What happened to AMD high end perfomance?

[mdzapeer]

At a time they were the best during the Athlon? So why don't they make chart topping stuff???

 

[jhu]

The low-hanging fruit has been picked. They would if they could.

 

[master_shake_]

right around the release of core2.

amd has being playing catch up and then give up since.

so... 2006? or 05.

 

[witeken]

You can't do much in this space without some serious R&D money. The Bulldozer philosophy has dramatically failed, too.

 

[Headfoot]

flamebait thread baits flame

 

[ClockHound]

They just went dormant for a decade or so to allow Intel to catch up. Seems to have worked.

Now they're waiting for their prince of investment to wake them up with the kiss of unlimited R&D. They're going to have to pay more to hookup at Tinder tho.

 

[Fjodor2001]

You can't do much in this space without some serious R&D money. The Bulldozer philosophy has dramatically failed, too.

AMD didn't have more R&D money than Intel in AMD K8 uArch era, yet they produced a better CPU.

R&D money is no guarantee for success. Just look at how poorly Intel's mobile phone CPU division has been performing despite all the billions of dollars in R&D they have wasted on it.

 

[Fjodor2001]

flamebait thread baits flame

+1

 

[ShintaiDK]

While you may never get what you pay for in R&D, you certainly dont get what you didnt pay for.

 

[TurtleCrusher]

AMD didn't have more R&D money than Intel in AMD K8 uArch era, yet they produced a better CPU.

R&D money is no guarantee for success. Just look at how poorly Intel's mobile phone CPU division has been performing despite all the billions of dollars in R&D they have wasted on it.

That and giving away bay Bay-Trail atoms to reduce ARM architecture marketshare. While I dig my Bay-Trail, Intel is bleeding cash in terms of mobile phones/tablets. Think about it, Intel has the best manufacturing process yet still cannot compete effectively with extremely low-margin, low R&D ARM licensees.

 

[Fjodor2001]

While you may never get what you pay for in R&D, you certainly dont get what you didnt pay for.

Too generalized of a statement to be used to determine any specific company's future success or lack thereof.

If you would have applied that statement to Apple in 2000 and some years prior to that, would you have guess what that company has turned into today? Their R&D budget was tiny back then and the company was close to bankruptcy.

Similarly, I guess you would have thought Nokias success in 1995-2005 was going to be an eternal sunshine story. Turned out it wasn't.

Lesson learned: Reality holds interesting twists in store for us that cannot be predicted simply by looking at the size of a company's R&D budget.

You should know this by now...

 

[myocardia]

They would still, if the lying, stealing man whom their founder, Jerry Sanders, picked to be his replacement had put their profits into R&D, which he did not. Added to that, he paid ~triple the value for ATI, some of which I can assure you went into his pockets, and they have struggled ever since. There is a very good reason that his nickname is 'Ruinz'...

 

[Arachnotronic]

That and giving away bay Bay-Trail atoms to reduce ARM architecture marketshare. While I dig my Bay-Trail, Intel is bleeding cash in terms of mobile phones/tablets. Think about it, Intel has the best manufacturing process yet still cannot compete effectively with extremely low-margin, low R&D ARM licensees.

Low R&D?

Apple, Samsung, and Qualcomm spend oodles on R&D.

 

[TurtleCrusher]

Low R&D?

Apple, Samsung, and Qualcomm spend oodles on R&D.

MediaTEK, a fabless semiconductor company is somehow doing much better in mobile than Intel, which has mostly gone all-in on mobile. Apple and Samsung spend much of their R&D on other ventures than mobile. Qualcomm commands 2/3 of the market on their own, so it'd be expected they spend a decent amount.

 

[ShintaiDK]

Low R&D?

Apple, Samsung, and Qualcomm spend oodles on R&D.

Samsung doesnt spend that much on R&D in terms of chip design. But it also shows.

 

[aigomorla]

I predict this thread will not end nice...

 

[ShintaiDK]

MediaTEK, a fabless semiconductor company is somehow doing much better in mobile than Intel, which has mostly gone all-in on mobile. Apple and Samsung spend much of their R&D on other ventures than mobile. Qualcomm commands 2/3 of the market on their own, so it'd be expected they spend a decent amount.

It still cost MediaTEK 1½B$ a year in R&D just to make SoCs. Close to what nVidia does and much more than AMD.

 

[TurtleCrusher]

It still cost MediaTEK 1½B$ a year in R&D just to make SoCs. Close to what nVidia does and much more than AMD.

Yet MediaTEK still has nearly a 3x mobile marketshare of Intel. Without gift wrapping with cash to use their chips. Profitable? Gaining marketshare? Putting up a fight against the big players?

Absolutely. This is without the luxury of the best fab processes available.

 

[Maximilian]

At a time they were the best during the Athlon? So why don't they make chart topping stuff???

Well... it went like this

Athlon 64 - Amazing
|
v
Athlon X2 - Amazing, until core2 came out then average
|
v
Phenom 1 - Sucked, also had a TLB bug that knocked ~10% performance off it
|
v
Phenom 2 - On par with core 2, intels i7 line was released shortly after and bested it
|
v
Bulldozer - Sucked
|
v
Piledriver - Sucked less but still sucked

So as you can see, out of the 6 things listed 3 of them sucked. Leaving AMD where it is today.

 

[frozentundra123456]

Still have an Athlon XP 2600 that I keep for nostalgia. First computer that I had that just "felt" fast. I dont really get attached to electronics, but I have more fond memories associated with that computer than any other one I have ever had, because with a cheap dgpu it really brought gaming to a whole new level.

After that though, I went Core and never looked back.

Edit: Kind of ironic that in the athlon days, AMD had the best gaming performance while Intel was more competitive in productivity tasks. Now just the opposite. Intel definitely has the upper hand in gaming, while AMD is more competitive in productivity apps, at least certain highly threaded ones. The only CPU intensive thing I do is gaming though, so I will just stick with my Sandy Bridge i5 for a few more years.

 

[myocardia]

Well... it went like this

Seriously, no K7 Athlon? It was amazeballs, and it beat Intel at Intel's own game. It not only was able to clock higher, it had better floating point than the Pentium III, at the same speed(s).

 

[ShintaiDK]

Yet MediaTEK still has nearly a 3x mobile marketshare of Intel. Without gift wrapping with cash to use their chips. Profitable? Gaining marketshare? Putting up a fight against the big players?

Absolutely. This is without the luxury of the best fab processes available.

In chip design its only the first chip that cost the big bux.

Thats why its all about volume and why there essentially can only be a monopoly in the end.

 

[Arachnotronic]

Yet MediaTEK still has nearly a 3x mobile marketshare of Intel. Without gift wrapping with cash to use their chips. Profitable? Gaining marketshare? Putting up a fight against the big players?

Absolutely. This is without the luxury of the best fab processes available.

Scholzdpx

MediaTek is undeniably far more successful than Intel in the mobile market, and the company sports robust margins and profits from mobile chip sales.

However, you should understand that success in a market isn't about how much money you spent on R&D last year; technical success is often a combination of a robust, effectively utilized R&D budget over many years.

There's more to the story than just the R&D budget or even the efficiency of that budget; customer relationships and corporate culture are also key. MediaTek has been in phone chips for a while, and it has built many customer relationships. This means that the company not only can sell parts/platforms today, but it can gather the input of its customers to help define competitive next generation products.

Additionally, "success breeds success," if you will. If you build a competitive product today, then it's much more straightforward to deliver a competitive product tomorrow because you've got a good baseline to build off of.

Finally, the last note I have on MediaTek for now, is that MediaTek leverages TSMC's R&D investment in process technology + IP, and it also licenses IP from companies such as ARM and Imagination Technologies. This significantly lowers the cost of entry into the mobile market because a lot of that heavy lifting is already done for it by other companies.

Back to Intel: Intel's core competency is CPUs/SoCs for PCs and servers. In the mobile market, CPU/GPU capability is important, but good IP for both of those can be licensed. The mobile market (in particular phones) is very dependent on communications/connectivity technology as well as imaging, and time-to-market is a very important consideration.

The mobile companies like Qualcomm and MediaTek are very good at spinning many derivatives of chips quickly. Intel still needs to develop this capability, and it doesn't come overnight no matter how much money they pour into it into a given year. It's a marathon, not a sprint.

Finally, last note on Intel, the time it takes to go from the proverbial drawing board to shipping silicon is measured in years and if you study Intel's R&D budget trend, you will notice that there was a marked step up in the 2011/2012 timeframe.

The products you are seeing from Intel now (Merrifield, Bay Trail, Cherry Trail, etc.) are all products of a much lower R&D budget from a company that is still trying to figure out its market. The R&D budget is now much higher, but the company still needs to learn its way around the mobile market.

 

[Abwx]

Intel's core competency is CPUs/SoCs for PCs and servers.

Their core competency is to be a foundry, untill you understand this you ll keep publishing articles that are completely out of phase while giving erroneous advices...

 

[NTMBK]

Nothing happened to AMD's top performance. That's kind of the problem. They haven't moved forwards significantly since Thuban...