AMD Not Competing with Intel Anymore, Goes Mobile

[AEKaBeer]

Originally Posted by Gillyske

Lol the company itself is thinking of abandoning desktop cpu's and fanbois are still hopeful. Its amazing.

Where in the whole 1 page article did you see them saying they're abandoning the cpu market? They're not competing with intel in the old ways doesn't mean they've closed up the cpu decision.

[Brett]

If anything changes it'll be more focus on competing with Atom and and Tegra. Fast set top boxes combined with tablets do most things a household PC is needed for and you don't need an i7 in a thin client.


edit:

Originally Posted by Gillyske

Lol the company itself is thinking of abandoning desktop cpu's

Link to a press release? Their news page has nothing.
They'll keep building desktop and server gear, put some more RnD into portables.

PS. not a fanboy.

[Thrawn]

Originally Posted by Bertross

Intel will never make monday from SB-E, $670 for entry level CPU + a $350+ motherboard??? who can afford that?

Intel makes a lot of money from SB-E (aka Xeons). This class chip may not sell in huge quantities but in terms of revenue it makes quite a sizable chunk (I think something like 1/3 of total processor revenues).

Basically in the market there's always a set amount of people willing to pay a particular price. Intel's goal is basically to make you pay as much money as possible from as 'small' a chip as possible.

Back when AMD was really competing, Intel had to use its best chips to fulfill mainstream desktop. Nowadays you can practically fit 4 units of Core i3 processors for the same die-size as a Bulldozer FX-4100. A Bulldozer nearly has the same die size as an SB-E.. so think about how much money AMD aren't making.

[Toliandar]

It will be a bad day when AMD retreats from the discrete cpu market. I can see why they may do it - comparing the successes of FM1, Bobcat etc to AM3 it's not hard to see where AMD's strongpoint is. Their next step should really be building on FM1 particularly on the performance per clock area which has consistently been the thorn in AMD's side for the last few years. Their graphics department is doing great and their tech will bleed down easy into their cpus I'm guessing; but to match that they need to improve their performance per clock.

They also really need to squeeze the manufacturers to get their yields up because IMO what hurts them the most right now as far as the laptop/tablet market goes is they aren't relatively cheap when you compare them to Intel offerings. Get the yields up and performance per clock up and they could really butcher Intel in the netbook/notebook/tablet market and as far as the tablet market goes - a new chip on the block can only help prices hehe.

[Thrawn]

Originally Posted by Toliandar

Their next step should really be building on FM1 particularly on the performance per clock area which has consistently been the thorn in AMD's side for the last few years. Their graphics department is doing great and their tech will bleed down easy into their cpus I'm guessing

Not much chance of AMD improving performance per clock anytime soon. That was meant to be done in Bulldozer.. but they failed so badly there. They need another radical new architecture to catch up with Intel and that would take years to materialize.

Funnily in the GPU side the boot is in the other foot. Intel is devoting a substantial amount of die-space and power budget to their GPU, yet it still sucks.

[Mau1wurf1977]

Originally Posted by Toliandar

I can see why they may do it - comparing the successes of FM1, Bobcat etc to AM3 it's not hard to see where AMD's strongpoint is.

To be honest I don't see that "strong point". AMD is so far behind in terms of performance per watt. Not just with BD or Phenom II, but also with Brazos.

Their main strategy has been to focus everything on graphics and that's about it.

Problem is that a low powered Ivy Bridge will just rip Brazos apart and while AMD is ahead in terms of graphics, Intel graphics have almost caught up and things such as Flash and Video acceleration work just as well.

I have a notebook with the E-300 CPU and it gets outperformed by a two generation out-of-date Pentium ULV. Now AMD has issues with Global Foundries and it seems the E-450 will be top-of-the-line ULV CPU for another year. Soon the Atom will be as fast as these Brazos chips...

[NNNG]

Originally Posted by Mau1wurf1977

To be honest I don't see that "strong point". AMD is so far behind in terms of performance per watt. Not just with BD or Phenom II, but also with Brazos.

Their main strategy has been to focus everything on graphics and that's about it.

Problem is that a low powered Ivy Bridge will just rip Brazos apart and while AMD is ahead in terms of graphics, Intel graphics have almost caught up and things such as Flash and Video acceleration work just as well.

I have a notebook with the E-300 CPU and it gets outperformed by a two generation out-of-date Pentium ULV. Now AMD has issues with Global Foundries and it seems the E-450 will be top-of-the-line ULV CPU for another year. Soon the Atom will be as fast as these Brazos chips...

Brazos competes with dual core atoms and it does so extremely well. It does not compete with ULV processors like Ivy Bridge based ultrabooks which will cost a lot more.

[IMtech]

Please tell me this is not true!!!

Yours kindly,

An avid AMD fanboi

[ppantel]

You guy's are scaring me lol! sounds like all companies are getting out of the desktop market. I was born and raised with a desktop in my hands when playing sierra games (like leisure suit larry) was cool!

now everyone's focusing on the laptop and tablet market!

what are us 3d artists and gamers going to do? buy servers with xeons then? its all apple's fault with their stupid ipad.

[Mau1wurf1977]

Originally Posted by NNNG

Brazos competes with dual core atoms and it does so extremely well. It does not compete with ULV processors like an i3.

That's where I disagree as there are several models of Brazos. The E-300 and E-450 are not found in netbooks, but 11.6" machines and also 14" and 15" ones were they DO compete against Intel ULV and even cheap Celeron machines.

Now in a 15" machine the SB Celeron just demolishes Brazos. So do all the ULV chips.

So this leaves us with the Atom as the only chip that Brazos can compete with. And in terms of price it also doesn't as you have to pay a premium to get a Brazos netbook over an Atom one.

Now Atom also hasn't stood still. They are now all dual core and the latest 1.66GHz model is faster! than the 1GHz Brazos. Back when Brazos came out it was mostly going up against single Core atom, but that train has left the station.

[Digit]

Originally Posted by Bertross

yup, just bad marketing, trying to sell what isn't true to the common folk but more technically speaking it is somewhat. I think yeah sure 8 cores sounds awesome lets use that....

imo the B3 P67 was more a disaster. Having to return a physical product is embarrassing.

Take the pildriver chip next term and see how it pairs to a 2600k then i bet we get a swing of change. Then lets talk a die shrink and a new socket and we talk business. Good things come to those who wait.

since PD isn't meant to be coming out until 2013, i'm fairly certain you should be comparing it to an IvyBridge CPU (because you know, Intel isn't just going to sit down and twiddle their thumbs), which would more than likely DEVASTATE the PD chip. AMD has stated in a roadmap that they're only hoping for around a 10% improvement over Bulldozer while I've read that Intel is going for a 15% improvement, which only widens the gap even further.

[stmok]

Originally Posted by OpenSystem

ITS ALL OVER!!! AMD Not Competing with Intel Anymore, Goes Mobile.

You presume way too much from what little was actually said by an AMD rep.

AMD have realised the long term threat to their existence isn't Intel any more. Its ARM-based providers like Nvidia, Qualcomm, Texas Instruments, etc.

Think about it, the majority market share of Tablets and Smartphones are dominated by ARM architecture. Mainly because their chips are much more affordable than x86 equivalents while providing sufficient performance for those markets. (There's no competition when Intel can only sell a few hundred thousand Atom chips versus ARM providers selling a couple of million in the same Quarter.)

ARM Holdings is pushing upwards in performance for the next few years. And if you have been keeping track, they have server and desktop intentions as their future. (The space where x86 market currently occupies.)

AMD is sticking with x86 architecture. The concern is if ARM squeezes out x86 and becomes the dominant consumer computing architecture; both AMD and Intel are f*cked.

Intel and AMD have the performance. But they don't have the cost (per chip) and power footprint in the same league as ARM. Its easier for ARM to climb upwards in performance than it is for Intel/AMD to climb down. (Since ARM do not have product lines that are higher than what they already have. They have a clear path.)

This is why Intel is so focused on die shrinks and manufacturing technology for future versions of the Atom and R&D on running processors with the lowest voltage as possible; while AMD is eager in getting Bobcat shrunk down first (Brazos 2.0 platform for 2012), then replacing it with the third generation in 2013 (Kabini for netbooks/nettops and Samara for tablets).

The real fight isn't between Intel and AMD. Its x86 vs ARM...Its what that AMD rep was implying.

[Toliandar]

But why is ARM doing so well on a power/performance measure? Is it because of a similar situation with consoles where because there is little variation in hardware that lower level access to hardware is more available/generally easier to optimise for the hardware?

[mAJORD]

it's the simplisity of the RISC instruction set..

x86 is old, and inefficient. Modern CPU's decode x86 instructions into RISC instructions internally to improve performance. To put it simply, This process "wastes" power

[Kurosaki]

Originally Posted by mAJORD

it's the simplisity of the RISC instruction set..

x86 is old, and inefficient. Modern CPU's decode x86 instructions into RISC instructions internally to improve performance. To put it simply, This process "wastes" power

Intel offered everyone a chance to leave this x86 space. The market said it prefer it this way