• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.

Next-Gen PS5 & XSX |OT| Console tEch threaD

Status
Not open for further replies.

pawel86ck

Banned
really? where did he state that? can you quote him?
all i read from his interview about api was:
For the Xbox, they have to put DirectX and Windows on the console, which is many years old, but for each new console that Sony builds, it also rebuilds the software and APIs in any way it wants.

i don't see any word about update. he means that ms build upon older version of dx but sony starts from 0 every generation that's why it's better then ms api.

In the original interview, Salehi said that DirectX 12 hadn't been updated for a long time, and curious omitted any mention of DirectX 12 Ultimate. It's unclear if he's had any time with the Xbox Series X development kit, which is regarded quite favorably by the industry judging by our conversations.
 
Last edited:

LordOfChaos

Member
So yes.. that trumps the PS5 SSD solution. There is nothing that PS has revealed so far that gives the GPU direct access to 100GB of data on SSD at this gens memory bandwidth speeds. Rough terms would be about 326GB/s for gddr5. Thats near instant.

Wait wha?

8-channel NAND is getting to 326GB/s how exactly? Bypassing system components in the way doeesn't remove the physicality of the bits available here.

This includes SSG. NAND is NAND.
 
Last edited:
If the gap remains between Sony and Microsoft, which is more than likely to happen again in a more devastating way, then Sony could easily strike more exclusives deals with 3rd party games, resulting in more games with new next gen experiences instead of half-baked experiences.
really? where did he state that? can you quote him?
all i read from his interview about api was:
For the Xbox, they have to put DirectX and Windows on the console, which is many years old, but for each new console that Sony builds, it also rebuilds the software and APIs in any way it wants.

i don't see any word about update. he means that ms build upon older version of dx but sony starts from 0 every generation that's why it's better then ms api.

How is that logical? The reason why they rewrite the api is because every generation they had a different ISA.

emotion engine, cell, amd architecture... vs... directx
 
Wait wha?

8-channel NAND is getting to 326GB/s how exactly? Bypassing system components in the way doeesn't remove the physicality of the bits available here.

Read the articles I've posted. Their solution appears to be based on Radeon Pro SSG solution that does exactly this using a basic PCIE4 SSD.

DF also examined this solution and came to a similar conclusion about the origin of the implementation.
 
Last edited:

Gamernyc78

Banned
I see based on the reveled aspects of their velocity architecture, that the XSX GPU has access to not only 6GB/s of compressed read write access to the SSD, 560GB/s to 10 GB of VRAM, and also direct access to a special 100GB partition on the SSD that does not require shuttling data across the PCIE bus at all.

So yes.. that trumps the PS5 SSD solution. There is nothing that PS has revealed so far that gives the GPU direct access to 100GB of data on SSD at this gens memory bandwidth speeds. Rough terms would be about 326GB/s for gddr5. Thats near instant.

What's great is that this part of the Xbox Velocity architecture spec is constantly repeated by them openly.

Only people rooting for this piece of hardware or that piece ignore this spec. I don't understand why we can't just examine and understand the tech.

Can anyone tell me of anything within the Sony i/o implementation that matches that?

🤦‍♂️🤦‍♂️Go ask the professionals. I can't believe your trying to say tht Microsofts ssd implementation is better 😂😂😂
 

LordOfChaos

Member
Read the articles I've posted. Their solution appears to be based on Radeon Pro SSG solution that dies exactly this using a basic PCIE4 SSD.


Yes, and SSG NAND runs at up to 6-8GB/s. Not hundreds. It doesn't magically make NAND faster, even removing PCI-e limits, the physical amount of chips, bit width, and clock speed have a certain bandwidth.

I think you might be confused with SSG's VRAM bandwidth at 484 GB/s, the HBM2 acts as a caching layer to the much slower 8GB/s NAND.

https://www.amd.com/en/products/professional-graphics/radeon-pro-ssg

https://www.pcgamer.com/amd-radeon-pro-ssg-pairs-vega-with-2tb-of-memory/

AMD says the SSG portion of the card can read data at up to 8GB/s and write data at up to 6GB/s. That's a lot slower than HBM2, but with the High-Bandwidth Cache Controller (HBCC) and thanks to the direct access, it's still over twice as fast as talking to SSDs over the PCIe bus.

It does that much, but you can't magic bits. I've also been pointing to SSG since the start of rumors about this generation, that's not what I'm contesting, but you seem to be conflating the actual VRAM bandwidth with the NAND bandwidth. Microsoft also has not mentioned the 100GB for SSG running any faster than their 2.4GB/s.
 
Last edited:

Bo_Hazem

Banned
Sometimes you get the feeling that people consider Cerny as an infallable all-knowing God, that singlehandedly designes the Playstation, while the Microsoft-engineers are bumbling idiots that has no clue what they are doing.

Comments such as "off the shelf", "just a pc", "brute force", "so many bottlenecks" and what have you are idiotic and should almost be a bannable offence. It ruins any form of discussion.

I'm not sure if describing a console like that is more offensive than calling other posts "idiotic". Just chill, mate. Here:

"Microsoft's new Xbox Series X is a very tall console that borrows from the design of gaming PCs"


"It's clear that the PS5 is nimble and smart while the Xbox Series X has brute force power, but there are many other aspects to consider. "


"Well, Salehi stated that the Xbox Series X generally works below the 12 TFLOPs that Microsoft claims, there are bottlenecks that limit the GPU, and you have to achieve perfect conditions of use of resources to be able to use those 12 TFLOPs, which in practice is very difficult to achieve. "


Let's just ban the whole internet if that makes you feel better, my friend :messenger_beaming: . It's free speech, as long as you don't insult anyone, just discuss and share thoughts and ideas.
 
Last edited:

joe_zazen

Member
There comes a point where these things are just elementary at lower resolutions. If the system has a 1080p external render target, you don't need as much ram (or bandwidth) to make that happen. I will assume the SSD will be the same configuration, MS is already using a budget controller (according to most reports, and the specs appear to align with that) so not much to gain by lowering the bar there. We don't yet know how things will be approached regarding the TFs either. First-party games could set aside 1.5 - 2TF for ML techniques that lower internal resolution targets a great deal and reduce texture bandwidth needs by a factor of 4.

lockhart is budget console so i doubt it will have 1TB storage. ferwer chips—>lower speed. and textures will be smaller, seems like a waste of money to have the same ssd setup.

they are cutting costs to the bone with LH (if the rumors of no disc drive are true) so I expect the bare minimum with the console and a $299 msrp with frequent $199 sales. Maybe I am wrong, and it will be $399-$449.
 
Last edited:
Apart from the savage words, I do wonder how much of a potential diff there really is. I get a feeling PS5 is really far ahead here. The LOD potentials on it are in another league. But I do wonder how much hdd on pc primarily will hold things back, in that order.

Thats why the XSX has mesh shaders and SFS... its incredible that people seem to invent awesome capabilities for the PS5, which the XSX actually has and have demonstrated, but claim that PS5's capabilities will be better.

This place is odd.
 

DaGwaphics

Member
lockhart is budget console so i doubt it will have 1TB storage. ferwer chips—>lower speed. and textures will be smaller, seems like a waste of money to have the same ssd setup.

they are cutting costs to the bone with LH (if the rumors of no disc drive are true) so I expect the bare minimum with the console and a $299 msrp with frequent $199 sales.

Lower capacities don't necessarily equate to slower SSD speeds, it depends on the number of nand chips and channels to the memory controller at play. Not to say that you couldn't withstand to lose a little speed here.
 

Bo_Hazem

Banned
How is that logical? The reason why they rewrite the api is because every generation they had a different ISA.

emotion engine, cell, amd architecture... vs... directx

AMD had to work closely with Microsoft to make directx12 work properly, with Vulkan being superior on the PC space already.

"He suggests that the PlayStation’s API and OS offer more freedom to developers as opposed to the Xbox’s DirectX-based custom version of Windows."


"With this architecture powering both the next generation of AMD Radeon graphics cards and the forthcoming Xbox Series X gaming console, we’ve been working very closely with Microsoft to help move gaming graphics to a new level of photorealism and smoothness thanks to the four key DirectX 12 Ultimate graphics features -- DirectX Raytracing (DXR), Variable Rate Shading (VRS), Mesh Shaders, and Sampler Feedback. "

 
Last edited:

DaGwaphics

Member
Thats why the XSX has mesh shaders and SFS... its incredible that people seem to invent awesome capabilities for the PS5, which the XSX actually has and have demonstrated, but claim that PS5's capabilities will be better.

This place is odd.

Also, those 500-600GB installs are going to get painful, for these Sony games that stream GB/s at every turn. :messenger_tears_of_joy:

Expect a major improvement over the 20-40mb/s reserved for stream assets in the current gen, I would not hold my breath for sustained streaming of even 2GB/s or more.

The proof will be in the pudding, both sides will show their recipes soon.
 

DForce

NaughtyDog Defense Force
I see based on the reveled aspects of their velocity architecture, that the XSX GPU has access to not only 6GB/s of compressed read write access to the SSD, 560GB/s to 10 GB of VRAM, and also direct access to a special 100GB partition on the SSD that does not require shuttling data across the PCIE bus at all.

So yes.. that trumps the PS5 SSD solution. There is nothing that PS has revealed so far that gives the GPU direct access to 100GB of data on SSD at this gens memory bandwidth speeds. Rough terms would be about 326GB/s for gddr5. Thats near instant.

What's great is that this part of the Xbox Velocity architecture spec is constantly repeated by them openly.

Only people rooting for this piece of hardware or that piece ignore this spec. I don't understand why we can't just examine and understand the tech.

Can anyone tell me of anything within the Sony i/o implementation that matches that?
And you've come to this conclusion based on DF SSD Theory video, too? I reference that video several times in this thread and I know how they were talking about how next gen consoles would work on a smaller scale in comparison's to AMD's SSG demonstration.


I find it a bit amusing that people are saying PlayStation 5's SSD is not secret sauce, but somehow Xbox's Velocity Architecture can help close the gap or exceed PlayStation 5's SSD raw speed.
 

DaGwaphics

Member
Hold on, what's this "instant access to 100 gb/s" jibberee-joo?

Just people misunderstanding a quote from MS. Basically, they referred to the fact that the entire game install is directly accessible by the CPU/GPU, this doesn't directly increase the theoretical maximums of the SSD in use. Obviously there would never be a need to access all assets at the same time, unless the entire game exists in a single small room that is always in full line of site.
 
Last edited:

SonGoku

Member
One thing though: why would only 10 GB of data be available for games in Scenario #2? I don't think that would make good sense from an engineering POV, even as a compromise for going mixed 1 GB/2 GB modules.
In order for CPU access to not take a disproportionate amount of bandwidth (48GB/s vs 80GB/s in the example) CPU data must be inside the fast pool
There's nothing inherently bad here, its a compromise, even the Xbox architect called it a trade off. This system was needed to feed the more powerful GPU (GB/s per TF is comparable to PS5). They didn't fuck up, its a calculated tradeoff.

In short:
If they use the full 13.5GB , average bandwidth available to GPU is similar to PS5 proportional to GPU power ( similar GB/s per TF)
If they limit game access to 10GB, the XSX GPU has slight surplus bandwidth advantage over PS5's (higher GB/s per TF)
Dunno, something feels like info on the memory setup is still not fully understood, I would assume the system would allow for data access from both pools
Its just how interleaved memory configurations work, MS would need to reinvent interleaved memory to overcome this reality, same limit is present on PS5 btw
Even to access 10MB worth of data all 10 chips or 6 chips must be accessed for XSX, same for PS5 to access that 10MB all 8 chips must be run simultaneously

To change how data is evenly distributed across all chips every time CPU access RAM would bring its own set of complications, latencies and performance penalties
 
Last edited:

scie

Member
I think it's all said and shown here, time to eat some juicy egg sandwiches then play Final Fantasy 7. You may consider watching what Mark Cerny said about how it's being read/transferred if you're interested in understanding the matter.

I have and still he didn´t say anything about access time which is instant on SSDs regarding of having an SATA SSD, NVME SSD... He talked about how data is accessed in different priority states and how it is transfered (read and write speeds) which still is something different of don´t having the need to search the data on SSD which applied to not having an access time basically. So what is your point?
 

Bo_Hazem

Banned
This ^

It's basic math, a 4 channel gen 3 drive can't hit XSX SDD performance (will be a little short). The average budget gen 4 (not on the market yet), will likely land at 3.5GBs.

I'm already using 3.5GB/s (raw) from my 1 year old PC. PCIe 4.0 should be maxed at 7GB/s. XSX is 2.4 GB/s which is average/below average speeds for PCIe 3.0.

970PRO_02_03.jpg
 

DaGwaphics

Member
I'm already using 3.5GB/s (raw) from my 1 year old PC. PCIe 4.0 should be maxed at 7GB/s. XSX is 2.4 GB/s which is average/below average speeds for PCIe 3.0.

970PRO_02_03.jpg

It depends on the number of channels. An 8 channel gen 3 controller could hit 3.5GB/s, a 4 channel controller only about half that (some as high as 2GB/s with caching tricks). From all reports XSX is using a 4 channel controller, which would limit the drive to a maximum of 3.5GB/s (some may reach a bit higher with caching tricks). I assume they don't hit that because they are not using the highest binned nand chips (neither is Sony, since a 12 channel config with the best chips would be faster).

8 channel designs come at a distinct cost premium. Not sure how you figure the XSX drive is below average in the PC space, the best selling NVMe drives are 1.7GB/s - 2GB/s. SATA SSDs are still outselling NVMe nearly 2:1, with speeds measured in MB/s.
 
Last edited:
In order for CPU access to not take a disproportionate amount of bandwidth (48GB/s vs 80GB/s in the example) CPU data must be inside the fast pool
There's nothing inherently bad here, its a compromise, even the Xbox architect called it a trade off. This system was needed to feed the more powerful GPU (GB/s per TF is comparable to PS5). They didn't fuck up, its a calculated tradeoff.

In short:
If they use the full 13.5GB , average bandwidth available to GPU is similar to PS5 proportional to GPU power ( similar GB/s per TF)
If they limit game access to 10GB, the XSX GPU has slight surplus bandwidth advantage over PS5's (higher GB/s per TF)

Its just how interleaved memory configurations work, MS would need to reinvent interleaved memory to overcome this reality, same limit is present on PS5 btw
Even to access 10MB worth of data all 10 chips or 6 chips must be accessed for XSX, same for PS5 to access that 10MB all 8 chips must be run simultaneously

To change how data is evenly distributed across all chips every time CPU access RAM would bring its own set of complications, latencies and performance penalties

Ah okay, that's a good way of looking at it, then. MS mentioned something about adding ECC to their memory, which since they also want to use XSX in servers that makes sense (server systems absolutely need ECC memory). Not sure how that will impact memory usage exactly, guess we'll have to wait and see.

The main thing still of concern there though is the amount of physical memory WRT bandwidth; PS5 and XSX offer the same amount of physical GDDR6 memory but PS5 manages it over just eight chips, which is an advantage in this type of setup when it comes to physical memory per TF and module bandwidth per physical GB (the second one moreso with XSX's slower pool being the point of focus).

Like you said, it's interesting tradeoffs and it will be something to see how devs work with them, particularly XSX because there are some scenarios where it has a capacity disadvantage vs. PS5 (and likewise, some scenarios where PS5 has a bandwidth disadvantage to XSX's faster pool).
 

Grodiak

Member
Is there really room for Lockhart? Seriously asking.

Even if there were easy ways to upscale quality between Lockhart and XsX, there is bound to be a ton of people touting how Lockhart is holding XsX back from its full potential. I just feel like that would kind of shift the focus point to shipping the cheaper box in bulk, leaving the XsX to be the "high-end" niche product - which would be a damn shame. Obv we don't know the costs of any of these boxes yet so... 🤷‍♂️
 
You are exactly correct LordofChaos. I wasnt trying to conflate that with actual access speeds of the NAND bandwidth. I think the Phil Spencer comment is taken as an equivalence not actual throughput. The VRAM access doesn't access all 100 GB simultaneously but can access any portion of data that GPU needs within that 100GB independently of VRAM at a rate similar to if it had a rating of GDDR5. I was surprised when I read that as well. How the actual implementation works, I haven't been able to pin down and I have looked in many places.

Let me share with you some of the resources I have been looking at and maybe if you can clarify it would help all of us.

1. https://www.gamersnexus.net/guides/2537-how-the-radeon-pro-ssg-works-basics
Here they state that "AMD tells us that processes moved to the SSG are functioning on average 3-4x the speed of using system storage."

2. https://www.pcworld.com/article/3099964/amds-new-ssg-technology-adds-an-ssd-to-its-gpu.html
"With SSG Technology, NAND is directly connected to the GPU core, residing within the graphics card itself."

3. This is a pretty good overall breakdown:


4. This is the DF overview of the same topic where they discuss tom warren's comment about SSD virtual ram usage and the SSG technology.


5. https://wccftech.com/spencer-on-xsx...ke-that-sony-is-investing-in-ssd-for-ps5-too/
Phil Spencer: Thanks to their speed, developers can now use the SSD practically as virtual RAM. The access times of the SSD come close to the memory access times of the current console generation. Of course, the OS must allow developers appropriate access that comes out of a pure storage medium. But then we will see how the address space will increase immensely - comparable to the change from Win16 to Win32 or in some cases Win64.

I dont know how else to take Spencer's comment than at face value. That is the origin of my comment.
 
Last edited:

PaintTinJr

Member
We don't know this, Cerny said it has to be higher than the internal to make up for the overhead of the IO unit handling the extra priority levels
By how much we don't know, could be 6GB/s 6.5GB/s 7GB/s etc.
I’m still intrigued by the number of priority levels mentioned. Anything more than two levels (high, low) hasn’t really been viable - without using a ring-bus - because a bridged type bus at saturation is so preoccupied with binary-back-off collisions for accessing the bus that providing any sort of isochronous throughput is too difficult at high load. I will be surprised if the PS5 doesn’t have at least one ring bus to accommodate those 6 levels of priority under heavy load.
 

MarkMe2525

Member
My friend, why not stop the psychic talk and personal mockery and keep it informative? Can you explain how 100GB-ready is when you can only transfer at the speed of 2.4GB/s (4.8GB/s compressed)? I think you're smarter than me as you've been showing here that I'm suffering from psychic or behavior issues, in that case you should be smarter than that misleading information that adds nothing new. It's like someone saying I can eat a burger in one bite, then the other responds that he can eat a whole cow in 2 months, it's not the same.

EDIT: PS5 is capped at 22GB/s compressed, and BGs BGs has seen 20GB/s in action. XSX is capped at 6GB/s.
I'm not pretending to know how it works. I would imagine it works in a similar way to this.
AMD has been working on this technology for years now and demonstrated it back on 2016. It needed 1.5 to 2GBps. Doesn't need 22GBps. Again that took me all of 3 minutes to find.
 

CrysisFreak

Banned
Is there really room for Lockhart? Seriously asking.

Even if there were easy ways to upscale quality between Lockhart and XsX, there is bound to be a ton of people touting how Lockhart is holding XsX back from its full potential. I just feel like that would kind of shift the focus point to shipping the cheaper box in bulk, leaving the XsX to be the "high-end" niche product - which would be a damn shame. Obv we don't know the costs of any of these boxes yet so... 🤷‍♂️
There is room for it from a business perspective for sure. It could, possibly, maybe, sell like hotcakes if it's a next-gen console for little money.
From an enthusiast perspective it is not so cash money of MS to release it. It WILL hold games back as a shitty casual baseline for sure. People should drop the scalability delusion.
 

LordOfChaos

Member


I'm not denying the impact of caching either, it just sounded like it was being claimed that the NAND would be over 300GB/s at first.

With the VRAM acting as another caching layer and a high cache hit rate, certainly you can access many of those assets as soon as you need them with this new paradigm. It doesn't, however, change the speed of the NAND physically, at all, so I think we agree.
 

Bo_Hazem

Banned
I'm not pretending to know how it works. I would imagine it works in a similar way to this.
AMD has been working on this technology for years now and demonstrated it back on 2016. It needed 1.5 to 2GBps. Doesn't need 22GBps. Again that took me all of 3 minutes to find.

Future will tell, with more unique 4K assets, travelling insanely fast, XSX would need more CPU work than PS5. Overall, on slower-paced games, PS5 will use less CPU, yet having a unified ram that is still around 16GB available compared to 10GB+3.5GB with drastic speed differences that need extra work from devs to optimize for.

 

Gamernyc78

Banned
Is there really room for Lockhart? Seriously asking.

Even if there were easy ways to upscale quality between Lockhart and XsX, there is bound to be a ton of people touting how Lockhart is holding XsX back from its full potential. I just feel like that would kind of shift the focus point to shipping the cheaper box in bulk, leaving the XsX to be the "high-end" niche product - which would be a damn shame. Obv we don't know the costs of any of these boxes yet so... 🤷‍♂️

Lockhart will not be a success and won't move many units in comparison to other consoles. As an option it's OK but as for profit I don't see it doing anything. It's caught in between too many things or below them if it exists and rumors are true.
 
Last edited:

pasterpl

Member
And you've come to this conclusion based on DF SSD Theory video, too? I reference that video several times in this thread and I know how they were talking about how next gen consoles would work on a smaller scale in comparison's to AMD's SSG demonstration.


I find it a bit amusing that people are saying PlayStation 5's SSD is not secret sauce, but somehow Xbox's Velocity Architecture can help close the gap or exceed PlayStation 5's SSD raw speed.

on the other side you have got people who claim that ssd can close the tflop gap...what I am saying that both sides present a lot of wishfu, not based on facts, thinking
 

DaGwaphics

Member
Future will tell, with more unique 4K assets, travelling insanely fast, XSX would need more CPU work than PS5. Overall, on slower-paced games, PS5 will use less CPU, yet having a unified ram that is still around 16GB available compared to 10GB+3.5GB with drastic speed differences that need extra work from devs to optimize for.


You don't think the OS on PS5 will need any ram? I wonder how they keep the kernel and the basic drivers/containers going to even access the SSD (hint you need the kernel storage drivers to have access to the SSD at all). Any streaming features are going to require RAM as well. You simply can't run an entire OS on disk without at least some memory footprint.
 

MarkMe2525

Member
Future will tell, with more unique 4K assets, travelling insanely fast, XSX would need more CPU work than PS5. Overall, on slower-paced games, PS5 will use less CPU, yet having a unified ram that is still around 16GB available compared to 10GB+3.5GB with drastic speed differences that need extra work from devs to optimize for.

Why would the CPU have to work more? Is this a guess? If it does have to work more, well frankly it's faster so that would pick up the slack. That whole split ram argument is also just an assumption your making. I've heard that it's one line of code that a developer must add to accommodate the XSX ram setup so my "assumption" is there isn't gonna be an appreciable difference in difficulty to code for.
 

Grodiak

Member
There is room for it from a business perspective for sure. It could, possibly, maybe, sell like hotcakes if it's a next-gen console for little money.

Yeah, this I am sure is the thought process. Also to be able to tout Xbox as "the most powerful" and have - I believe someone has been using "Little Timmy?"'s father buy the cheaper but most likely "similarly" powerful - could work for their advantage. Marketing message translating to all series boxes. I guess I am just hoping for a strong next gen for all parties to make sure nothing IS hold back :D

Lockhart will not be a success and won't move many units in comparison to other consoles. As an option it's OK but as for profit I don't see it doing anything. It's caught in between too many things or below them.

I guess this is a pricing and PR question. I CAN see it selling big among casuals if it is cheap enough and still can play all the next-gen titles, per CrysisFreak's notion above. Could very well lead to a Xbox One vs X situation - albeit if launched at the same time the whole situation is a bit different.
 

Darius87

Member
In the original interview, Salehi said that DirectX 12 hadn't been updated for a long time, and curious omitted any mention of DirectX 12 Ultimate. It's unclear if he's had any time with the Xbox Series X development kit, which is regarded quite favorably by the industry judging by our conversations.
let's all believe windows central now! :messenger_grinning_squinting:
again can you quote his words about the update of dx from his interview? misinformation from others sites doesn't count.
 

joe_zazen

Member
There is room for it from a business perspective for sure. It could, possibly, maybe, sell like hotcakes if it's a next-gen console for little money.
From an enthusiast perspective it is not so cash money of MS to release it. It WILL hold games back as a shitty casual baseline for sure. People should drop the scalability delusion.

it also fits in with their tech approach to gaming. They want the content to work across a wide spectrum. X1 in years 1 and possibly 2, and lower end pcs for the whole gen. Baseline is low with or without xss, so might as chuck it out there.

if sony has nothing that looks and or plays like nothing on the xbox xeries platform, then xss will have been a good call. If sony somehow makes a Zelda botw must play exclusive that outclasses things on any other platform...
 

GetSchwifty

Banned
it also fits in with their tech approach to gaming. They want the content to work across a wide spectrum. X1 in years 1 and possibly 2, and lower end pcs for the whole gen. Baseline is low with or without xss, so might as chuck it out there.

if sony has nothing that looks and or plays like nothing on the xbox xeries platform, then xss will have been a good call. If sony somehow makes a Zelda botw must play exclusive that outclasses things on any other platform...
The Last of Us: Part 2 could be Sony's Breath of The Wild moment if they hold it back for PS5 launch. It could be considered the best launch title of all time. I think they should do it. TLoU2 is already delayed indefinitely...
 
I'm not denying the impact of caching either, it just sounded like it was being claimed that the NAND would be over 300GB/s at first.

With the VRAM acting as another caching layer and a high cache hit rate, certainly you can access many of those assets as soon as you need them with this new paradigm. It doesn't, however, change the speed of the NAND physically, at all, so I think we agree.

Yes I think that is what they mean by instant in this regard. Sorry for the confusion earlier. In all cases checking local cache, then vram then this Virtual mem pool/buffer is faster than doing a read/write access to stream off any SSD.


The reddit post and this conversation with you have actually been the most enlightening I've seen on the subject. Now if we can just pin down the memory architecture access process.

I'm still trying to understand 10/6 split in reasonable terms. Somewhere in there there is a gap as to whether or not the CPU/GPU access locks out the other processors access to the RAM resource that its not locked to. I dont think the memory diagrams explain it well yet. Although I think rnival is very close.
 
"That "instant" access might be a slight exaggeration, since that expanded pool of data still seemingly has to come from the system's NVMe storage at a 2.4GB/s transfer rate. Even expanded to 4.8GB/s thanks to a new decompression stack, that's well below the 336 to 560GB/s access for data stored on the system's 16GB of RAM. It's also not clear why Microsoft specifically cites a 100GB limit for those "instant" assets amid the 1TB of internal storage built into the system."

- https://www.google.com/amp/s/arstec...lag-limiting-tech-of-the-xbox-series-x/?amp=1
 
I'm not denying the impact of caching either, it just sounded like it was being claimed that the NAND would be over 300GB/s at first.

With the VRAM acting as another caching layer and a high cache hit rate, certainly you can access many of those assets as soon as you need them with this new paradigm. It doesn't, however, change the speed of the NAND physically, at all, so I think we agree.

I think the idea of this plus SFS... is pretty revolutionary. You have access to 10GB plus 100GB but still only pull exactly what you need within the fustrum of the players view. I think both Sony and MS made the right calls in ensuring that their solutions are efficient. Now the question is how effective are they.

The game I'm most interested in this year is Flight Simulator 2020. This solution is exactly the right kind of solution to enable highly detailed texture and LOD for their air maps.
 
Last edited:

Gamernyc78

Banned
Yeah, this I am sure is the thought process. Also to be able to tout Xbox as "the most powerful" and have - I believe someone has been using "Little Timmy?"'s father buy the cheaper but most likely "similarly" powerful - could work for their advantage. Marketing message translating to all series boxes. I guess I am just hoping for a strong next gen for all parties to make sure nothing IS hold back :D



I guess this is a pricing and PR question. I CAN see it selling big among casuals if it is cheap enough and still can play all the next-gen titles, per CrysisFreak's notion above. Could very well lead to a Xbox One vs X situation - albeit if launched at the same time the whole situation is a bit different.

Even with low pricing my comment still holds. If it's not 200 or below it's doa. Pricing isn't everything although it's gd to have a low price. Pstv was dirt cheap after few months and still sold like shit. Microsoft image and what ppl perceive as value just isn't there anymore. Lockhart isn't something I think casuals are asking for so the price doesn't matter.

Thts just my opinion of course.
 
Last edited:
"That "instant" access might be a slight exaggeration, since that expanded pool of data still seemingly has to come from the system's NVMe storage at a 2.4GB/s transfer rate. Even expanded to 4.8GB/s thanks to a new decompression stack, that's well below the 336 to 560GB/s access for data stored on the system's 16GB of RAM. It's also not clear why Microsoft specifically cites a 100GB limit for those "instant" assets amid the 1TB of internal storage built into the system."

- https://www.google.com/amp/s/arstechnica.com/gaming/2020/03/going-for-speed-the-load-busting-lag-limiting-tech-of-the-xbox-series-x/?amp=1

I cannot disagree with this... however according to the supposed layout this access does NOT go through the system PCIE but is an on-die connection to the GPU. We dont have the SSD to GPU specification yet as to how this is done. Time will reveal.
 

GetSchwifty

Banned
Yes I think that is what they mean by instant in this regard. Sorry for the confusion earlier. In all cases checking local cache, then vram then this Virtual mem pool/buffer is faster than doing a read/write access to stream off any SSD.


The reddit post and this conversation with you have actually been the most enlightening I've seen on the subject. Now if we can just pin down the memory architecture access process.

I'm still trying to understand 10/6 split in reasonable terms. Somewhere in there there is a gap as to whether or not the CPU/GPU access locks out the other processors access to the RAM resource that its not locked to. I dont think the memory diagrams explain it well yet. Although I think rnival is very close.

If anything, Sony's implementation seems far closer to realising the dream of SSG than XSX's implementation. Sony's SSD solution even seems to go far beyond what SSG does.

Radeon SSG basically involves being able to load larger data-sets, far quicker than what is capable through normal storage on a PC. Consoles are completely different and based upon HSA and doesn't have the same type of constraints as a PC.

Therefore, neither console is SSG, as SSG is a technology to mitigate the problems along the IO pipeline specific to PCs.

However, both next-gen consoles are trying to do something similar to SSG in that they are expanding the available RAM pool available to the GPU and obviously refinining their APIs to customise it more specifically for the purposes of gaming.

There isn't any secret sauce going on, but PS5's solution (based on currently available information), seems far better, faster, more robust and more tightly integrated and considered than XSX's one. Of course though, the proof will be in the pudding.
 

pawel86ck

Banned
let's all believe windows central now! :messenger_grinning_squinting:
again can you quote his words about the update of dx from his interview? misinformation from others sites doesn't count.

Here you go:
For the Xbox, they have to put DirectX and Windows on the console, which is many years old,
"Many years old" and "not updated" is the same thing, so please dont suggest windows central is lying.
 
Last edited:
Status
Not open for further replies.
Top Bottom