You don't like streaming at all. As you expressed, the price doesn't matter. It was never a question about whether you liked streaming or not.

I mean you guys could just continue doing what you're already doing and ...not buy it...

Begun the Streaming Wars have!

https://www.amazon.com/PlayStation-Now-Subscription-12-Months/dp/B019CYYSFC?th=1

Thoughts on the price?

Personally I will likely re-sub for 2 more years at this price.

Began the Streaming Wars have!

https://www.amazon.com/PlayStation-Now-Subscription-12-Months/dp/B019CYYSFC?th=1

Thoughts on the price?

Personally I will likely re-sub for 2 more years at this price.

I see people are really losing themselves in this summary of the report. My take away from the report is that it seems entirely accurate based on the restriction placed on the summary. Specifically the below line in the footnotes:

¹ In assessing potential conspiracy charges, the Special Counsel also considered whether members of the Trump campaign “coordinated” with Russian election interference activities. The Special Counsel defined “coordination” as an “agreement—tacit or express—between the Trump Campaign and the Russian government on election interference.”

This statement provides a very narrow window of what Special Counsel would consider "coordination or collusion". This entails that some tangible agreement was established between the two parties specifically on election interference. The likelihood of this is that in truth, some members of Trump's campaign got various information from various sources and may have even alluded to or used this information. However without a specific agreement this could not be seen as proper coordination as per the directive of the Special Counsel.

The second portion, obstruction of justice is specifically not charged due to the follow:

Generally speaking, to obtain and sustain an obstruction conviction, the government would need to prove beyond a reasonable doubt that a person, acting with corrupt intent, engaged in obstructive conduct with a sufficient nexus to a pending or contemplated proceeding. In cataloguing the President's actions, many of which took place in public view, the report identifies no actions that, in our judgment, constitute obstructive conduct, had a nexus to a pending or contemplated proceeding, and were done with corrupt intent, each of which, under the Department's principles of federal prosecution guiding charging decisions, would need to be proven beyond a reasonable doubt to establish an obstruction-of-justice offense.

This is of particular importance. Because the underlying crime of specific agreements with Russia were not present. The idea that there was obstruction to prevent the investigate of a crime that did not exist cannot be proven. In other words, Trump's opposition could be viewed as legitimate due the fact that this specific crime did not take place. In other words it cannot be proven that Trump was attempting to prevent an investigation into an actual crime, rather he could justify his actions in stating that he truly believed he did not commit a crime and was being targeted.

Now, ultimately this is important because it tells us two things.

1) It does not state that there were no crimes. However it does state that there was not a coordinated conspiracy. This is an important distinction to make.

2) Further it does not mean Obstruction of Justice was not committed. Rather, proving it beyond a reasonable doubt would be dubious given that "coordination" was not proven.

3) This does not state other possible crimes or even whether the President was aware of or approved of other actions by Russian nationals. Example, the campaign could have been aware of Russia activities but took no active part in the process, hence no conspiracy nor collusion. The full report will have this information documented, however the lesser indictments of other members of the campaign implies that these individuals were aware of Russia activities and in communication with various entities however not taking an active role in the processes of interference.

It's hilarious how closely this process mirrors Netflix and video streaming. People forget Netflix wasn't even the first to implement video streaming, Amazon was and further Netflix only allowed less than 20 hours of streaming per month on a small catalog of titles at SD initially. Netflix won because it made its streaming business cheap enough, for what people wanted which was lots of content and options at a low price. It's the same exact process and will scale up quickly. My favorite article is from 2007 which announced Netflix streaming business . With them even going so far as to tout Blockbuster's advantages because DVDs aren't going anywhere for a long time and Netflix didn't have the infrastructure to be successful long-term in the video distribution business... LOL.

https://www.nytimes.com/2007/01/16/technology/16netflix.html

I've talked about it before but it's totally worth stating again.

Higher cloud resources and bandwidth costs MONEY. Google themselves stated they will offer a 720P stream option. The underlying hardware would need to be able to render the game in 4K/60fps before the issue of even bandwidth comes up. We can extrapolate the approximate costs of game streaming by the services that already exist (Most are built with an average of 20-25 hours of gaming per month, in the interest of this exercise we will assume 20 hours per month average because the proposed pricing of GeForce Now):

Shadow:

The current PC streaming service costs $35 per month although it first launched at $50 per month. The specs are slightly lower than that of Google Stadia. However this does NOT include games. At 20 hours per month, this would cost $1.75.

PaperSpace:

They have an excellent breakdown of what you should expect price-wise. https://www.paperspace.com/pricing. The equivalent of Stadia specs would cost you $0.78 per hour or about $15.60 for 20 hours of gaming. Again NO GAMES, no OS.

Geforce NOW. Although still in beta it is proposing a cost of $25 per 20 hours of gaming. This would be for up to 1080P streaming and include only free to play games users would need to bring there own through services such as Steam, etc. The cost per hour would be approximately $1.25.

PS Now. When first launched it presented a rental proposition of 4 hours for $3. Or about $0.75 per hour. At its current monthly offering using the 20 hour metric it would be $1 per hour of gaming. However if we use its $100 per year metric. ($8.33 per month, the price for 20 hours falls to $0.42 per hour.)

Direct comparison of price-feature

Shadow: $1.50 - $1.75 per hour No Games, Includes OS , 4K/60

Paperspace $0.51 - $0.78 per hour No Games, No OS VM dependent.

Geforce Now $1.25 per hour FtP Games, Includes OS, 1080P/60

PSNow $0.41- $1 per hour 700 Games, Custom OS, 720P/30 Downloadable

From our outline above we can see typical PS4 level games would likely fall in the $0.50 and below price per hour for the cloud resources. Where as we could double or even triple that cost per hour as we talk about 4K/60 cloud hardware. Again, this is before we have the discussion of games content which will further the cost. Considering Google's talk of having streams from between 720P/30 to 4K/60 I think they are setting up a tiered model.

My personal guess is as follows for Stadia pricing:

Base tier: $6-$15 or Ad- supported month/ FtP games 720P/30. Purchase games or rentals per hour price: $0.30-$0.75

Mid tier: $25, AA games 1080P/60 some games included. First party games included. Per hour price $1.25

Top tier: $35, AAA games 4K/60 many games included. Some third party games included. Per hour price $1.75

The above seems very reasonable for the market price of running these cloud services because an hour of high-end gaming is NOT free.

I think this is brilliant and likely to take off. It gives a basic VR experience for relatively cheap if you already own a switch. No, it's not a PSVR killer because its a tier above that. Like saying a Mustang is a Ferrari killer... nah. The order in terms of quality and tech is:

$5 VR kits for phones, Google cardboard, This Thing, Occulus go/Samsung VR, PSVR, Occulus Rift/Vive.

@ronvalencia said:

@qx0d said:

https://www.youtube.com/watch?v=wi9DtYSOMks

https://www.gameinformer.com/gdc-2019/2019/03/19/google-announces-stadia-a-powerful-new-game-streaming-service

Google CEO Sundar Pichai admits he's not a big gamer, but he's got big plans for gamers around the world. Today Google revealed its new game streaming platform, Stadia, with a stated goal to bring the best games to everyone in the world. "When we say for everyone, we really mean it. It's one of our most cherished values at the company," Pichai says.

Here's everything we know about the platform right now.

Reasons for NVIDIA skipping HBM v2, the argument for GDDR6

From http://www.pcgameshardware.de/Radeon-RX-Vega-64-Grafikkarte-266623/Tests/Benchmark-Preis-Release-1235445/3/

Vega 64's HBM v2 has 484 GB/s theoretical with 303 GB/s practical memory bandwidth before memory compression is applied. Efficiency is 62.60 percent

RX-580's GDDR5 has 256 GB/s theoretical with 193 GB/s practical memory bandwidth before memory compression is applied. Efficiency is 75.39 percent

-----------

For Xbox Anaconda and PS5, assume GDDR6 has similar practical memory bandwidth efficiency as GDDR5 i.e. 75.39 percent efficient .

Candidates

PS4 like with 256 bit PCB motherboard

GDDR6-13000 x 256 bit yields 313.62 GB/s practical memory bandwidth estimate. <----- Not brainier to why NAVI's RX-580 replacement to be Vega 56/64 range.

GDDR6-14000 x 256 bit yields 337.75 GB/s practical memory bandwidth estimate. <----- Not brainier to why NAVI's RX-580 replacement to be Vega 56/64 range.

X1X like with 384 bit PCB motherboard

GDDR6-13000 x 384 bit yields 470.43 GB/s practical memory bandwidth estimate. <----- this configuration will crush Google's Stadia specs.

GDDR6-14000 x 384 bit yields 506.62 GB/s practical memory bandwidth estimate. <----- this configuration will crush Google's Stadia specs.

MS's X1X has GDDR5-6800 which is GDDR-7000 chips which is one step bellow GDDR5-8000. Pattern, MS selects 1 step below.

Sony's PS4 Pro has GDDR5-7000 which is one step bellow GDDR5-8000.Pattern, Sony selects 1 step below.

Both Sony and MS will crush Google's Stadia box.

GDDR6-13000 x 384 bit config's 470.43 GB/s has higher gap against Vega 64's HBM v2's 303GB/s when compared to X1X vs PS4 Pro.

GDDR6-13000 x 384 bit config would be my candidate for Xbox Anaconda to replace Xbox One X i.e. drop in replacement for GDDR5 with GDDR6.

You can't hyperfocus on the bandwidth as the main decider.

The cost of having a 385 bit bus is going to increase your cost, heat, and space which all factor into the end price. HBM2 is a fraction of the size and space and heat of a GDDR6 chip. In a blade server or console these factors are high priorities as well and its not outside reason that consoles could utilize small pools of HBM2 like they uses to use eScram or other high bandwidth memory pools in the past. Remember, PS4s unified memory without smaller high bandwidth pools was the exception. Not the rule.

@ronvalencia said:

@michaelmikado said:

@ronvalencia said:

P

@michaelmikado said:

@ronvalencia:

No the video is wrong, what he is suggesting is literally impossible the GPU and GPU cannot be on two separate packages while the CPU utilizes vRAM from the GPU. This configuration is only feasible when the cpu and GPU are single packages and bus with access to the same memory controller.

Edit: to be fair later in article they claim that the HBM2 is shared which contradicts what they claimed earlier but still state that they are separate packages. It sounds like there’s a mix of marketing and technical information. So will wait for more info but the fact that they contradict themselves on whether the HBM2 is just vRAM or not would be a basic spec to get.

FALSE.

The original Xbox 360 has separate CPU and GPU/NB/MCH packages with unified GDDR3 memory architecture.

Separate CPU package is connected to GPU package which is then connected to unified GDDR3 memory.

PC's CPU can access GPU's VRAM via PCI-E links. 1990s PCI protocols still has server RAM expansion cards via PCI expansion slots! PCI-E still runs with PCI protocols.

PC's Windows NT/HAL wasn't designed to register memory pools in GPU's VRAM as system memory. Linux's flexibility doesn't have Windows NT's rigidity.

The 360 having separate packages was only possible by having direct access to each other's cache and aren't required to go through a memory bus. While I can't speak on Linux, unless they've replicated the xbox 360 cache scheme at the server level which would arguably be a larger accomplishment. It makes no sense from a performance stance and even if they did, the cost would be prohibitive for that kind of customization when there are many off the shelf options. Ignoring the fact that the cache differences don't mean they are the same chips they could have done any number of configurations to the cache as stated.

Did you forget GCN's X86-64 pointer compatibility?

GCN is design for fusion with x86 CPU regardless if it's discrete CPU+discrete GPU combo or an APU.

X86 CPU can pass it's X86-64 pointer to GCN IGP or GCN dGPU.

Both X86 and GCN are little endian data formatted.

PS5 dev kits are Ryzens with Vega 56/64 or VII which can run PS4 games e.g. GTS at 8K resolution. Sony could release high end desktop AMD PC that can run PS4 games.

PS4's fusion link is better than PCI-E 16X version 2.0 but slightly inferior to PCI-E 16X version 3.0 and substantially inferior to incoming PCI-E 16X version 4.0.

---

MI60/MI50, ZEN 2 and X570 chip-set supports PCI-E version 4.0

This is exactly what I stated though. GCN is able to use either traditional vRAM or system RAM by design, specifically in an APU. The slide you show literally explain the exact problem I described. Trying to have the GPU and CPU both access the RAM simultaneously through different buses without any direct communication between the two is the issue. AMD's solution is the same as the 360 where the GPU and CPU communicate directly via cache without attempting to have two different access points to the memory controller. The speed of the bus doesn't matter if the flow of data cannot be managed correctly between the two computing units competing for memory resources. Again, I'm not saying Google didn't do that, only that in order for this to work they would need to give the GPU and CPU direct access to each other. It also doesn't mean they aren't using 7601s as they could just customize the caches to be shared with the GPU cores to allow only stacks of VRAM.

The problem with this is that it would require a very custom setup when they basically already have premade equivalent off the shelf parts. There would not be allot of benefit in designing a custom spec in this manner besides the heat and space savings.