A more realistic threat to console gaming
Over the past while I've been arguing why the Steam Box isn't really a viable alternative to console gaming and why I don't think it poses a threat there (caveat: I haven't said it won't succeed, or even that it shouldn't succeed. Just that I don't think any successes it gets will be in large part from sales cannibalized from the console market). This article however presents a possibility.
iPhones, iPads and Android devices are used as gaming machines as often as they are for social media. They also vastly outnumber console sales by an insane margin. They haven't been a threat to the console gaming market because their performance hasn't really allowed them to do so. If this chipset delivers and actually can meet or exceed the PS4 and Xbox One consistently and in all areas (not just graphics) then in 3-5 years you might start seeing the true slow death of consoles.
I'm skeptical as usual. I'm not sure I believe that passively cooled SoC's leapt so far forward "overnight" that they could out consistently out perform an actively cooled gaming console running on modern chipsets. To me, that sounds fishy as all hell. Both the PS4 and Xbox One run on 8 core, 64-bit processors which get reasonably close to 2GHZ. The systems have 8GB of RAM and while they have around the same number of compute units as these chips I'll wager that they are clocked higher.
Why? Because if NVidia was this close to being able to run that much power on a passively cooled SoC then they could have built even faster dedicated graphics that are actively cooled. And if they could have done that, then Sony and Microsoft would have been clawing at each other to be the first to put either a mobile SoC into their console and dramatically reduce the devices footprint or scoffing at AMD for their VASTLY inferior graphics chipsets.
It just doesn't seem likely to me.
Although, there are a number of factors not directly related to the SoC that affect performance like RAM. And, while not always and not as impactful as RAM, there is disk space as well. There isn't a cell phone I'm aware of and only very few tablets (especially ARM based) which have even the 500GB that the current gen consoles have which has already been panned as insufficient by many.
Then there is the clock speed of the compute units. While I'll imagine that being based on the Kepler architecture, they are quite likely theoretically able to get up to speeds that could outperform the chips in their console brethren. But I think actual units that ship in devices like phones and tablets will be clocked much slower. I honestly can't see a competitively clocked mobile GPU being able to run passively, not to mention the impact on battery life.
But, with a lower clock speed and more cores, you can generally get the same performance with less battery usage and heat. Those 2 things are things that mobile electronics manufacturers gobble up.
I expect that while some of the performance increase these chips are capable of will be utilized, I think it will look and feel incremental. Most vendors will opt for clock speeds that give a noticeable enough difference without sacrificing too much battery life or generating too much heat.
And, as always, there is the argument I made against the Steam Box. OS can play a big role in performance as well. There are already cards on the market that make the GPU's in even the PS4 look like a joke. But, between the OS level optimizations for gaming and the compile time optimizations for systems, games run on this "inferior" hardware can make the superior hardware look like the waste of money. And the primary consumers of NVidia's SoC's run an OS which ISN'T optimized for console quality gaming (Android). It is also a hardware agnostic platform, so developers can't target this new architecture specifically.
So, maybe my 3-5 years was still a bit too optimistic. Maybe 5-10 years. I'm still seeing a lot of hurdles for this. And I still doubt the calibre of this claim. IF the only thing an OEM needs to do is bundle enough fast RAM into a cell phone without active cooling to achieve PS4/Xbox One like performance AND doing that doesn't absolutely rape battery life then I'll fall back on my 3-5 year guess. Otherwise, I'd have to throw in with the 5-10 years instead.
3-5 years because the eco-system would still need to catch up. Even if the tech is there today, the software titles aren't. Most developers won't code for a phone or tablet expecting this kind of power until it is either already the norm or approaching it at a fast pace. It will still take a couple of years for this level of performance to become mainstream. And that is when you would expect to see a shift that starts eroding consoles dominance.
But, I'll put out another suggestion. Perhaps this news has nothing to do with the mobile segment. Perhaps NVidia wants to start seeing their SoC's in desktops, laptops and perhaps even consoles. If this news is a bid to get these chips into devices where fans are commonplace and accepted then maybe they are trying wrestle dominance for Intel and AMD on yet another battlefield. I don't think it will disrupt anything in less than the 3-5 year timeframe, but it makes a much more valid argument if the performance they boast is only when actively cooled. It shows that their solutions have more or less caught up the x86 chips and that they may soon surpass them.
NVidia has long been the leader in graphics tech. But, done right, this may also position them as the dominant CPU force through their SoC solutions.
iPhones, iPads and Android devices are used as gaming machines as often as they are for social media. They also vastly outnumber console sales by an insane margin. They haven't been a threat to the console gaming market because their performance hasn't really allowed them to do so. If this chipset delivers and actually can meet or exceed the PS4 and Xbox One consistently and in all areas (not just graphics) then in 3-5 years you might start seeing the true slow death of consoles.
I'm skeptical as usual. I'm not sure I believe that passively cooled SoC's leapt so far forward "overnight" that they could out consistently out perform an actively cooled gaming console running on modern chipsets. To me, that sounds fishy as all hell. Both the PS4 and Xbox One run on 8 core, 64-bit processors which get reasonably close to 2GHZ. The systems have 8GB of RAM and while they have around the same number of compute units as these chips I'll wager that they are clocked higher.
Why? Because if NVidia was this close to being able to run that much power on a passively cooled SoC then they could have built even faster dedicated graphics that are actively cooled. And if they could have done that, then Sony and Microsoft would have been clawing at each other to be the first to put either a mobile SoC into their console and dramatically reduce the devices footprint or scoffing at AMD for their VASTLY inferior graphics chipsets.
It just doesn't seem likely to me.
Although, there are a number of factors not directly related to the SoC that affect performance like RAM. And, while not always and not as impactful as RAM, there is disk space as well. There isn't a cell phone I'm aware of and only very few tablets (especially ARM based) which have even the 500GB that the current gen consoles have which has already been panned as insufficient by many.
Then there is the clock speed of the compute units. While I'll imagine that being based on the Kepler architecture, they are quite likely theoretically able to get up to speeds that could outperform the chips in their console brethren. But I think actual units that ship in devices like phones and tablets will be clocked much slower. I honestly can't see a competitively clocked mobile GPU being able to run passively, not to mention the impact on battery life.
But, with a lower clock speed and more cores, you can generally get the same performance with less battery usage and heat. Those 2 things are things that mobile electronics manufacturers gobble up.
I expect that while some of the performance increase these chips are capable of will be utilized, I think it will look and feel incremental. Most vendors will opt for clock speeds that give a noticeable enough difference without sacrificing too much battery life or generating too much heat.
And, as always, there is the argument I made against the Steam Box. OS can play a big role in performance as well. There are already cards on the market that make the GPU's in even the PS4 look like a joke. But, between the OS level optimizations for gaming and the compile time optimizations for systems, games run on this "inferior" hardware can make the superior hardware look like the waste of money. And the primary consumers of NVidia's SoC's run an OS which ISN'T optimized for console quality gaming (Android). It is also a hardware agnostic platform, so developers can't target this new architecture specifically.
So, maybe my 3-5 years was still a bit too optimistic. Maybe 5-10 years. I'm still seeing a lot of hurdles for this. And I still doubt the calibre of this claim. IF the only thing an OEM needs to do is bundle enough fast RAM into a cell phone without active cooling to achieve PS4/Xbox One like performance AND doing that doesn't absolutely rape battery life then I'll fall back on my 3-5 year guess. Otherwise, I'd have to throw in with the 5-10 years instead.
3-5 years because the eco-system would still need to catch up. Even if the tech is there today, the software titles aren't. Most developers won't code for a phone or tablet expecting this kind of power until it is either already the norm or approaching it at a fast pace. It will still take a couple of years for this level of performance to become mainstream. And that is when you would expect to see a shift that starts eroding consoles dominance.
But, I'll put out another suggestion. Perhaps this news has nothing to do with the mobile segment. Perhaps NVidia wants to start seeing their SoC's in desktops, laptops and perhaps even consoles. If this news is a bid to get these chips into devices where fans are commonplace and accepted then maybe they are trying wrestle dominance for Intel and AMD on yet another battlefield. I don't think it will disrupt anything in less than the 3-5 year timeframe, but it makes a much more valid argument if the performance they boast is only when actively cooled. It shows that their solutions have more or less caught up the x86 chips and that they may soon surpass them.
NVidia has long been the leader in graphics tech. But, done right, this may also position them as the dominant CPU force through their SoC solutions.
Comments
Post a Comment