Arm Co-Founder Doesn't Think NVIDIA Owning the Company Would be in Its Best Interests

[PowerPC]

Planned obsolescence by virtue of evolving software. You want to buy a computer that is only good for 2 to 3 years before games, streaming services, applications, etc stop working?

I said that people will choose whatever is the most efficient CPU for the tasks they currently need to do. This has nothing to do with my own opinion. And the normal tasks don't change every 2 to 3 years. Again, most people just watch Youtube or stream Twitch, edit a video, or do some specific task like that over and over. And ARM is about finding those tasks and delivering a better product for those tasks. Otherwise, ARM wouldn't be so big in other areas. You could say the same argument about planned obsolescence in phones and I never heard this issue raised there. They could just as likely figure out how to make ARM work for most people for their normal PC tasks, and why would you then prefer x86, if it is much less efficient? People buy what they need now. Imagine you could get an ARM PC that does everything you do now, but faster, cheaper and more efficiently than a x86 PC would? I believe you and most people would go with ARM at that point. Future-Proofing is always just a speculation game.

I guess developing software for ARM is like developing software for consoles. You need to keep the hardware in mind more and optimize to its capabilities. This is not even a bad thing, in my opinion. Just another way to make software. It all depends on software in the end and how well developers will be able to make it. Apple is making strides on that front. It also seems like it will become incredibly simple to port x86 apps to ARM automatically and make them run almost just as good on ARM. This is what Apple does really well with Rosetta 2 on the ARM Mac Mini Dev kit, as leaks have shown.

And to finish, look at the problems x86 development has had, especially with Intel? ARM is way ahead of the curve when it comes to performance increases throughout the years. To the point where Apple feels more comfortable to develop its own chips and not have to rely on even a company like Intel that can't deliver on its promises. Here's a good summary on that (starts at 6:56):

 

[TheoneandonlyMrK]

I said that people will choose whatever is the most efficient CPU for the tasks they currently need to do. This has nothing to do with my own opinion. And the normal tasks don't change every 2 to 3 years. Again, most people just watch Youtube or stream Twitch, edit a video, or do some specific task like that over and over. And ARM is about finding those tasks and delivering a better product for those tasks. Otherwise, ARM wouldn't be so big in other areas. You could say the same argument about planned obsolescence in phones and I never heard this issue raised there. They could just as likely figure out how to make ARM work for most people for their normal PC tasks, and why would you then prefer x86, if it is much less efficient? People buy what they need now. Imagine you could get an ARM PC that does everything you do now, but faster, cheaper and more efficiently than a x86 PC would? I believe you and most people would go with ARM at that point. Future-Proofing is always just a speculation game.

I guess developing software for ARM is like developing software for consoles. You need to keep the hardware in mind more and optimize to its capabilities. This is not even a bad thing, in my opinion. Just another way to make software. It all depends on software in the end and how well developers will be able to make it. Apple is making strides on that front. It also seems like it will become incredibly simple to port x86 apps to ARM automatically and make them run almost just as good on ARM. This is what Apple does really well with Rosetta 2 on the ARM Mac Mini Dev kit, as leaks have shown.

And to finish, look at the problems x86 development has had, especially with Intel? ARM is way ahead of the curve when it comes to performance increases throughout the years. To the point where Apple feels more comfortable to develop its own chips and not have to rely on even a company like Intel that can't deliver on its promises. Here's a good summary on that (starts at 6:56):

Yeah , no.

Arm is not in any way at all anywhere close to consoles.

They're set in stone, arm have an awful lot of core's with differentiation, plus a very configurable GPU from 2/16 cores and a vast amount of accelerator's that are completely optional.

Apple's arm does not equal Qualcomms or Samsung and there are many , many more arm chip suppliers with different ideas too.

X86 has fewer side options cos it's just better at general.

Arm chips being low power get to play on new node's first an artificial advantage that doesn't sit well with attaining HP performance out of those, simple.

And your making a lot of assumptions regarding use cases, genuinely arm cannot do what I want on three of the four computers I have to use, only my backup/file server Could happily use arm.

So no thanks dualy x86 with HT is too slow arms not even capable of running these software.

Oh and who buys a computer solely with the few tasks they have in mind, most exceed that due to the intelligent ideology that they may do more and that they might need something unexpected to work , the lack of cd/dvd drives for example has probably annoyed everyone who works with PC's at some point.

 

[PowerPC]

Arm is not in any way at all anywhere close to consoles.

Not literally, obviously. My whole post is about the way you develop for these things and how hard or easy it is. Again, I'm not a chip maker, I'm a software developer. I also don't know all the abbreviated words from that industry that you use, like HP. It would be much easier to talk if you wouldn't use abbreviations. Works better in general, when you are talking about something from your field with someone who isn't in your field.

But the point of the similarity between ARM chips and consoles was about how people go about developing for these things. You have to account for the very special nature of a console being specialized for games, just like you would have to account for a more specialized nature of an ARM designer chip that works best for a certain number of applications but can also simulate all kinds of other applications like Apple is doing with Rosetta 2.

And the meta point is that people, and therefore also companies, will go for that, just economically, because they can reduce the cost of the chips and at the same time improve the performance and efficiency on certain key tasks. It's kind of a no-brainer. We're at performance levels with CPUs that already exceed what we need most of the time. But what we're still lacking in right now (and always will be) is efficiency at scale and prices. Prices that should be going down for these things, but are actually climbing the whole time in the last years. For an absolute commodity like a computer these days, that's kinda unprecedented. Commodities should usually get lower and lower in price as time goes by. This is where ARM will shine the most. So we could be seeing a major shift to ARM if just the price is lower, which we already know it definitely is.

 

[TheoneandonlyMrK]

Not literally, obviously. My whole post is about the way you develop for these things and how hard or easy it is. Again, I'm not a chip maker, I'm a software developer. I also don't know all the abbreviated words from that industry that you use, like HP. It would be much easier to talk if you wouldn't use abbreviations. Works better in general, when you are talking about something from your field with someone who isn't in your field.

But the point of the similarity between ARM chips and consoles was about how people go about developing for these things. You have to account for the very special nature of a console being specialized for games, just like you would have to account for a more specialized nature of an ARM designer chip that works best for a certain number of applications but can also simulate all kinds of other applications like Apple is doing with Rosetta 2.

And the meta point is that people, and therefore also companies, will go for that, just economically, because they can reduce the cost of the chips and at the same time improve the performance and efficiency on certain key tasks. It's kind of a no-brainer. We're at performance levels with CPUs that already exceed what we need most of the time. But what we're still lacking in right now (and always will be) is efficiency at scale and prices that should be going down for these things, but are actually climbing the whole time in the last years. For an absolute commodity like a computer is these days, that's kinda unprecedented. Commodities should usually get lower and lower in price as time goes by. This is where ARM will shine the most. So we could be seeing a major shift to ARM if just the price is lower, which we already know it definitely is.

Sorry , but the arm ,one handed interface I use for the internet's makes typing tiresome then you have auto correct.

And that plus Media consumption and communication are arm's mainstay.

So accelerator development here by arm and apple and intel have vastly paid off, simple but arm are not alone, AMD have some but access to all of it anyway as does intel tbf.

What I meant by a process advantage is this.

Apple and Qualcomm use the cutting edge At risk(might not get many working) production node's at foundries (Tsmc) (5nm At the minute but 3nm soon) , the other slight setback is that the process to make those chips is being actively developed to scale it up.
Small chips work well for this since the size helps keep low yields to output useful amounts of chip's.

They're designed and manufactured for very low power use, not speed and performance.

The improvement developed goes into the creation of a high power version of that node, which also enjoys a quicker , streamlined ramp to full production runs.

Sooo, a cutting edge non server, arm design is actually held back in performance due to node's, though it gets it's already efficient ass massaged by the same thing.

It gain's the maximum efficiency from a node with the maximum performance always blocked to it.

And time and again single threads at speed have rained supreme , thing's are changing but not the way the chips are made. Well not introduction ideology.

As for useable.

My arm phone does all the usual very well but nothing like the work , actual work of any other.

And the one pc I have to do actual work on is not upto the minor tasks I require if it.
I'm only talking looking at technical drawings here too no simulation running, my home pc does that.

There's a million different uses for a billion different computing devices so to assume we're all doing what you think is wrong, sure a statistical and factual number of people are surfing now but I also do so much more than that with them, a lot do.

 

[PowerPC]

Sorry , but the arm ,one handed interface I use for the internet's makes typing tiresome then you have auto correct.

And that plus Media consumption and communication are arm's mainstay.

So accelerator development here by arm and apple and intel have vastly paid off, simple but arm are not alone, AMD have some but access to all of it anyway as does intel tbf.

What I meant by a process advantage is this.

Apple and Qualcomm use the cutting edge At risk(might not get many working) production node's at foundries (Tsmc) (5nm At the minute but 3nm soon) , the other slight setback is that the process to make those chips is being actively developed to scale it up.
Small chips work well for this since the size helps keep low yields to output useful amounts of chip's.

They're designed and manufactured for very low power use, not speed and performance.

The improvement developed goes into the creation of a high power version of that node, which also enjoys a quicker , streamlined ramp to full production runs.

Sooo, a cutting edge non server, arm design is actually held back in performance due to node's, though it gets it's already efficient ass massaged by the same thing.

It gain's the maximum efficiency from a node with the maximum performance always blocked to it.

And time and again single threads at speed have rained supreme , thing's are changing but not the way the chips are made. Well not introduction ideology.

The process advantage of ARM is there simply because ARM is a much more simple chip. It has an insane power-efficiency advantage over other types of designs. This gives ARM a huge edge on cost-effective manufacturing and a faster development process. I believe the second part is true, but I'm not an ARM architect. But it clearly must have been a big part of the reason why Apple is going with ARM, because they are able to design it on their own for their needs and clients. It's not possible to design something for the needs and wants of your clients if the design is too complicated. ARM PCs aren't even a new idea. The first Computers ARM released was a desktop PC line called Acorn Archimedes.

The thing is also that Intel is falling quickly and AMD might not be actually the one taking its place in the long run. It could actually be ARM that might take the most advantage from Intel's downfall.

 

[TheoneandonlyMrK]

The process advantage of ARM is there simply because ARM is a much more simple chip. It has an insane power-efficiency advantage over other types of designs. This gives ARM a huge edge on cost-effective manufacturing and a faster development process. I believe the second part is true, but I'm not an ARM architect. But it clearly must have been a big part of the reason why Apple is going with ARM, because they are able to design it on their own for their needs and clients. It's not possible to design something for the needs and wants of your clients if the design is too complicated. ARM PCs aren't even a new idea. The first Computers ARM released was a desktop PC line called Acorn Archimedes.

The thing is also that Intel is falling quickly and AMD might not be actually the one taking its place in the long run. It could actually be ARM that might take the most advantage from Intel's downfall.

So the A12 bionic is a simple chip, assumptions are carrying you far from the truth.

Arm chips only appear as good as they do because of a vast array of coprocessors ,and accelerator's.

The raw core's are just too weak but perspective is key.

And like I said 30% of arm's efficiency comes from the node at 5nm , they're not at all out of reach of x86 in terms of innovation or progress, even Intel aren't as off the ball as your implying.

 

[PowerPC]

So the A12 bionic is a simple chip, assumptions are carrying you far from the truth.

Arm chips only appear as good as they do because of a vast array of coprocessors ,and accelerator's.

The raw core's are just too weak but perspective is key.

And like I said 30% of arm's efficiency comes from the node at 5nm , they're not at all out of reach of x86 in terms of innovation or progress, even Intel aren't as off the ball as your implying.

ARM is a much simpler architecture and it definitely does overshadow intel in price for performance. RISC is simpler by definition in terms of hardware needed, period. Cost-effectiveness compared to x86 additionally includes the much more costly development process x86 seems to be having, I'm not just talking about hardware cost. If you believe Intel won't have any more delays after this while ARM is delivering for Apple quicker than anybody anticipated, you don't actually realize what Intel is dealing with.

Look at this video from 2018

Look at the whole thing, he does a benchmark.

After the video tell me how the "raw cores are just too weak" when he is comparing it core for core two years ago? And the price is lower on ARM without massive scale like Intel, so how? Do you really think that it can't happen on the PC side as well? ARM was just too busy taking the mobile market for themselves these last years. It seems like now they have the time and money to show what they can really do in the PC market.

 

[Initialised]

Everyone loves Nvidia. I wonder why.....

They don't want to get PhysXed out of existence?

 

[RoutedScripter]

That's right, absolutely agreed, Nvidia should never buy ARM, for many reasons, I don't have time right now to explain.

Tech is already consolidated enough, this is just more ridicolousness ontop of it. And hopefully it goes all to UK, so you have more diveristy, I'm sick of the reliance on Eastern Asian hardware, badly translated BIOS screens and customer feedback not really being respected well.