Summary of Benchmark Results: 3900X vs 9900k

The way i look at bye intel or by amd.either processor will serve its purpose.
 
I'm just glad that AMD is back in the mix. I didn't think I would see the day, that Intel would be put in 2nd place. Especially after the Blowdozer.
 
HubDoubleDub said:
that Intel would be put in 2nd place
Click to expand...

AMD's going to have to buy out TSMC, pay UMC to upgrade, and hook up with Samsung and probably TI if they want a chance at putting Intel into 2nd place. Being fabless means that they're going to be number two for a long, long while, and they only have a limited time to capitalize on Intel's 10nm mistakes.
 
IdiotInCharge said:
AMD's going to have to buy out TSMC, pay UMC to upgrade, and hook up with Samsung and probably TI if they want a chance at putting Intel into 2nd place. Being fabless means that they're going to be number two for a long, long while, and they only have a limited time to capitalize on Intel's 10nm mistakes.
Click to expand...

AMD is better off being content in 2nd place as long as they can maintain a close second place.

Had that 10nm mistake happened to AMD, they would be bankrupt
 
RamonGTP said:
Had that 10nm mistake happened to AMD, they would be bankrupt
Click to expand...

You mean, had it happened to TSMC ;)

That's the challenge for AMD that I'm highlighting. Intel brings in the big bucks and maintains the marketshare because they can independently supply the market. When they screw up, they screw up, but it's just them. AMD has a supply chain that's entirely outside of their control and affected by market forces that they may have trouble adapting to in the future.

To further the idea: investing in AMD products for smaller scale stuff or stuff that's agnostic is fine, but for places that run applications that need to be tuned to the architecture they're running on, AMD is still in the 'proving' stages, not just in terms of technology but also in terms of resiliance.

At some point, they likely need to either a) get their own fabs again or b) merge with a company that does, with Samsung being a good example of a candidate from an external perspective.
 
IdiotInCharge said:
You mean, had it happened to TSMC ;)

That's the challenge for AMD that I'm highlighting. Intel brings in the big bucks and maintains the marketshare because they can independently supply the market. When they screw up, they screw up, but it's just them. AMD has a supply chain that's entirely outside of their control and affected by market forces that they may have trouble adapting to in the future.

To further the idea: investing in AMD products for smaller scale stuff or stuff that's agnostic is fine, but for places that run applications that need to be tuned to the architecture they're running on, AMD is still in the 'proving' stages, not just in terms of technology but also in terms of resiliance.

At some point, they likely need to either a) get their own fabs again or b) merge with a company that does, with Samsung being a good example of a candidate from an external perspective.
Click to expand...

I'm not sure they really need to. Nvidia does fine as a fabless company. The key for AMD if they remain fabless is to always have a backup plan should Fab Company #1 bail on them, fail, or try to strongarm them on price. In other words, have a uarch that is easily ported to a different process. For the most part, Zen has been pretty successful from this regard - and splitting I/O and core chiplets makes it even easier, as they can (and do) source different parts from different Fab companies. When GloFo bailed on 7nm, they could go to TSMC (in fact, I wonder if that was the plan all along).

When AMD had their own fabs, they were always behind Intel on process (except once where it was a functional tie). Now they can keep up with Intel on process by using competing fabs. But I agree that it IS a risk, and one AMD needs to closely manage. I just don't think they have to merge with a fab company or anything. They can play the game like Nvidia has.
 
Last edited:
DuronBurgerMan said:
Nvidia does fine as a fabless company.
Click to expand...

This is a technically valid but perhaps not broadly applicable point- while Nvidia is fabless, they're also ordering from Samsung, and comparing GPUs to the desktop CPU market seems a bit like comparing apples to potatoes. Especially when the comparison is with Intel. I'll bet that Nvidia has more influence with TSMC than AMD does, and that Apple has more influence than Nvidia- and those are just companies that we can name from public knowledge.

AMD has to compete with Intel while also competing with other companies for fab allocations.

DuronBurgerMan said:
In other words, have a uarch that is easily ported to a different process. For the most part, Zen has been pretty successful from this regard - and splitting I/O and core chiplets makes it even easier, as they can (and do) source different parts from different Fab companies.
Click to expand...

I'll absolutely agree that AMD has been successful in approaching their architecture design in a way that lets them hedge their bets with respect to ordering production runs. Whatever the myriad of technical limitations that CPU segmentation into chiplets imposes on Zen, they've done a great job in countering the real-world effects with current software and workloads.

DuronBurgerMan said:
When AMD had their own fabs, they were always behind Intel on process (except once where it was a functional tie).
Click to expand...

AMD never seemed to have Intel's execution. As soon as they messed up they were run over; the same happened to 3Dfx by Nvidia. Intel and Nvidia don't always have products that are superior across the board, but they always have products that are both competitive and desirable, and they always have something new on the way.

DuronBurgerMan said:
I just don't think they have to merge with a fab company or anything. They can play the game like Nvidia has.
Click to expand...

Referring back to the first point- AMD absolutely can play fabs against each other, however, the question as to how much capacity there is for them to eat up versus what Intel is able to ship remains. I bring up the idea of a merger because competing with Intel successfully will mean that manufacturers like Samsung and TSMC are going to have to build out more capacity specifically for AMD.

And if they're building to accommodate AMD, they're likely going to want more than the promise of future orders. A merger would bring technology to both sides, and a mega-conglomerate like Samsung could allow AMD to scale their technology and production far faster than AMD could do on their own.
 
AMD was always playing this race from behind. Always. Intel plays with the home field advantage in every game. And by that I don't mean the stuff the AMD shills spew from time-to-time, that there is some big Intel conspiracy or something.

No, it's just a case of Intel creating the x86 market, Intel having the cash to execute consistently, without having to rely on some brilliant wizard in the basement to bail them out of their troubles. And, of course, Intel having the capacity to make a mistake and recover from it.

When Athlon came out, AMD pushed harder, and executed more consistently than they ever have. Again with Athlon 64, and Athlon X2. They came within a hair of taking the market share crown from Intel. Less than a 5% difference! But then Phenom (I and II) just couldn't keep up. AMD ran out of steam, and Intel had deep enough pockets - and multiple design teams - to generate new ideas, and they finally leapfrogged AMD. They had the cash, talent pool, and market gravitas to win even when they lost ground for a time.

It's funny, because after that AMD just felt like they were tired, used up, old. Rehashing old designs, slowly losing their place. Demoralized or something, I don't know. And then Bulldozer. Lol. What a joke that was.

AMD is making their play again. My prediction? They ride this Zen train for a while, and really make Intel sweat for a few years. With the improved financials, maybe they can extract their GPU position from the toilet, too, and give Nvidia some competition again. But in 4 or 5 years, they'll run out of steam again, and Intel will still be in the dominant position, and I don't think that will ever change so long as x86 remains a thing.

OTOH, it doesn't have to. IMHO, AMD's real market function is to keep Intel honest, and light a fire under their ass occasionally, whenever they get too comfy and lazy.
 
You must log in or register to reply here.
Back
Top