Intel Rumored to Have Offered $2 Billion for RISC-V Chipmaker SiFive

I agree, but that narrow window of opportunity just meant it wasn't a great product for wide spread adoption. It leaves you with ... some NUC-style computers (POS devices, higher powered HTPC type streaming boxes, etc), maybe some very entry level servers, and lower/mid-level NAS devices.
All of those already dominated by ARM.
 
Gaming entertainment is huge, huge $, the many big AAA titles run on x86, not arm. If that will change is to be seen. Seeing Tesla go with Ryzen and RNDA2 for their information/gaming system in their car was very interesting. It automatically will have thousands of games available to it, even if using Linux. Now who wants to sit in a car and play games on a fixed location monitor? How would the backseat riders even enjoy that feature? To me it seems stupid and a waste but someone will love it. The point is leveraging what x86 can do well or ecosystem, can allow use cases that beat out ARM.
 
Gaming entertainment is huge, huge $, the many big AAA titles run on x86, not arm. If that will change is to be seen. Seeing Tesla go with Ryzen and RNDA2 for their information/gaming system in their car was very interesting. It automatically will have thousands of games available to it, even if using Linux. Now who wants to sit in a car and play games on a fixed location monitor? How would the backseat riders even enjoy that feature? To me it seems stupid and a waste but someone will love it. The point is leveraging what x86 can do well or ecosystem, can allow use cases that beat out ARM.

I agree. But just to argue a counterpoint:

Up until recently, every console generation was different architecture, different OS, different everything. The gaming industry does not need backwards compatibility or even a like architecture to move forward. If ARM, or anything else, starts to present a better value proposition, the games will follow very quickly.

And you could almost argue they already are: love it or hate it, mobile gaming (largely on ARM) is HUGE, and includes Nintendo right now in its camp.
 
I agree. But just to argue a counterpoint:

Up until recently, every console generation was different architecture, different OS, different everything. The gaming industry does not need backwards compatibility or even a like architecture to move forward. If ARM, or anything else, starts to present a better value proposition, the games will follow very quickly.

And you could almost argue they already are: love it or hate it, mobile gaming (largely on ARM) is HUGE, and includes Nintendo right now in its camp.

How recently? The original XBOX is almost 20 years old now.
 
ARM is getting very close to x86 performance cpu wise. Couple it with an nvidia gpu and you got some serious power.
 
As recently as the Xbox One/PS4, so very recently.
Yup. This gen and last gen PS/XB were pretty much the only ones to be similar architecture back to back.

Before that, just looking at Xbox (OG): Intel x86/nVidia -> XB360: PPC/ATI -> XB1 -> AMD x86/AMD Polaris

I guess you could also say Wii -> Wii U ... both were PPC based CPUs with ATI/AMD graphics, but they weren't entirely backwards or cross compatible and the GPUs were pretty different critters, even though they both were sourced from the same company. Wii U could run Wii stuff via a virtualization layer though.
 
I agree. But just to argue a counterpoint:

Up until recently, every console generation was different architecture, different OS, different everything. The gaming industry does not need backwards compatibility or even a like architecture to move forward. If ARM, or anything else, starts to present a better value proposition, the games will follow very quickly.

And you could almost argue they already are: love it or hate it, mobile gaming (largely on ARM) is HUGE, and includes Nintendo right now in its camp.
x86 has a very big library that goes back decades, most of those games would never get back ported, maybe emulated. Of course new games could tap into an Arm Eco System but the number of Tesla's systems would be rather small making those efforts probably not worth it. These types of systems are unique to drive sells for a particular brand otherwise Tesla would probably just use something else if everybody else had it. Since the PC market is still viable for games as well, this makes new games playable as well on the Tesla System. It makes a lot of sense if other companies follows, they would use this type of approach. Nvidia solutions could use ARM and maybe GeForce now and allow networking of Switches if Nintendo wants to be involved ( I doubt it).

Tesla, 10's of thousands of games ready to play transparently, including tomorrow games with no real difficult extra developer work. Now if people find this valuable in a Tesla is to be seen.
 
ARM is getting very close to x86 performance cpu wise. Couple it with an nvidia gpu and you got some serious power.
This is a hard one to prove, and to argue-

The best example I have of an ARM CPU is Apple's M1, but having fast CPU cores is not what makes it 'fast'. It's having extremely, anal-retentively, well-tuned software, that has been tuned iteratively alongside the hardware.

What we don't really see is what happens when software steps outside of what the hardware is tuned for. For stuff like content creation, we've already seen benchmarks for stuff that the M1 doesn't have hardware acceleration for, and it turns out it's not as fast as a 'comparable' x86 system.
 
x86 has a very big library that goes back decades, most of those games would never get back ported, maybe emulated. Of course new games could tap into an Arm Eco System but the number of Tesla's systems would be rather small making those efforts probably not worth it. These types of systems are unique to drive sells for a particular brand otherwise Tesla would probably just use something else if everybody else had it. Since the PC market is still viable for games as well, this makes new games playable as well on the Tesla System. It makes a lot of sense if other companies follows, they would use this type of approach. Nvidia solutions could use ARM and maybe GeForce now and allow networking of Switches if Nintendo wants to be involved ( I doubt it).

Tesla, 10's of thousands of games ready to play transparently, including tomorrow games with no real difficult extra developer work. Now if people find this valuable in a Tesla is to be seen.
Ever heard of a little something... what's it called... aah yes, emulation?
 
This is a hard one to prove, and to argue-

The best example I have of an ARM CPU is Apple's M1, but having fast CPU cores is not what makes it 'fast'. It's having extremely, anal-retentively, well-tuned software, that has been tuned iteratively alongside the hardware.

What we don't really see is what happens when software steps outside of what the hardware is tuned for. For stuff like content creation, we've already seen benchmarks for stuff that the M1 doesn't have hardware acceleration for, and it turns out it's not as fast as a 'comparable' x86 system.
I'm not saying its going to happen overnight, but the breach between ARM and x86 has reduced tremendously in just a few years. Hopefully x86 will kick into 6th gear and we may see substantial performance gains soon. Othewise ARM will catch up, hey maybe we'll finally get a Windows on ARM that doesn't suck.
 
I'm not saying its going to happen overnight, but the breach between ARM and x86 has reduced tremendously in just a few years. Hopefully x86 will kick into 6th gear and we may see substantial performance gains soon. Othewise ARM will catch up, hey maybe we'll finally get a Windows on ARM that doesn't suck.
Well, the bigger change is the difference between ARM and what users need on a 'desktop'.

Turns out that if you accelerate all of the intensive tasks, i.e. video transcoding, the CPU cores don't actually have to be that powerful.

That's the big difference here: ARM SoCs are being tuned heavily toward user workloads whereas desktop operating systems and x86 aren't. That's also nothing new, since game developers have been doing the same with consoles for decades.

So the real shift would be for traditional desktop hardware and software vendors to start approaching design like Apple has done with the M1 and IOS / MacOS.
 
Desktop doesn’t really need all that much firepower either. I could probably do 80% of what I need on a 8gb raspberry pi…
 
Desktop doesn’t really need all that much firepower either. I could probably do 80% of what I need on a 8gb raspberry pi…
yeah... no

I could go as low as a core i3 to handle most of my work and even some gaming. I have much slower PCs at home just for browsing and RDP, but they struggle even for video conferencing in teamviewer.
 
yeah... no

I could go as low as a core i3 to handle most of my work and even some gaming. I have much slower PCs at home just for browsing and RDP, but they struggle even for video conferencing in teamviewer.
Zoom runs fine in the web client on Raspberry Pi OS, with the caveat that I've run mine on a 8GB pi 4 in a FLIRC case and at 2147mhz - I wouldn't make any promises on a 2GB pi with a bare board at stock speed. Given the bulk of my day is spent in putty sessions, audio only zoom, e-mail, or slack, a pi could cover that.
 
Zoom runs fine in the web client on Raspberry Pi OS, with the caveat that I've run mine on a 8GB pi 4 in a FLIRC case and at 2147mhz - I wouldn't make any promises on a 2GB pi with a bare board at stock speed. Given the bulk of my day is spent in putty sessions, audio only zoom, e-mail, or slack, a pi could cover that.
Great, unfortunately Teams alone requieres much more resources on the desktop, so its a no go for me on anything less than the core i3 I sometimes use.
 
Ever heard of a little something... what's it called... aah yes, emulation?
Mentioned it in the quote you posted from me? Problem with emulation is it takes work, is not always consistent and can have its own issues or bugs with the tens of thousands of game titles and cause another developer work load for in like Tesla's case, a very tiny market to have to support, plus many games would never get any updates since published. Why emulate when you can have the real McCoy? New devices, using arm and what ever GPU with new games supporting it makes more sense vice using emulation on software not likely to get updated. Of course applications, unless a wide spread adoption of a standard ARM platform for developers would most likely be needed for making it worth while. PC ECO system which goes along with the Consoles as well, x86 centric makes that a pretty big barrier for gaming unless Cell phone gaming is good enough, plenty of stuff there but not so much AAA big titles.

Just a unique use case (which may not sell) where x86 is the better solution over ARM for the time period. I am sure other types of use cases could also be better with x86 over ARM.
 
Become a Patron!
Back
Top