have all my original playstation, fat Ps2, fat ps3, ps3 is the only one which i had to buy a slim model because of the ylod.
And please Sony allow us to switch storage capacity and not make it too proprietary lol.
Or at least like the ps4 pro let us use an external hdd.
Maybe they will put a mini oled touchscreen on the front with on/off switch. I personally would have loved to see it lokking something like that. But only in my dreams.
Last edited:
D
Deleted member 11852
Guest
As long as they put clearly visible power and eject buttons on the front I'll be happy. I also don't need a 6" LED strip to tell me the power/console state.
Your in luck, for next gen power aspirations LED's are out and lasers are in.
Edit: balls, technical thread . Something something blue laser wavelength :/
Tkumpathenurpahl
Veteran
Yup. In fact, I've got some exclusive footage of an XSX as it idles in standby:
Back with another of my twice yearly stupid questions.
Re: The chip shot from the Scarlett reveal. I am on record (I think here) in being skeptical of that shot being of Anaconda. Something about it just doesn't seem correct. I think the consensus landed around 350mm for that die. Just out of curiosity, has the possibility this is Lockhart been considered? I don't mean a designed Lockhart chip, but a "failed" Anaconda. There has been a rumor around for awhile, that just will not quite die, that one of the next-gen consoles is a MCM. The consensus seems to be that Lockhart is, at least in part, a way to take advantage of a "failed" Anaconda chip as has been discussed. However Lockhart would almost certainly not include ray-tracing. If the ray-tracing cores were on a separate die(s) then it might help explain at least some of the disparity between the CU counts (64- almost certainly a dev kit number if true at all) and the measured size of the chip. If you off-load all the RT tech to a separate die(s) and use a MCM for Anaconda, you would also save on the overall size of the APU. Thus higher yields (cheaper overall) and you wouldn't be tossing whatever amount of RT dedicated silicon that actually worked in a given "failed" Lockhart APU. Is this a good idea or am I missing something fundamental here?
As an aside to the above: My understanding is that RT likes very fast RAM. Any reason you couldn't have a dedicated pool of special RT RAM to help out? Maybe HBM, EDRAM or ESRAM, but only for the RT cores. Maybe a tech I am not considering? Just seemed a way to relieve the potential bandwidth pressure on the main pool of RAM and get RAM more suited to RT use.
Re: The chip shot from the Scarlett reveal. I am on record (I think here) in being skeptical of that shot being of Anaconda. Something about it just doesn't seem correct. I think the consensus landed around 350mm for that die. Just out of curiosity, has the possibility this is Lockhart been considered? I don't mean a designed Lockhart chip, but a "failed" Anaconda. There has been a rumor around for awhile, that just will not quite die, that one of the next-gen consoles is a MCM. The consensus seems to be that Lockhart is, at least in part, a way to take advantage of a "failed" Anaconda chip as has been discussed. However Lockhart would almost certainly not include ray-tracing. If the ray-tracing cores were on a separate die(s) then it might help explain at least some of the disparity between the CU counts (64- almost certainly a dev kit number if true at all) and the measured size of the chip. If you off-load all the RT tech to a separate die(s) and use a MCM for Anaconda, you would also save on the overall size of the APU. Thus higher yields (cheaper overall) and you wouldn't be tossing whatever amount of RT dedicated silicon that actually worked in a given "failed" Lockhart APU. Is this a good idea or am I missing something fundamental here?
As an aside to the above: My understanding is that RT likes very fast RAM. Any reason you couldn't have a dedicated pool of special RT RAM to help out? Maybe HBM, EDRAM or ESRAM, but only for the RT cores. Maybe a tech I am not considering? Just seemed a way to relieve the potential bandwidth pressure on the main pool of RAM and get RAM more suited to RT use.
Last edited:
I read a dev said 10 to 20 % compared to DX12, Vulkan PC but the biggest advantage is to work with an unique configuration.
Last edited:
1.386225x
I fail to see how Sony can pull a significant advantage in SSD solution against MS unless MS is unwilling to match Sony.
I have no doubt that Sony could really have invented a very fast controller and read optimized I/O and software stack for their SSD solution, but I do doubt that that kind of technique and technology is exclusive to Sony, or that a similar or equivalent solution doesn't exist from the many other SSD vendors. Sure Sony may have a patent for a really fast read-centric SSD, but surely other vendors especializing in SSD solutions must have a stack of patents Sony can only dream of.
Assuming MS doesn't have their own ultra-fast solution, they can always go to third party companies to collaborate or get help for their software stack and I/O. Again, unless MS is unwilling to match Sony.
MS's SSD shouldn't that far away in performance if MS is willing to match Sony's touted highly optimized and customized nand flash SSD. And I believe this to be the case. MS's and Sony's SSD solutions are not that far away from each other in performance.
Or is that Sony's software stack and I/O optimizations really that magical that other SSD companies, let alone MS, cannot have something of an equivalent?
Tkumpathenurpahl
Veteran
Maybe it's just that Microsoft are somewhat limited by their need to accommodate the PC ecosystem? Requiring a PCIE 4 SSD is a reasonable mandate for PC gamers to play Microsoft first party exclusives. Requiring a bespoke XBox motherboard - with, for example, soldered NAND - isn't as reasonable.
The software stack they need to do it themselve. The SSD in datacenter are much faster than customer SSD.
https://thessdguy.com/baidu-goes-beyond-ssds/
Here using a FGPA and a custom software stack, Baidu was able to multiply by 7 to 10 the write speed tailoring the software.
http://ranger.uta.edu/~sjiang/pubs/papers/ouyang14-SDF.pdf
They are able to reach sequential write speed but in application. They reach 95% of the raw bandwidth of the SSD.
PC software aren't optimized for SSD. Sony decided to create an optimized NAND flash for gaming. And they use a software stack wit han ARM CPU, ASIC decompressor and a custom controller for improve read speed and have a guaranteed latency as low as possible.
I would not be surprised to have a Sony SSD reaching 8 GB/s of uncompressed data on PCIE 3 or maybe a custom bus.
EDIT: Fast SSD are customized, here it is a mix of hardware and software.
Last edited:
Personally I think that if a 1P studio doesn't think a game is possible on pc then they should just release on Xbox and xcloud (once upgraded).
Simply say not compromising on Xbox hardware or software, once pc's catch up which it will then it will get those games.
I actually think that's a very strong message and says not abandoning PC.
- Status
Similar threads
- Locked
