@daniel_su123: They aren't going to go around having everyone brand their services with an Xbox or Microsoft label any more than Amazon forces Netflix to brand their streams with "Prime" labels because Netflix runs off Amazon servers. Just like Amazon prime, they will both offer the infrastructure to competitors for the competitors, not their branding, as well as having their own cloud offering. There's literally no tangible benefit to MS forcing them to carry their brand name on their gaming services. The end customers aren't about to go out and purchase Azure instances as more than they would purchase AWS instances from branding on Netflix streams.
One of the things that we know is that Microsoft is positioning XCloud as a White Label service. This means that XCloud will be available to any company that is wanting to create a Game Streaming service. I think this is a genius move of Microsoft, positioning XCloud as a white label service to push Xbox Live, Mixer etc to 2 Billion gamers.
Since XCloud is based on Xbox hardware and developers can develop on both Xbox Consoles and XCloud with no changes.
I can see a situation where Ubisoft's service is based on XCloud, where Ubisoft's games have Xbox Live, Mixer, Xbox Controller and many of Microsoft's services as the backbone, but all under Ubisoft's sub service.
Walmart, Verizon is also rumoured to be developing a Game Streaming service, they could base their service on XCloud, where those games run natively on Xbox Consoles and Streamed to any device and have Xbox Live, Mixer, Xbox Controller support.
The reality of the industry moving forward is that Publishers and Cloud Giants are going to become more increasingly powerful and that they will explore their own service, however, Microsoft is still retaining an ecosystem of Xbox Live, Mixer, Xbox Controllers etc despite Ubisoft, EA, Activision developing their Streaming service.
I think that is the initial take away similar to Surface hardware. Microsoft Xbox branding may eventually become more "niche" as it supports the infrastructure of other game streaming brand and services. Similarly in the game space we have always had MS branded hardware such as controllers, keyboards, mice, etc.
I also made a thread about this exact thing with all three major consoles role and positions
What are you talking about?? There doesn’t need to be hardware backwards compatibility with PS4 Pro because there are no PS4 Pro exclusive games.
And AGAIN the silicon is shared between the Xbox console and server class and there is no server class Navi GPUs at this time and manufacturing partners have already stated that there will not be.
Images of a new AMD-made PCB have appeared online, showcasing what seems to be AMD's first Navi GPU offering, a reference design which will likely be used to create AMD's own Navi graphics cards as well as its workstation/enterprise variants.
You can't be serious. This is only article you could find to support your claim that Navi is a datacenter GPU and was written months ago and says LIKELY as in they are completely guessing and anyone writing before the Computex keynote would have guessed.. This runs contrary to Lisa Su's report of Vega's future from Computex as the leading GPU architecture in its datacenter initiatives. There is literally no evidence to support that Navi architecture has been adapted to datacenter applications at this time and she specifically mentions the MI60 which basically a Vega 64 for datacenter applications.
What are you talking about?? There doesn’t need to be hardware backwards compatibility with PS4 Pro because there are no PS4 Pro exclusive games.
And AGAIN the silicon is shared between the Xbox console and server class and there is no server class Navi GPUs at this time and manufacturing partners have already stated that there will not be.
PS4 Pro's Far Cry 5 used rapid pack math FP16 to improve performance.
The use of FP16 compute in games will reduce resource utilisation while having no impact on rendered images, while also offering a 2x performance increase in GPUs where features like AMD's Rapid Packed Math are available. So far only the PS4 Pro console and Radeon Vega series GPUs have this feature.
---
AMD Radeon Instinct MI6 is Polaris 10 based GPU for server which needs to be replaced by NAVI.
Try again.
Wait, what? This gives more ammunition to the idea that Microsoft's coming servers and virtue of sharing the silicon, the next Xbox is also Vega based. It would mean PS4 pro enhanced games would run with FP16 on their servers. As for Navi, you are making an assumption that PS5 games will even be available via streaming on MS's cloud to begin with and even if they were Vega 64 by and large should be enough to handle the propose Navi performance.
What are you talking about?? There doesn’t need to be hardware backwards compatibility with PS4 Pro because there are no PS4 Pro exclusive games.
And AGAIN the silicon is shared between the Xbox console and server class and there is no server class Navi GPUs at this time and manufacturing partners have already stated that there will not be.
Right so its not relevant to PS5 SSD solutions at this point.
NAVI has significant differences from GCN, hence PS5's PS4 BC would need to present PS4 behavior to PS4 games which includes run-time timings on data resource arrivals.
PS4 Pro's "butterfly" design which doubles the layout of PS4 and down clocking indicates PS4 games are less robust on running scaling hardware.
Significant parts of DirectX includes resource tracking which enable games to scale up or down on different hardware. Different hardware has different resource arrival timings which can create glitches when data resource didn't arrive at expected time.
At this point, there's many unknowns with PS5.
Yes this is why may have Sony filed a patent based on CPU spoofing rather than straight sandbox emulation.
With the increasing demand and necessity of backward compatibility in newer consoles, the machines are desired to run previous-gen games without any synchronisation errors, processing incapabilities and other issues. Cerny and Pilgrim’s patent claims to have fixed these problems.
“In order to deal with problems arising from differences in the behaviour of CPUs to legacy devices, new devices may mimic the specific capabilities of legacy device CPUs when executing legacy applications”, the patent page included. “An important function of the ability to mimic legacy devices is to trick legacy applications to behave as if legacy applications are running on legacy devices. As many applications are designed to run on different processors, modern processors often implement opcodes or registers that allow software applications to discover processor details. To facilitate backwards compatibility, the processor on the new device can execute the opcode in such a way that the information of the different processors is returned in response to the call from the legacy application, or the register value can be provided. Different information is referred to herein as “spoofing processor ID”. The spoofing processor ID selectively identifies the particular capabilities of the new device either differently than actually supported or not at all when it is actually supported.”
How is this at all relevant that Intel doesn’t have 7nm???? No one is arguing transistor density but that space, heat, and power are still concerns and it’s not going to magically make room for 8 cores on most main steam PCs. At the absolute most optimistic projection it may replace the quad core processors with a 6 core in low-mid tier PCs. It’s literally completely irrelevant to the point that mainstream octacore will not be a thing unless most manufacturers switch to AMD.
Your argument on Intel's lacking 7nm is flawed since TSMC 's measures their density differently.
Intel i7-7820X has 12 core CPU imprint with 8 CPUs being active (is close to $499 price range. My point, BOM cost is the same for all 12 core Intel Skylake X CPUs.
Intel x299 motherboards are in AM4 X370/X470 motherboards price range. Intel can respond and it's waiting for AMD..
To keep things into perspective
Absolutely none of this is Relevant to discussion we were having about core counts. At no point were transistor density or performance mentioned. Only that we have a long way to go before 8 cores become the minimum core count in mainstream PCs. Intel is not tracking to have an 8 core low-end processor anytime in the near future. We weren't discussing performance at all.
Again, Intel is waiting for AMD to be competitive and Sony has declared 8 core Zen v2 for PS5's games, hence PC market's competitive pressures would be geared towards it.
Intel's 8 core lowest retail price is with defect chips i.e. the customer still pays for 8 core chip imprint with cores disabled. Fully activated CPU cores has a higher profit margin. BOM cost is the same across different SKUs within the same chip design.
(Socket 2066) X299 motherboard's cost is already AM4's X370/X470 price range, hence Intel already setting up motherboard infrastructure conditions to counter AMD's Ryzen 7 and 9 3000 series.
Socket 3647 is Intel's new premium workstation and server motherboard infrastructure. Xeon Bronze is it's lower cost entry point for Socket 3647 motherboards.
Again. We aren't talking performance, price, or literally any other metric!! The user only commented on raw core count and I explained the expectation that 8-core 16 thread minimums are not realistic in the mainstream low and mid-tier market at this point because no matter what the reason is Intel isn't there yet. There are no 8 core/16thread low-end Intel processors. Nothing about the specs or pricing or business reasons changes that.
Following AMDs presentation I’ll bite on this one.
Sony: using between 8.5-10.2 TFlop Navi high end part giving effective performance between a Vega 56 and a Vega 64 with the 1.25x increase in performance per clock.
MS: more complicated. Lockhart uses a 6Tflop low end Navi Part, effective performance is like a 7.5 AMD GPU.
Anaconda is using a Vega 64 platform at minimum 13.7 Tflops and be overclocked to 14, which it’s also dual purposes for Azure and xCloud to keep the prices low. It is effectively its “Pro” version except releasing simultaneously with it core Lockhart.
Games will be designed as Lockhart for the baseline. Anaconda is the 4K/60FPS machine due to effectively being the performance of its target system Lockhart. Anaconda may be more powerful but may suffer from being built for multiple purposes such as cloud AI and machine learning.
NAVI landed somewhere mid 2019
Current XCloud prototype is using Xbox Scorpio not Vega 64. Xbox Team recycles Azure's network infrastructure with custom Xbox Scorpio server racks.
Sony-MS deal to stream PS5 games would require NAVI based servers i.e. both MS and Sony has a common purpose to reduce semi-custom Zen v2/NAVI based APU server purchase risk.
This year's Radeon RX 5700/5800's release would be following by Radeon Pro NAVI variants.
Both Sony and MS doesn't need Vega II's FP64 math HPC capabilities.
Xbox Team has another team handling XCloud which is different from Office-Azure-and Server-Azure teams.
There are LLVM code differences between NAVI's GFX1010 vs Vega 64's GFX901
Xbox Anaconda is a drop-in successor to Xbox Scorpio, hence it wouldn't be using Vega 64/VII.
Obviously current xCloud is scorpio, however this is not true in the least for the future iteration.
Phil Spencer specifically says that same silicon in both the next Xbox and for their Azure servers will cater to AI compute. The only thing on the market from AMD which has both a consumer and Server computer at their Vega chips which also work well for machine-learning. Navi is not a server grade at this point.
PHIL SPENCER: Yeah. If you go and you watch that video again, one of the things to take notice of is the silicon we're using to stream these games is actually the silicon from our console. And it turns out that consoles have very compatible kind of design criteria to what a blade in a server looks like. You want high power to energy use, low-price, cooling is an issue, all the things that these devices care about in the home work really well on a server blade. The thing that's interesting for us as we roll forward is we're actually designing our next gen silicon in such a way that it works great for playing games in the cloud and also works very well for machine-learning and other non-entertainment workloads. So as a company like Microsoft, we can dual-purpose the silicon that we're putting in. We have a consumer user for that silicon and we'll have enterprise use for those blades as well. It's all in our space around driving down cost to serve, and your cost to serve is made up by two things, how much was the hardware and how much time was that hardware monetized? So we can monetize that hardware over more cycles in the day, in the 24 hours, through game streaming and other things that needs CPU and GPU in the cloud, and we will drive down the cost to serve in our service. So the design as we move forward is done hand-in-hand with the Azure silicon team. And I think that creates a real competitive advantage. If you look on the map of where Azure lives today on the planet, it lives really close to a lot of people who don't play our traditional games today. This is about global scale through the infrastructure that Azure has built and making sure that our silicon and the work that we're putting into those data centers is used for multiple uses inside of Microsoft.
Further, if the Azure servers are Vega 64 based they would out perform the rumored Navi parts on raw compute and be satisfactory for cloud game streaming. This is consistent with all developer rumors of Anaconda being Vega based with AI extensions which most developers won't use while PS5 having a lower TFlop count but performing closer in performance. The developer rumors already exist and Microsoft themselves state that the hardware of the next Xbox would need to be dual purpose. There is only one GPU line from AMD which would fit that criteria.
Following AMDs presentation I’ll bite on this one.
Sony: using between 8.5-10.2 TFlop Navi high end part giving effective performance between a Vega 56 and a Vega 64 with the 1.25x increase in performance per clock.
MS: more complicated. Lockhart uses a 6Tflop low end Navi Part, effective performance is like a 7.5 AMD GPU.
Anaconda is using a Vega 64 platform at minimum 13.7 Tflops and be overclocked to 14, which it’s also dual purposes for Azure and xCloud to keep the prices low. It is effectively its “Pro” version except releasing simultaneously with it core Lockhart.
Games will be designed as Lockhart for the baseline. Anaconda is the 4K/60FPS machine due to effectively being the performance of its target system Lockhart. Anaconda may be more powerful but may suffer from being built for multiple purposes such as cloud AI and machine learning.
This is a nuts hot take, unless PCs switch to AMD in mass the majority of PCs built will be quad core for the next few years so you wouldn’t get much out of an 8 core 16 thread part anyway. Intel isn’t even on 7nm yet to even make mass adoption of octacore feasible for most of the world.
Intel's 14 nm is like TSMC's 10 nm.
Intel's 10 nm CPUs are also year 2019 release.
Try again.
How is this at all relevant that Intel doesn’t have 7nm???? No one is arguing transistor density but that space, heat, and power are still concerns and it’s not going to magically make room for 8 cores on most main steam PCs. At the absolute most optimistic projection it may replace the quad core processors with a 6 core in low-mid tier PCs. It’s literally completely irrelevant to the point that mainstream octacore will not be a thing unless most manufacturers switch to AMD.
Your argument on Intel's lacking 7nm is flawed since TSMC 's measures their density differently.
Intel i7-7820X has 12 core CPU imprint with 8 CPUs being active (is close to $499 price range. My point, BOM cost is the same for all 12 core Intel Skylake X CPUs.
Intel x299 motherboards are in AM4 X370/X470 motherboards price range. Intel can respond and it's waiting for AMD..
To keep things into perspective
Absolutely none of this is Relevant to discussion we were having about core counts. At no point were transistor density or performance mentioned. Only that we have a long way to go before 8 cores become the minimum core count in mainstream PCs. Intel is not tracking to have an 8 core low-end processor anytime in the near future. We weren't discussing performance at all.
Log in to comment