These are the I/O co-processors that I'm talking about and thank you for the SRAM correction. Your right that Microsoft didn't give us a breakdown of their APU like Sony has. They could have some dedicated I/O hardware in it or maybe they don't have I/O hardware within the APU.
I really hope that Microsoft becomes a bit more transparent with their I/O system because that would clear alot of the confusion up.
I find this kind of selective..Whatever Louise said, i wouldn't take it seriously in any way ( especially when he/she is only XSX/W10 dev and also he/she follows these guys ) :
If you want an explanation of what that sentence means or is describing read up on Sparse Virtual Textures one of the oldest standardized approaches. You can also look up explanations to megatextures too....or maybe I was right? Probably not, but this reporting from guru3d had the same take:
"Sampler feedback solves this by allowing a shader to efficiently query what part of a texture would have been needed to satisfy a sampling request, without actually carrying out the sample operation."
AMD RDNA 2 and GeForce RTX Architecture to Support DirectX 12 Ultimate
AMD’s forthcoming graphics architecture that will power the future of PC and console gaming, will support the new DirectX 12 Ultimate API Microsoft announced today. DX Ultimate basically means suppo...www.guru3d.com
It doesn't and serves no purpose but as a means to ruin every discussion. Its hard enough to have any discussion already and adding another source of noise does not help.
Would also like to add that Microsoft would probably have to make sacrifices in the APU to have a better I/O solution. This could potentially lead them to having a narrower GPU which would decrease the flop count. Flops is alot easier to market than a superior I/O system plus it leads to other advantages such as increased resolution.
I don't think one is making dumb mistakes compared to the other it's just their goals for next gen are different.
Didn't Cerny mention in his presentation that RT is not a focus for the PS5?
They definitely have dedicated I/O hardware and a custom I/O unit. It's just that they involve one of the CPU cores wheras the PS5 everything will be hardware accelerated. On both systems, devs just have to update their I/O code for current gen games.
These are the I/O co-processors that I'm talking about and thank you for the SRAM correction. Your right that Microsoft didn't give us a breakdown of their APU like Sony has. They could have some dedicated I/O hardware in it or maybe they don't have I/O hardware within the APU.
I really hope that Microsoft becomes a bit more transparent with their I/O system because that would clear alot of the confusion up.
As it shouldn't be, since these consoles can only do very light RT. But if the HW supports it (which it is confirmed to do) they obviously are going to add it to the API, they aren't going to leave it completely out because it is not the main focus.
The Xbox Series X is also 'overperforming' in terms of IO rate thanks to the brand new DirectX extension, DirectStorage. According to Goossen, it would take a PC with 13 Zen 2 cores to match the full IO rate of the new console.
.... and we have just entered "fan" territory. seriously though, it was a valid argument. You aren't applying the same logic to whichever side you like.
My point was not that the hardware doesn't support RT. It most likely does. My point was that if Sony's focus is not RT, they might not add the function in API immediately if they have to finish other stuff first. It might as well be there though, because Kirby Louise has openly said she doesn't have a PS dev kit because it's too expensive for her to acquire. So she might be simply wrong on that one. Then again, the GI in UE5 can possibly be accelerated with RT hardware, and they didn't use it. So... Yeah. I don't know.
They didn't use because they wanted to show what Lumen can do without adding RT to the mix. It was in one of the interviews.
He definitely downplayed it. Which I found surprising.
After listening to it again, yeah, it was more in that direction. He did mention later that RT would be used primarily for audio and GI. Not exactly like that, but he implied it, because the others are too expensive.
Agreed.
Actually, to me it seems like she follows a lot more PC pages than anything else. Not that it matters that much to me. I don't look at people as being universally right or universally wrong.
thicc_girls_are_teh_best
informed me of a hot chips event in August where we can probably obtain a diagram of the APU or at least a breakdown of it. MS has DLSS for AMD... implemented in XSX HW. So... they are fully gunning for RT implementation this Generation. Probably to show in Halo infinite first.
It's why I said that Microsoft needs to be more transparent in that department and show us how their APU is built so we can make better comparisons.
They also have software that they use in their I/O system. Not everything is hardware accelerated.
If I were to make an educated guess.i would say that both companies are using hardware and software acceleration in their I/O systems.
Sadly, for many people, the focus of this thread is to imagine ways of minimizing the difference in SSD speed (magic multipliers and compression schemes that turn 4.8 into 7) and downplay anything PS5 related, either by assuming incompetence in the design that will lead to the SSD design/speed advantage being negated or by assuming Sony doesn't care or have time to even add RT instructions to the API.
The same as they are now or smaller.
I don't think "downplay" is the right term. He said it was there and you can use it if you want but there is a big cost for it.
first, RAM is 16GB so even if the console could instantly (and I use the word 'instantly' colloquially) transfer/stream some portion of SSD it could be no more than RAM capacity. second, if it is faster than the compressed speed, lets say 2x, for a special part of SSD than that part would be open to contention among many games in the library, which one gets to stay there, and what happens when you switch games? third, a partition within an SSD can not be software for it to work it really needs a hardware difference if there is an even speedier transfer than the compressed value. So overall that instantaneous seem a lot like marketing/PR mumbo jumbo that is not real, not even representative.
The. If you think about it, a 8-12 hour game will be about 100gig I guess. You won’t need all that data at once. If they can transfer data at 4.8gps then surely that be fast enough?
It will be more than enough. I really don't know why people are feeling insecure about the SSD speed.
Actually, it's supposed to reduce CPU overhead;
www.eurogamer.net
That is assuming that the data actually requires to be stored in the 16GB of RAM first. It doesn't have to be if 100GB of the SSD is also seen as RAM by the system.
The SSD has one speed. The compression of the files on it will determine how effective the SSD output is. It might be as low as 2.4 GB/s for badly compressed files, and theoretically reach higher than 4.8GB/s if it is well compressed.
The fact it only uses 10% of the core is impressive. Imagine Sony has two whole co-processors doing the same work. Remember the advantage of hardware acceleration is that the same process would have a high amount of overhead on the CPU. That doesn't exist here. I also suspect there is some other hardware acceleration somewhere that enables the CPU overhead to be negligible.
I don't think it will use zero. The CPU still needs to tell the I/O which data to read and where to put it in RAM. The I/O handles the actual transfer and decompression by itself.
