News Intel and AMD forge x86 ecosystem advisory group that aims to ensure a unified ISA moving forward

Admin

Administrator
Staff member

bit_user

Titan
Ambassador
It's a little hard to see what's in this for Intel, but my best guess is that they caught flak from big customers for reversing course on AVX-512 and doing AVX10.

Also, if they want to stave off the assault from ARM and RISC-V, then there's an incentive for getting these ISA extensions deployed and utilized in the field, on a more aggressive time table. Having a situation where AMD drags its feet on a certain ISA extension could mean that software developers are less likely to utilize it, and that ultimately hurts Intel and undermines the point of even having the extension.

The article said:
Intel and AMD are the only two primary x86 architecture licensees that build new processors in high volumes, creating a duopoly.
VIA is another licensee and has been engaged with Zhaoxin to design and build new x86 CPUs. Given the potential size of the Chinese market, I think they might qualify as high-volume.
 
Last edited:

Kamen Rider Blade

Distinguished
Dec 2, 2013
1,457
1,002
21,060
Maybe we can get Intel to allow AMD in on the ATX Spec.

There are a few things I want to modify as well.

First, we need to get rid of 12VHPWR & 12v-2x6 and kill it.

Start over from scratch to make a BETTER/SAFER High Wattage Plug.

The current plug is just badly architected with not enough safety margin from the outset.

Also, the Backside mounted MoBo Connectors are dumb.

We need to nip that in the bud.

We need "Right-Angle Receptacles & Plugs to become the standard.

We also need to bring back all our PCIe Rear Add-In slots.

M.2 Connectors can be adapted to PCIe slots, not the other way around.

Don't allow MoBo makers to force M.2 Connectors by default on the MoBo to eat away our PCIe slots.
 
Last edited:

ekio

Reputable
Mar 24, 2021
142
158
4,760
Amd and Intel KNOW x86 is a dead technology that is kept alive through bandaids and they know their value is bound to keeping it alive as long as they can.

The healthy way of moving on would be to embrace arm or risc-v with advanced x86 to modern isa software translators for the transition phase, but that would mean the end of the duopole…
 

Kamen Rider Blade

Distinguished
Dec 2, 2013
1,457
1,002
21,060
Amd and Intel KNOW x86 is a dead technology that is kept alive through bandaids and they know their value is bound to keeping it alive as sling as they can.

The healthy way of moving on would be to embrace arm or risc-v with advanced x86 to modern isa software translators, but that would mean the end of the duopole…
x86 is 4 Life! & 4 EVAR!

But in all seriousness:

The ARM vs x86 debate largely doesn't matter. It's outdated.
Conclusion: Implementation Matters, not ISA

Jim Keller agrees that the ISA doesn't really matter either
 

Kamen Rider Blade

Distinguished
Dec 2, 2013
1,457
1,002
21,060
Amd and Intel KNOW x86 is a dead technology that is kept alive through bandaids and they know their value is bound to keeping it alive as long as they can.

The healthy way of moving on would be to embrace arm or risc-v with advanced x86 to modern isa software translators for the transition phase, but that would mean the end of the duopole…
What we need is for nVIDIA & Qualcomm to be allowed to buy x86 Licenses and join the x86 Family of CPU makers.
 

bit_user

Titan
Ambassador
But in all seriousness:

The ARM vs x86 debate largely doesn't matter. It's outdated.
I have a lot of respect for most of what they publish, but the author clearly didn't do his research, as he wasn't even aware of APX.

Jim Keller agrees that the ISA doesn't really matter either
I think he's been out of the game, for too long. Either that, or he felt it would be uncouth to undercut his former employers (AMD and Intel), if he started badmouthing their ISA. But, when time came for him to integrate CPU cores into Tenstorrent's products, did he try to license IP from AMD or Intel? No. He first went to ARM, and then decided RISC-V was better for the flexibility it gave him.

What we need is for nVIDIA & Qualcomm to be allowed to buy x86 Licenses and join the x86 Family of CPU makers.
At this point, I'm sure they wouldn't even bother if they could.
 
  • Like
Reactions: artk2219

bit_user

Titan
Ambassador
Amd and Intel KNOW x86 is a dead technology that is kept alive through bandaids and they know their value is bound to keeping it alive as long as they can.
As long as there's demand, why wouldn't they keep developing it?

The healthy way of moving on would be to embrace arm or risc-v
AMD has built ARM CPUs in the past (indeed, Jim Keller was even working on one - the K12) and is rumored to be doing it again. Intel has been dabbling with RISC-V, but it's unclear how seriously.

with advanced x86 to modern isa software translators for the transition phase, but that would mean the end of the duopole…
These tools already exist. AMD and Intel don't need to develop them.
 
  • Like
Reactions: artk2219

acadia11

Distinguished
Jan 31, 2010
968
31
19,010
In other words they'll get close but stay out of ARMs reach!

All puns intended who ever stated the ARMs race with x86s is an out-dated way of looking at the CPU world today has got the right idea. CPU's are more complex than that today and "x86" and
"ARM" have evolved into architectures that adopt solutions that are similar and novel to today's problems. Not the rudimentary and outdated modes of CISC and RISC computing definitions that worked 40-50 years ago, they mix and match concepts now.
 
Last edited:
  • Like
Reactions: artk2219

DS426

Prominent
May 15, 2024
278
207
560
Amd and Intel KNOW x86 is a dead technology that is kept alive through bandaids and they know their value is bound to keeping it alive as long as they can.

The healthy way of moving on would be to embrace arm or risc-v with advanced x86 to modern isa software translators for the transition phase, but that would mean the end of the duopole…
x86 is dead? Prove it -- when did this happen?

No, friend. x86 has stiff competition and needs to innovate. Apparently they've realized this, otherwise they wouldn't be collaborating on x86-ng (next generation).
 
  • Like
Reactions: artk2219
This mostly seems like everyone decided to get ahead of the more significant proposals for x86 modification. I can't imagine something like x86S ever happening without AMD and Intel being on the same page. It also seems like the best path forward for continuing x86 with no obvious disadvantages to those involved.
 
  • Like
Reactions: artk2219

abufrejoval

Reputable
Jun 19, 2020
615
454
5,260
I'm pretty sure the hyperscalers helped push those love birds together: they want choice and weren't happy about Intel trying to force them back into their camp via ISA extensions that are needed to keep up with RISC-V and ARM (while they are busy work on option C).

Now only consumers need to do the same when it comes to having new hardware forced on them, unless they really deliver measurable value instead of hostile agents presented as empty co{m}p[i}lot promises.
 

NinoPino

Respectable
May 26, 2022
496
310
2,060
x86 is 4 Life! & 4 EVAR!

But in all seriousness:

The ARM vs x86 debate largely doesn't matter. It's outdated.
Is outdated because the ISA relevance was demonstrated a lot of times.
If you want a demonstration look at the actual superiority of ARM versus x86 when power usage matters.

Jim Keller agrees that the ISA doesn't really matter either
Not at all. Jim Keller in the famous 2021 interview with Anandtech said also this :

"So fixed-length instructions seem really nice when you're building little baby computers, but if you're building a really big computer, to predict or to figure out where all the instructions are, it isn't dominating the die. So it doesn't matter that much."

"So if I was just going to say if I want to build a computer really fast today, and I want it to go fast, RISC-V is the easiest one to choose."

"Instructions (sets) that have been iterated on, and added to, have too much bloat."

And recently he said also that the decision of AMD, of dropping the ARM frontend from the final Zen project was a big mistake.

Another demonstration of the importance of a good ISA is Intel's recent effort to streamline the x86 ISA adding 3 operand instructions, conditional prefixes and dropping legacy x86.

And all of this leaving the RISC vs CISC debate to the history because all modern x86 are RISCs with x86 bloatware inside.
 

danny009

Honorable
Apr 11, 2019
521
38
10,920
Great to see some good news about tech for once and not AI crap. I do love x86 it is great and honestly ARM is just a joke like Windows Phone. Dislike my opinion please
 
  • Like
Reactions: artk2219

bit_user

Titan
Ambassador
"ARM" have evolved into architectures that adopt solutions that are similar and novel to today's problems. Not the rudimentary and outdated modes of CISC and RISC computing definitions that worked 40-50 years ago, they mix and match concepts now.
ARM is still register-to-register and has fixed-size instruction words. It's much more RISC than not.
 
  • Like
Reactions: artk2219
x86 is 4 Life! & 4 EVAR!

But in all seriousness:

The ARM vs x86 debate largely doesn't matter. It's outdated.


Jim Keller agrees that the ISA doesn't really matter either
Yeah x86 isn't going anywhere anytime in the next few decades. Binary compatibility for 99% of desktop software cements it.

Plus nobody has made a real x86 CPU in two decades.
 
  • Like
Reactions: artk2219