What if Nvida bought AMD next year????

  • Thread starter Man-wai Chang ToDie (33.6k)
  • Start date
P

Paul

Man-wai Chang ToDie (33.6k) said:
.... AMD has no good reason to continue, does it?

They're a fabless design house, roughly equivalent
in function to Nvidia. They're going to have to
"sink or swim" on their own, as no one will rescue
them. The only asset they have, is their staff.

Paul
 
E

Ed Medlin

Conor said:
Are you aware that AMD and nVidia are the same company now?


--
Conor

I only please one person per day. Today is not your day. Tomorrow isn't
looking good either. - Scott Adams

AMD bought ATI..........not Nvidia.


Ed
 
D

Dave

Man-wai Chang ToDie (33.6k) said:
.... AMD has no good reason to continue, does it?

Are you SERIOUS?!? (typing this question on my newly built Phenom 9850 / HD
4830 Crossfire system which neighbors and friends are drooling over...)

Why would you think AMD has no good reason to continue? Over many years,
AMD/Intel have swapped back and forth, as far as performance leader and
leader in bang/buck goes. That is still true.

I know AMD/ATI have not been as quick in coming out with new chips lately,
and it to some degree seems like they are now trying to play "catch up" with
Intel.

But don't count AMD/ATI out. The near future is clearly multi-core, as far
as CPUs and GPUs goes. After some minor stumbling, AMD seems to be headed
in the right direction, and is STILL price competitive with Intel for many
builds (depending on what you are aiming for as a finished result, of
course)

Looking farther into the future, AMD/ATI is in a MUCH better position for
long-term success than Intel or nvidia. The future is to integrate the GPU
and CPU into one chip, which will basically be a multi-core GPU with some
firmware to allow it to function as a multi-core CPU. The CPU as we now
know it will soon be obsolete. Yeah, I said obsolete. The GPU is getting
so powerful that very soon a CPU will be redundant / unnecessary. At that
point, the CPU functions will be merged into the GPU. If you think Intel
will be able to accomplish that gracefully on their own, you are wrong.
(Any good high-end gaming INTEL chip graphics cards on the market? Or even
rumored to come out soon?) AMD already has ATI. So putting out a high-end
GPU chip that also functions as a CPU will be extremely easy for AMD,
obviously. The only way Intel can remain long-term competitive is to buy
nvidia.

AMD is more likely to prosper far into the future, unless Intel buys
vidia. -Dave
 
P

Paul

Dave said:
Are you SERIOUS?!? (typing this question on my newly built Phenom 9850
/ HD 4830 Crossfire system which neighbors and friends are drooling
over...)

Why would you think AMD has no good reason to continue? Over many
years, AMD/Intel have swapped back and forth, as far as performance
leader and leader in bang/buck goes. That is still true.

I know AMD/ATI have not been as quick in coming out with new chips
lately, and it to some degree seems like they are now trying to play
"catch up" with Intel.

But don't count AMD/ATI out. The near future is clearly multi-core, as
far as CPUs and GPUs goes. After some minor stumbling, AMD seems to be
headed in the right direction, and is STILL price competitive with Intel
for many builds (depending on what you are aiming for as a finished
result, of course)

Looking farther into the future, AMD/ATI is in a MUCH better position
for long-term success than Intel or nvidia. The future is to integrate
the GPU and CPU into one chip, which will basically be a multi-core GPU
with some firmware to allow it to function as a multi-core CPU. The CPU
as we now know it will soon be obsolete. Yeah, I said obsolete. The
GPU is getting so powerful that very soon a CPU will be redundant /
unnecessary. At that point, the CPU functions will be merged into the
GPU. If you think Intel will be able to accomplish that gracefully on
their own, you are wrong. (Any good high-end gaming INTEL chip graphics
cards on the market? Or even rumored to come out soon?) AMD already
has ATI. So putting out a high-end GPU chip that also functions as a
CPU will be extremely easy for AMD, obviously. The only way Intel can
remain long-term competitive is to buy nvidia.

AMD is more likely to prosper far into the future, unless Intel buys
vidia. -Dave

You did hear about how AMD has split into two companies, one
part having the fab, and the other part having the rest of the
company ? In the near term, it means business as usual, as the
fab part will continue to make processors. But in the long term,
the technical issue will be whether AMD shops around for fab
capability, for example going to TSMC. And if they do, then the
question would be, how do you build a "fast" processor, when
you don't own the fab. The thing about having your own fab,
is you can do lots of tweaking, and you own the whole business.
When you're fabless, the odds are greater that your products
will suck. For example, if I went to TSMC, I might be able to
build a 2GHz processor, whereas if I tweaked my own stuff,
I could build a 3GHz processor.

AMD has a world of possibilities before it, and should
select a strategy to suit their new "asset light" corporation.
That means they cannot continue to do exactly what they've
been doing, so expect the business decision to split the
company, to have some side effects. Just how crazy they get,
will depend on what their management has been smoking. And
how long the cash lasts.

Paul
 
M

Matt

Paul said:
They're a fabless design house, roughly equivalent
in function to Nvidia. They're going to have to
"sink or swim" on their own, as no one will rescue
them. The only asset they have, is their staff.

Paul


Oh, I'm sure they have a lot of secret software and data and prototyping
and test equipment that helps them do what they do, besides patents and
copyrights.
 
M

Matt

Paul said:
And if they do, then the
question would be, how do you build a "fast" processor, when
you don't own the fab. The thing about having your own fab,
is you can do lots of tweaking, and you own the whole business.
When you're fabless, the odds are greater that your products
will suck. For example, if I went to TSMC, I might be able to
build a 2GHz processor, whereas if I tweaked my own stuff,
I could build a 3GHz processor.


Are you saying that AMD can't build its own prototypes? That they have
no prototyping fab?
 
P

Paul

Matt said:
Are you saying that AMD can't build its own prototypes? That they have
no prototyping fab?

A prototype is no good, unless it matches the intended
production process. So if AMD had a small fab still under
their control, it wouldn't match TSMC.

Paul
 
D

Dave

You did hear about how AMD has split into two companies, one
part having the fab, and the other part having the rest of the
company ? In the near term, it means business as usual, as the
fab part will continue to make processors. But in the long term,
the technical issue will be whether AMD shops around for fab
capability, for example going to TSMC. And if they do, then the
question would be, how do you build a "fast" processor, when
you don't own the fab.

You think AMD has always manufactured all of their own chips? IBM and
others have been pumping out AMD designed chips for AMD for (from my memory
anyway) decades. In fact, until you mentioned it here, I wasn't even aware
that AMD had significant fab capacity owned directly by AMD.

If AMD is comfortable letting go of their in-house fab, that tells me that
AMD's fab partners have proven extremely reliable and cost-effective for
AMD.

If anything, this further strengthen's AMD's future profitability. -Dave
 
P

Paul

Dave said:
You think AMD has always manufactured all of their own chips? IBM and
others have been pumping out AMD designed chips for AMD for (from my
memory anyway) decades. In fact, until you mentioned it here, I wasn't
even aware that AMD had significant fab capacity owned directly by AMD.

If AMD is comfortable letting go of their in-house fab, that tells me
that AMD's fab partners have proven extremely reliable and
cost-effective for AMD.

If anything, this further strengthen's AMD's future profitability. -Dave

I've worked for a company which had its own fabs. I am familiar with
the advantages and disadvantages of having your own fab. It was great,
being able to drive over to the fab, and talk to the people actually
designing the CMOS process and ask questions about its fundamental
properties.

Processor design, sometimes involves custom logic cell design. Sure,
you can get somebody at IBM to do it for you (IBM has designed their
own processors), but the thing is, IBM is very expensive for stuff
like that. If you're going with IBM, bring a "fat wallet".

It is not impossible to do it, it is just inconvenient. And there'd be
a difference between IBM and TSMC, both in terms of mindset and
capabilities. TSMC might be cheaper, but would they necessarily
bend over backwards to help you ? (My company worked with a number
of different external fabs, and with some of them, you could tell
they just weren't interested in keeping you happy.)

Being fabless is fine, if you were always fabless. Your design and
design process flow, evolve over time, to work smoothly with
external companies. But for a company that always had its own fab,
it can come as a shock, and for a company like AMD, they really
cannot afford to drop the ball. So whatever they do in response
to the decision to split the company, they cannot afford to make
mistakes.

For a company that is struggling to remain competitive, this
isn't making it any easier. It's a distraction.

Paul
 
M

Matt

Paul said:
A prototype is no good, unless it matches the intended
production process. So if AMD had a small fab still under
their control, it wouldn't match TSMC.

Paul


Maybe they have folks onsite at TSMC and they transmit their
experimental designs electronically?
 

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments. After that, you can post your question and our members will help you out.

Ask a Question

Top