Nvidia's Plan for ARM to Take X86 's Throne | Exclusive

erek

[H]F Junkie
Joined
Dec 19, 2005
Messages
10,785
"The X86 architecture has reigned supreme for many, many years - but that may just change. According to our exclusive info, Nvidia have plans for the ARM architecture to take the throne from X86, and could also lead to them presenting a real challenge to Intel and AMD in the CPU space. Could Nvidia's purchase of ARM lead to a huge change for the future of PC and laptop processors? We also have yet another leak for the RTX 3060 Ti GPU, with new benchmarks giving us another look at the expected performance of the RTX 30 Ampere graphics card. Will AMD's RDNA 2 RX 6700 card be able to compete in this critical area of the GPU market? Finally, we have an update on the availabilty of RX 6000 custom cards, after we saw a paper launch for the top end cards, the RX 6800 XT and RX 6800."

 
member when MIPS was going to take the throne away from x86?

member when PowerPC was going to take the throne away from x86?

member when SPARC was going to take the throne away from x86?

member when Alpha was going to take the throne away from x86?

member when PA-RISC was going to take the throne away from x86?

member when RISC-V was going to take the throne away from x86?
 
Well Apples M1 is nipping on the heals of both AMD and Intel. The M1X could very well take the lead in the 65w space. An iteration or 2 on that paired up with NVidia’s existing IP could very well accomplish the task. ARM is scaling very well in the Datacenter work world and I expect them to aim there first. Having to pair their pretty A100’s with AMD’s EPYC’s has to leave a bit of a dirty taste and I’d imagine they want to correct that sooner than not.
 
According to our exclusive info

"According to this one weird trick".

member when MIPS was going to take the throne away from x86?
member when PowerPC was going to take the throne away from x86?
member when SPARC was going to take the throne away from x86?
member when Alpha was going to take the throne away from x86?
member when PA-RISC was going to take the throne away from x86?
member when RISC-V was going to take the throne away from x86?
member when we didn't take clickbait fanfiction video titles as news
 
Last edited:
member when MIPS was going to take the throne away from x86?

member when PowerPC was going to take the throne away from x86?

member when SPARC was going to take the throne away from x86?

member when Alpha was going to take the throne away from x86?

member when PA-RISC was going to take the throne away from x86?

member when RISC-V was going to take the throne away from x86?
Except when people and the industry said all of that, it wasn't 2020. ;)
 
member when MIPS was going to take the throne away from x86?

member when PowerPC was going to take the throne away from x86?

member when SPARC was going to take the throne away from x86?

member when Alpha was going to take the throne away from x86?

member when PA-RISC was going to take the throne away from x86?

member when RISC-V was going to take the throne away from x86?

The money + aggression/determination of Nvidia is a new factor in all this though. Whether that was at least one of the missing ingredients all those other times remains to be seen.
 
member when MIPS was going to take the throne away from x86?

member when PowerPC was going to take the throne away from x86?

member when SPARC was going to take the throne away from x86?

member when Alpha was going to take the throne away from x86?

member when PA-RISC was going to take the throne away from x86?

member when RISC-V was going to take the throne away from x86?

None of those had the market penetration of ARM nor the resources of NVIDIA. ARM has proven itself as a scalable architecture and NVIDIA will probably push it further at a more brisk pace. If anyone can turn ARM into the next x86 (for desktops because numerically it's already beaten x86), it's NVIDIA. Do I expect ARM to replace my x86 desktop in the next 5 years? No I don't but we might see ARM start making inroads within a decade, especially if NVIDIA can convince MS and the biggest video game publishers to build ARM based engines/games.
 
What else does anyone think Nvidia bought ARM for.

Of course there coming for Intels cheddar.

ARM powers the fastest super computer in the world for now... and it will probably be the most power efficient SC arch for years. (although the new power may be very similar) Yes Nvidia is going to be selling 100% Nvidia ARM super computers sooner then later. Yes they are going to be going after Apple. As I read more reviews for Apples M1 stuff... the more impressed I am, Apples has probably the best laptops in their price bracket now with their M1 machines. A beefed up M1x if the rumors are true are going to be impressive. When Apple gets to version 2... I for sure expect Nvidia will be in control of ARM and probably well on the way to brining a Nvidia CPU to windows ARM. Will they be able to make that work with windows games ? Who knows probably.... I think MS wants it to happen, and they really just need a good chip supplier.
 
Last edited:
No I don't but we might see ARM start making inroads within a decade, especially if NVIDIA can convince MS and the biggest video game publishers to build ARM based engines/games.
Microsoft used PowerPC for the xbox 360 after switching away from x86 in the original xbox... and back to x86 they went

the microsoft surface RT was also arm based, how did that go?

unless microsoft can make their own version of windows 10 with an x86>ARM emulation layer (rosetta2) that's as efficient as apples (and therefore bring over decades of x86 software), and the big guys (lenovo, hp, dell, microsoft) release some arm based hardware that's competitive i just don't see it happening

I see pcs just going away before x86 goes away, ironically replaced by arm devices
 
Yeah this isn't going to be a thing. Never.

Remember when AMD was going to switch to ARM? They smartened the hell up and made a real cpu.
 
unless microsoft can make their own version of windows 10 with an x86>ARM emulation layer (rosetta2) that's as efficient as apples (and therefore bring over decades of x86 software), and the big guys (lenovo, hp, dell, microsoft) release some arm based hardware that's competitive i just don't see it happening

isn’t that what they’re trying to do now, though?
 
Reading some of these posts as well, I just want to point out, we all like to refer to our computers as "fast". They are slow AF. Challenger Hellcat is fast. F22 is fast. Falcon Rockets are fast. Computers don't move. # of operations per second is relative and extremely complex based on the what it is doing, what its designed to do, and our perception of how long we think it should take a computer to do some things. X86 wins all day because the software says. Just like the C language aint going anywhere because it isnt going to be replaced by something better. X86 is the pinacle of bit banging bullshit that computers do. Get back to reality, quantum computing and only quantum computing will replace X86. And when that happens I want to live in a black hole.
 
Reading some of these posts as well, I just want to point out, we all like to refer to our computers as "fast". They are slow AF. Challenger Hellcat is fast. F22 is fast. Falcon Rockets are fast. Computers don't move. # of operations per second is relative and extremely complex based on the what it is doing, what its designed to do, and our perception of how long we think it should take a computer to do some things. X86 wins all day because the software says. Just like the C language aint going anywhere because it isnt going to be replaced by something better. X86 is the pinacle of bit banging bullshit that computers do. Get back to reality, quantum computing and only quantum computing will replace X86. And when that happens I want to live in a black hole.
Quantum is niche and is only applicable is certain scenarios. Python has replaced C as #1 Language. Not saying Python is better, it's easier.
 
Quantum is niche and is only applicable is certain scenarios. Python has replaced C as#1 Language. Not saying Python is better, it's easier.
Python has not replaced C. Python is most popular because its easy. The machines that make all of the other things you like and use every day, do not run Python.
 
Python has not replaced C. Python is most popular because its easy. The machines that make all of the other things you like and use every day, do not run Python.
I know, I said that. C is a poor analogy when you say #1 language in reference to ARM and x86. Almost as bad as your Quantum example.
 
I know, I said that. C is a poor analogy when you say #1 language in reference to ARM and x86. Almost as bad as your Quantum example.
Agree to disagree on this.

Quantum computing isnt likely either, but sure as hell ARM is not replacing X86. Its just not better.
 
Agree to disagree on this.

Quantum computing isnt likely either, but sure as hell ARM is not replacing X86. Its just not better.
Just because you make poor analogies doesn't mean I disagree with your point of view.
 
Reading some of these posts as well, I just want to point out, we all like to refer to our computers as "fast". They are slow AF. Challenger Hellcat is fast. F22 is fast. Falcon Rockets are fast. Computers don't move. # of operations per second is relative and extremely complex based on the what it is doing, what its designed to do, and our perception of how long we think it should take a computer to do some things. X86 wins all day because the software says. Just like the C language aint going anywhere because it isnt going to be replaced by something better. X86 is the pinacle of bit banging bullshit that computers do. Get back to reality, quantum computing and only quantum computing will replace X86. And when that happens I want to live in a black hole.

I think we all just witnessed the most profound post of the millennium....
 
Yeah this isn't going to be a thing. Never.

Remember when AMD was going to switch to ARM? They smartened the hell up and made a real cpu.
I wouldn't say never. I've said before that Nvidia and not Apple can bring ARM to the level of performance that would compete with x86. Will that dethrone x86? No, that's nonsense talk. For one, x86 applications are everywhere and going ARM means a hindrance in compatibility and performance. Secondly, x86 will get better. It always gets better. X86 is in a bad spot right now because Intel had dominated the market and became complacent, and we praise AMD when they catch up to Intel. Now that both companies have reasons to improve their performance, we're going to see x86 get faster again.

If anything ARM will live along side x86 for a long time. Eventually either x86 dies or ARM dies, and that depends on who becomes the most cost effective solution.
 
  • Like
Reactions: travm
like this
member when MIPS was going to take the throne away from x86?

member when PowerPC was going to take the throne away from x86?

member when SPARC was going to take the throne away from x86?

member when Alpha was going to take the throne away from x86?

member when PA-RISC was going to take the throne away from x86?

member when RISC-V was going to take the throne away from x86?
 

Attachments

  • OIP.jpg
    OIP.jpg
    27.1 KB · Views: 0
If anything ARM will live along side x86 for a long time. Eventually either x86 dies or ARM dies, and that depends on who becomes the most cost effective solution.
Is that true, I feel that if x86 stay alive the next 100 years that would not necessarily mean ARM market share on smartphone, tablet, cars, television and so on would become nearly nill, ARM is so omnipresent, I think they are selling over 25 billions of them a year now, that more than 3 by humans on earth in just a year. And for many of those application, it is not obvious shifting to x86 would bring much (there is little legacy software useful for them and so on).

The talk about never (about C, or X86) do sound like really premature, if computing was 1,000, 2,000 year's old, statement like that could make sense, but thinking that in 3,000 year's C/C++ and X86 will still be significant and expressing it as a certain fact is pushing it, imo (or even just by the end of this century).
 
member when MIPS was going to take the throne away from x86?

member when PowerPC was going to take the throne away from x86?

member when SPARC was going to take the throne away from x86?

member when Alpha was going to take the throne away from x86?

member when PA-RISC was going to take the throne away from x86?

member when RISC-V was going to take the throne away from x86?

In many cases those processors did take the throne from x86. All of the processors you mentioned were capable and powerful, and when doing math or transactions they could haul rocks way better than x86.

Intel was worried that a better architecture was going to eat their lunch. By 1995 Intel was so badly wedded to the x86 instruction set they could no longer do a clean slate processor. They still tried - does anyone remember the Intel 960? The Itanium? The Itanium was Intel and HP working together to try to take throne away from Intel's own x86 environment.

What none of those processors could do well is run Windows software. And none of the companies who made software for Windows could afford to do a native port of their product to a hardware platform that was less than 3% of the consumer market and was running on a temporary, unoptimized windowing system (that is, one of the many versions of UNIX).

Things are turning the corner in 2020. Microsoft has most of its operating system separated from the metal. Virtualization, DLL translation layers and software compiled for virtual processes gives Windows a lot of flexibility, and this will grow as we go forward. We've been watching it in the graphics arena for the last 6 years - rather than Microsoft trying to write a wrapper for the nVidia and Radeon drivers, nVidia and Radeon now write their drivers to perform specific DX12 functions. This switch is now maturing in the operating systems, both for Windows and (to a lesser degree) Linux. nVidia isn't going to make an ARM processor that replicates x86 functions, they are going to decide how to develop an ARM processor that best does all the base processes in Microsoft Windows.

That doesn't mean they'll win. But Intel processors no longer have a direct link to the core of Windows, and Windows has moved all of Intel's x86 assemblers to an x86 abstraction layer. At some point all of those custom x86 extensions are going to weigh the processor down. The AMD processors that are currently whupping on Intel don't run all of the x86 extensions in metal - in a lot use cases the AMD processor is more like a RISC processor and the x86 library is virtualized, an ability that AMD acquired when they purchased NexGen.
 
Last edited:
Didn't some company recently announce a CPU that has better performance than AMD & Intel ?

Nvidia should easily be able to acquire that company if needed
 
Oh boy I can't wait for new limitations set out by Nvidia on motherboards.

Do you think their chipsets will compete with nforce as being the hottest thing on the Earth?
 
Oh boy I can't wait for new limitations set out by Nvidia on motherboards.

Do you think their chipsets will compete with nforce as being the hottest thing on the Earth?
Can't be any worse than the extreme market segmentation, and market stagnation, caused by Intel over the last 5-10 years.
At least these will have much better iGPUs in them.
 
Can't be any worse than the extreme market segmentation, and market stagnation, caused by Intel over the last 5-10 years.
At least these will have much better iGPUs in them.
Not nearly as bad as the wild west pre 1366/990fx was.

Where only nforce boards allowed sli and you had to use nvidias sata controllers and Amd chipsets and intel chipsets were relegated to only crossfire but you got to use Intels and AMD's sata controller.
 
Not nearly as bad as the wild west pre 1366/990fx was.

Where only nforce boards allowed sli and you had to use nvidias sata controllers and Amd chipsets and intel chipsets were relegated to only crossfire but you got to use Intels and AMD's sata controller.
mmm, I remember those proprietary days of the mid to late 2000s - late DDR1 to early DDR3 era.
I will never forget how those last-gen nForce chipsets ran hotter than the GTX 480 - those were good times. :D
 
At some point all of those custom x86 extensions are going to weigh the processor down. The AMD processors that are currently whupping on Intel don't run all of the x86 extensions in metal - in a lot use cases the AMD processor is more like a RISC processor and the x86 library is virtualized, an ability that AMD acquired when they purchased NexGen.
Which is it then?
 
In many cases those processors did take the throne from x86. All of the processors you mentioned were capable and powerful, and when doing math or transactions they could haul rocks way better than x86.

Intel was worried that a better architecture was going to eat their lunch. By 1995 Intel was so badly wedded to the x86 instruction set they could no longer do a clean slate processor. They still tried - does anyone remember the Intel 960? The Itanium? The Itanium was Intel and HP working together to try to take throne away from Intel's own x86 environment.

What none of those processors could do well is run Windows software. And none of the companies who made software for Windows could afford to do a native port of their product to a hardware platform that was less than 3% of the consumer market and was running on a temporary, unoptimized windowing system (that is, one of the many versions of UNIX).

Things are turning the corner in 2020. Microsoft has most of its operating system separated from the metal. Virtualization, DLL translation layers and software compiled for virtual processes gives Windows a lot of flexibility, and this will grow as we go forward. We've been watching it in the graphics arena for the last 6 years - rather than Microsoft trying to write a wrapper for the nVidia and Radeon drivers, nVidia and Radeon now write their drivers to perform specific DX12 functions. This switch is now maturing in the operating systems, both for Windows and (to a lesser degree) Linux. nVidia isn't going to make an ARM processor that replicates x86 functions, they are going to decide how to develop an ARM processor that best does all the base processes in Microsoft Windows.

That doesn't mean they'll win. But Intel processors no longer have a direct link to the core of Windows, and Windows has moved all of Intel's x86 assemblers to an x86 abstraction layer. At some point all of those custom x86 extensions are going to weigh the processor down. The AMD processors that are currently whupping on Intel don't run all of the x86 extensions in metal - in a lot use cases the AMD processor is more like a RISC processor and the x86 library is virtualized, an ability that AMD acquired when they purchased NexGen.
And the 960 like the 29K were microcontrollers for things like printers and specialty hardware. There were some DOD contracts (still) but they were not general use CPUs.

Point being. While both Intel and AMD have dabbled in non x86 architectures both companies keep coming back. Each time they do an x86 design it becomes more and more advanced and complex.....which is what ARM will run into more and more and is doing now. Its not getting simpler or less complex to do better than x86 and other options in certain spaces. It is getting more complex or getting wedded to ASICs. Which is familair ground.....right? X86 used to have the FPU separate as an x87. The memory controller? Separate. 16 the 32 then 64. SSE, SSEII, DNow!, etc.

For ARM to win it will have to get bigger, better, and more complex. Not smaller, less functional, and less powerful.
 
Didn't some company recently announce a CPU that has better performance than AMD & Intel ?

Nvidia should easily be able to acquire that company if needed

I was referring to Nuvia.
If nVidia acquire them, then can rename themselves to nuVidia

https://www.extremetech.com/computi...faster-than-zen-2-while-using-much-less-power

Basically: some former Apple and Google processor engineers decided to found their own company to do ARM processors for the server space.

Anandtech decided to post their latest ad as a news article.

The company claims that their processors will be "+50% to +100% peak performance of the other cores, either for the same power as other Arm cores or for a third of the power of x86 cores." Then they shared a single benchmark based on a simulation of actual performance, leaving out the top part of it for some reason.

https://www.anandtech.com/show/1596...0-st-performance-over-zen-2-for-only-33-power

I like seeing ARM showing up in more spaces, but this felt so much like an ad designed to drum up investment capital for the "next big thing". I know simulations are a thing but why do a simulation of your OMGWTFast processor, telling everyone how fast it will be, and then leave out the part of the simulated benchmark that's supposed to be so awesome?

TL;DR: more ARM. Discuss.
 
Back
Top