Now that Intel's Getting Into GPU Market, Thoughts on an NVIDIA & AMD Merger?

Aardvarck

Limp Gawd
Joined
Sep 21, 2004
Messages
231
Sure, it's another couple of years, but what we'll end up seeing is Intel get up to the level of AMD as far as being able offer both CPUs and GPUs as part of their product set.

Is a merger between Nvidia & AMD even a remote possibility in this case?

I would assume that their joining forces would give Intel quite a run for their money and only serve to empower them and make them much stronger.

This is purely a thought. I'm mainly interested in getting everyone's expertise as to how this could or could not be a reality.
 
Sure, it's another couple of years, but what we'll end up seeing is Intel get up to the level of AMD as far as being able offer both CPUs and GPUs as part of their product set.

Is a merger between Nvidia & AMD even a remote possibility in this case?

I would assume that their joining forces would give Intel quite a run for their money and only serve to empower them and make them much stronger.

This is purely a thought. I'm mainly interested in getting everyone's expertise as to how this could or could not be a reality.
\

I'm not sure if that will happen somewhere down the line, but i sure hope it doesn't. I hope there's like 8 companies selling dedi gpus instead of just 3. That way those companies cant overcharge for "RnD" since it is actually quite cheap to make the actual product.
 
Highly doubt that will happen and if it does something about Intel's ownership of the x86 architecture not being transferred even in the event of a buyout/merger would most likely prevent an AMD/Nvidia "merger" from being able to compete with Intel in the x86 market, unless AMD is buying NVIDIA which they can't afford.

...or something along those lines.
 
Only if Intel completely decimates both AMD and Nvidia for a few years... and even then probably not.
 
I highly doubt Intel is going to make a high end performance part. It will be low to mid range at best.
 
I get the feeling the Intel GPU will be an obscure part made to fill a part of the market that almost nobody cares about. Kind of like their Optane ssds.

Pretty much, with current push for ai/deep learning, gpu's are far more important so it makes sense for Intel to move into that market. I can't see them bothering to compete with amd or nvidia on the consumer side.


edit: type, stupid phone autofill.
 
Last edited:
I get the feeling the Intel GPU will be an obscure part made to fill a part of the market that almost nobody cares about. Kind of like their Optane ssds.

At least those are the fastest out there and based on a real improvement in design, latency is king in the end. Also byte-addressed is so much better than having to write entire blocks and deal with garbage collection etc, that was only a stop-gap to accommodate existing flash tech in the endless march towards better storage.

Anyways, a safe bet intel gpus will still be shit for a good while longer.
 
You can't merge with AMD they lose their access to the x86 license.

Maybe this is a dumb question and shows some ignorance... but didn't AMD invent the x64 instruction set (1st) which is pretty much right in line with the Intel x64 set? Can you run a PC without x86 and only on x64?

I know there are still millions of 32b OS's out there, but the majority now, and the vast majority of new sales are 64b.
 
Maybe this is a dumb question and shows some ignorance... but didn't AMD invent the x64 instruction set (1st) which is pretty much right in line with the Intel x64 set? Can you run a PC without x86 and only on x64?

I know there are still millions of 32b OS's out there, but the majority now, and the vast majority of new sales are 64b.

I'll let someone with far more programming knowledge than me answer, but I will say that you can run PC's on non-native instruction sets but there are two problems. 1.) You run into compatibility problems as legacy software can't run on the non-native hardware without emulation. This limits the software availability considerably and there is only one way around that. 2.) You can emulate the old instruction sets through software. However, there is always a performance hit doing this. A substantial one in most cases. This is why Intel's IA64 instruction set and Itanium never saw real success outside of a single niche.
 
AMD and Intel both invented 64-bit architectures. AMD invented their architecture as an extension to x86. Intel cross-licensed the x64 architecture as it allowed them to make 64-bit chips with the x86 instruction set.
 
AMD and Intel both invented 64-bit architectures. AMD invented their architecture as an extension to x86. Intel cross-licensed the x64 architecture as it allowed them to make 64-bit chips with the x86 instruction set.

Intel had it's own IA64 implementation which was implemented on it's Itanium CPU's first which is separate from x86-64 / EM64T (now known as Intel 64).
 
Intel had it's own IA64 implementation which was implemented on it's Itanium CPU's first which is separate from x86-64 / EM64T (now known as Intel 64).

Which is precisely why they cross-licensed AMD's architecture. Itanium wasn't called Itanic for no reason...
 
Maybe this is a dumb question and shows some ignorance... but didn't AMD invent the x64 instruction set (1st) which is pretty much right in line with the Intel x64 set? Can you run a PC without x86 and only on x64?

I know there are still millions of 32b OS's out there, but the majority now, and the vast majority of new sales are 64b.
Technically it is a cross license AMD gets x86 and Intel x86_64. The way it could be done would require special software. Since Windows is not open source it would require a lot of work.

If AMD wishes they could make a trail run (64 bit only) on a Linux distribution to see how much of a black whole the x86 compatibility really is.
Not to sure that x86_64 is a complete replacement.
 
IA-64 remains interesting because the architecture relies on smart compilers, and it's really interesting that such compilers are only now potentially feasible for building a broad sample of products with the availability of machine learning.

AMD64 was a brilliant move on their part- double the registers and their width, nothing particularly difficult, but it did enable >2GB of RAM for a single process and >4GB of RAM for the system at a time when those walls were starting to become apparent. Oh, and Intel was still pushing Netburst to places that it simply wouldn't go while AMD was running their evolved K7 that was meant to square off with Intel's brief detour away from P6, which gave them the two aces they needed, and that carried them all the way until they pulled their own Netburst (higher clocks, longer pipelines, lower IPC...) with Bulldozer :D.

Regardless, in response to the OP I'd always thought that we'd get Intel picking up 'ATi' and Nvidia picking up AMD, with Intel concurring the x86 license transfer in the interest of avoiding anti-trust actions.
 
I'm thinking the Intel GPU's will be geared more towards business use and sold to OEM's providing PC's with a bit more graphical power. They'll bundle a CPU and GPU for a discount. It's not in Intel's interest to go after high end GPU market at this time. Low to low/mid, with slightly better performance than an APU would be my guess. And they'll sell the piss out of them.

A merger between Nvidia and AMD is not going to happen.
 
\

I'm not sure if that will happen somewhere down the line, but i sure hope it doesn't. I hope there's like 8 companies selling dedi gpus instead of just 3. That way those companies cant overcharge for "RnD" since it is actually quite cheap to make the actual product.

Oversaturation is not the answer. The 90's proved that. Too many manufacturers, all around the same performance, all vying for your money. They cannot all stay afloat.

Use the auto industry as an example. The big three not only had 10 other companies to compete with, they also had themselves to compete with. AMD said it recently, that they went from offering 50 versions of a cpu to just a few. Oversaturation is bad.
 
I'm thinking the Intel GPU's will be geared more towards business use and sold to OEM's providing PC's with a bit more graphical power. They'll bundle a CPU and GPU for a discount. It's not in Intel's interest to go after high end GPU market at this time. Low to low/mid, with slightly better performance than an APU would be my guess. And they'll sell the piss out of them.

They'll likely start low- they pretty much have to. But when they do it, and they get the architecture right and get the drivers right- and they're already pretty close!- GPUs are very easy to scale up.

A merger between Nvidia and AMD is not going to happen.

I don't think it will happen either, but AMD could easily get far enough behind to no longer be sustainable, and really, Nvidia is the second-best company next to Intel in this market.
 
This isn’t Intel’s first forway into the dGPU market. Their last one failed pretty bad.

I really hope there aren’t any mergers. We need more competition. Honestly, I wish NVIDIA somehow got an X86 license.
 
I really hope there aren’t any mergers. We need more competition.

Nvidia pushing AMD's x86 tech to the limit with Intel pushing AMD's GPU tech to the limit- wouldn't that be more competitive than Intel and Nvidia dominating their markets and AMD occasionally almost catching up?
 
You can't merge with AMD they lose their access to the x86 license.

That really no longer applies as Intel uses AMD tech as well. Intel might make a bit of fuss if it really hated the partner that AMD was acquired by. They hashed a bit of this out when AMD divested itself of Fabs which they originally had to have to keep the license.If someone like Samsung bought AMD they would not get in the way of that, however if Nvidia were to buy AMD it would be a huge deal and likely not go through due to anti trust and grievances by Intel.
 
That really no longer applies as Intel uses AMD tech as well. Intel might make a bit of fuss if it really hated the partner that AMD was acquired by. They hashed a bit of this out when AMD divested itself of Fabs which they originally had to have to keep the license.If someone like Samsung bought AMD they would not get in the way of that, however if Nvidia were to buy AMD it would be a huge deal and likely not go through due to anti trust and grievances by Intel.
nope
 
tenor.gif
 

You do realize that x86 patents have expired? Now there are specific add-on's that are still under patents that they both cross license to each other, trust me a deal would be hammered out, just like it was when AMD ditched the Fab's despite that being a clause for termination. You speak with a absolute when it's already been proven false. As times advance and things get more difficult it's not uncommon to share tech and cross license, however it makes it difficult to enforce clauses in agreements when the other guy can hurt just as bad by pulling a license from you as well. Before x64 I would think you were right but now I just see them hammering out a deal they both could live with.
 
nVidia doesn't have much reason to buy AMD with or without Intel entering the graphic card market. Or at-least they would have no more business reason than they have now.

nVidia surely doesn't think it needs AMD to compete with intel in the graphic card segment. So what possible reason is there to buy AMD? To compete in the processor segment? 1. That's not a lucrative segment in the future with IoT, cloud, blockchain and AI. and 2. If nVidia wants to compete in the processor segment, they have as much reason to by AMD now as at any other time

AMD just can't afford nvidia
 
AMD could spin off just RTG - no reason x86/CPU business even needs to come into play here.

It is interesting though - nVidia would be interested in AMD for the x86/CPU end of the deal (which they can't get). Intel would be interested in AMD for RTG (which they already have the beginnings of a working relationship with on that end, not to mention previous dealings with x86/x64). Intel would be much more likely partner, but I don't see it occurring. AMD as a company buyout would only occur, in my opinion, to a foreign player looking to an American business front and toehold into the electronics market - say Bitmain or Huawei something of the sort. The local big boys would likely just let AMD go into insolvency and then look to pick up the patents/IP in the resulting firesale.
 
You do realize that x86 patents have expired? Now there are specific add-on's that are still under patents that they both cross license to each other, trust me a deal would be hammered out, just like it was when AMD ditched the Fab's despite that being a clause for termination. You speak with a absolute when it's already been proven false. As times advance and things get more difficult it's not uncommon to share tech and cross license, however it makes it difficult to enforce clauses in agreements when the other guy can hurt just as bad by pulling a license from you as well. Before x64 I would think you were right but now I just see them hammering out a deal they both could live with.

It is not just about patents. It is more complicated then what you describe here. And you should google it if not just try and find the discussion somewhere on here.
You can find the details of the old deal between Intel and AMD and the current deal is still not public.
 
Nvidia pushing AMD's x86 tech to the limit with Intel pushing AMD's GPU tech to the limit- wouldn't that be more competitive than Intel and Nvidia dominating their markets and AMD occasionally almost catching up?


Assuming NVIDIA has good X86 engineers. It would be more like AMD using NVIDIA tech in their line while Intel devolps their own GPU.

Best thing that can happen would be NVIDIA somehow getting an X86 license, and Intels GPU being a smash hit. Then we have 3 CPU makers, and 3 GPU makers .
 
Back
Top