The AMD Execution Thread [2007 - 2017]

Status
Not open for further replies.
The point is not "AMD sucks because they can't match Intel" on a CPU basis. The point is the transformation of the company from a bloodied over-matched rival of Intel to an innovator on the semi-custom, APU, embedded front. AMD is going to be a totally different and relevant company imho going forward...instead of the company that Intel crushed. I see opportunity...
 
Well that's good. I'll just buy Intel rigs now. Will convert my htpc over to them too next year

What does pure x86 CPUs have to do with HTPC's? That specific market is perfect for AMD's roadmap is it not?
 
What does pure x86 CPUs have to do with HTPC's? That specific market is perfect for AMD's roadmap is it not?

I rather one family of chips with one set of updates for it.

My desktop and htpc both have and chips so it's one driver package to get. That won't be a possibility going forward with amd
 
So you'll go Intel for your HTPC because you're lazy? Well that's a new one for picking one vendor over another.... then again maybe not so new.
 
Very interested to hear what it is they've been working on with Samsung: https://www.koreatimes.co.kr/www/news/tech/2013/09/133_135262.html

But industry and AMD officials who are directly involved in the issue told The Korea Times that those AMD-manufactured chips will be featured in new game consoles such as Microsoft’s Xbox and upcoming consoles to be introduced by Samsung.

“AMD is in talks with Samsung and Microsoft to ship its custom chips..."

The creation of the Semi-custom business unit is proving to be the right decision.
 
So you'll go Intel for your HTPC because you're lazy? Well that's a new one for picking one vendor over another.... then again maybe not so new.

I am lazy.

To be fair I was elected tech guru for my entire family (extended also) cause you know I do it for a living and I just want to have to go to friends and family's houses when I get off work to go and fix more computers and not get paid.

So I tend to buy the same exact set ups for everyone. Everyone upgrades at the same time , everyone gets the same laptop and so on . When I upgrade the 6 media center's I take care of , we will go from e-350s (dual core brazos at 1.6ghz) to most likely an i3 broadwell chip .

When I upgrade the gaming rigs I take care of we will go from fx 8150s to whatever the hid end intel chip will be since there is no new amd high end. That will happen in 2015 when star citzen hits since we all want to play
 
You're whole family wants to play Star Citizen? I want your family.
 
not my whole family but about 8 of us. You can join us

One of us

One of us

Just come to jersey , hopefully you like Italian food.

you should see the crazy that our civ 5 games become
 
I am lazy.

To be fair I was elected tech guru for my entire family (extended also) cause you know I do it for a living and I just want to have to go to friends and family's houses when I get off work to go and fix more computers and not get paid.

So I tend to buy the same exact set ups for everyone. Everyone upgrades at the same time , everyone gets the same laptop and so on . When I upgrade the 6 media center's I take care of , we will go from e-350s (dual core brazos at 1.6ghz) to most likely an i3 broadwell chip .

When I upgrade the gaming rigs I take care of we will go from fx 8150s to whatever the hid end intel chip will be since there is no new amd high end. That will happen in 2015 when star citzen hits since we all want to play

I fail to see how having Broadwell in the HTPCs is going to make things easier. If anything, it will be worse, because the Star Citizen gaming rigs will have discrete graphics, and probably Radeons (it's a Mantle game) so you'll have to install Catalyst drivers.

On the HTPCs, with Broadwell, you'll have to install Intel graphics drivers; with AMD APUs, it will be the same Catalyst drivers as in the gaming rigs.
 
The few gaming pcs aren't what I worry about. Its the htpcs and regular desktops in the array of computers.

The desktops used for excel /word and media duties like editing photos from vacations and the like will be using the same set up as the htpcs right down to the motherboard.
 
I guess since it appears AMD will be exiting the "big core" x86 market the question should be how far up the food chain can the APU migrate? Consoles, laptops, tablets, low end desktop are a given but how far up can it reach in traditional higher end rigs with a dedicated CPU and a discrete GPU? I see no reason why a media centric PC couldn't use an APU...am I missing something?

It will be interesting to see how the APU (beginning with Kaveri) scales in regards to performance, memory requirements, and what effect HSA and Mantle will have to bridge any performance gap. As any good business person knows "not all revenue is good revenue"...so AMD exiting their traditional market and migrating to greener pastures is a savvy and sound move to me.

Go go semi-custom gadget arms!
 
I guess since it appears AMD will be exiting the "big core" x86 market the question should be how far up the food chain can the APU migrate?
Does it still make a big core component in the APU? If not, it's at a high water mark.
A product goes as high as its glass jaws allow.

The APUs it is selling as semicustom products right now leverage technology that had significant contributions from staff that have left or been hired away since. Whole graphics teams have been lost, as have many of the CPU designers.
AMD is moving into a market with competitors that have many of its former employees responsible for its current semicustom wins.

AMD hasn't divorced itself of a need to advance the technology it has spent most of its time merely leveraging, but the brain drain is obvious.
The semicustom business may be constraining in various ways, since involving other IP holders means AMD spends more of its time on tech it can't patent and may be constrained from adopting elsewhere.
That might be good enough short-term, but AMD can't relinquish its relevance in non-APU products because it needs to add more value than being just a high-end integrator. It's not clear it will have much of a lead there, either (see the companies that employ so many of the employess that made that happen).

Consoles, laptops, tablets, low end desktop are a given but how far up can it reach in traditional higher end rigs with a dedicated CPU and a discrete GPU?
Wouldn't they want to update their system infrastructure and sockets?
I see the same chipset and memory technology as far as the eye can see.
The roadmap that shows the FX line not changing also doesn't indicate the featureset of the APUs changing either, beyond a different processor core that has no presence in products that have CPU performance as a marketing point.
 
I thought about chipsets and sockets earlier...not sure what the implications are there. While your brain drain points are valid to a point, I would counter they have brought back a few top level people from Apple and others. The sad truth is going head to head with Intel with the HUGE inherent disadvantages (no fabs, larger process node, 1/10th R & D budget) all made for a losing hand. AMD has at the least had the courage (perhaps out of necessity) to branch out dramatically from their roots and try a different angle for serving the consumer and enterprise electronics markets.

They are due a level of credit for buying ATI (although they overpaid) to achieve this APU vision they had, getting an ARM licence, buying Sea Micro, getting the console wins, developing a potentially widely accepted api (Mantle), helping to drive the open HSA movement, and returning to profitability. That was a looooong road back from oblivion and they are certainly not out of the woods by a long shot...thus the $3.65 share price.

I find it ironic I have gone from their fiercest critic to an AMD apologist!! To make it worse, I have put my money where my mouth is and bought waaay more shares than I should have. "A fool and his money..."
 
Does it still make a big core component in the APU? If not, it's at a high water mark.
A product goes as high as its glass jaws allow.

The APUs it is selling as semicustom products right now leverage technology that had significant contributions from staff that have left or been hired away since. Whole graphics teams have been lost, as have many of the CPU designers.
AMD is moving into a market with competitors that have many of its former employees responsible for its current semicustom wins.

AMD hasn't divorced itself of a need to advance the technology it has spent most of its time merely leveraging, but the brain drain is obvious.

People have left, people have joined, it's not easy to tell what the net effect might be from the outside.

The semicustom business may be constraining in various ways, since involving other IP holders means AMD spends more of its time on tech it can't patent and may be constrained from adopting elsewhere.
That might be good enough short-term, but AMD can't relinquish its relevance in non-APU products because it needs to add more value than being just a high-end integrator. It's not clear it will have much of a lead there, either (see the companies that employ so many of the employess that made that happen).

Other IP holders? There's ARM I guess, and Sony/MS might have done a little bit of work on the console APUs, but probably not much. I think eventually AMD plans to have its own ARM core, so they would only rely on the ISA (and more minor stuff from ARM et al.).

Wouldn't they want to update their system infrastructure and sockets?
I see the same chipset and memory technology as far as the eye can see.
The roadmap that shows the FX line not changing also doesn't indicate the featureset of the APUs changing either, beyond a different processor core that has no presence in products that have CPU performance as a marketing point.

The new CPU core is a given. A new GPU is, I think, fairly likely and could very well be behind the very vague "Next gen AMD Radeon graphics" bulletpoint. Other chip-level changes could be there, like a new memory hierarchy, e.g. with a small, fast, inclusive L2 for each module and a shared L3, accessible to the GPU as well. The latter is likely to be about twice as big as that of Kaveri (assuming a 20nm process). There could be package-level improvements, like stacked memory. In fact I think this kind of needs to happen. As for CPU performance not being a (major) marketing point, that was/is true of Llano/Trinity/Richland, but that's mostly because CPU performance just wasn't good. It should be better with Kaveri—though probably not enough for AMD to emphasize it—and if Excavator brings further substantial improvements, this could very well change. Currently, AMD has a 2.9GHz (Turbo) / 2.1GHz (base) quad-core (Richland) with a 25W TDP. Assuming Excavator brings a cumulative IPC improvement of 20~30% over Richland (including the intermediate Steamroller step) perhaps with a ~10% clock speed bump in this power enveloppe, this ought to be more than adequate for the great majority of consumers.

I mostly share overclocked_enthusiasm's view that AMD has a lot of potential, but with a difficult road ahead, and a few objectives they can't afford to miss. I think the quality of the Steamroller architecture, which should become clear very soon, will be a good indicator of the company's future prospects.
 
The CFO last week at the CS conference talked about going from 4-5% to "25%" in the server market over the next 5 years.

http://www.youtube.com/watch?v=BOYdKht1YwE

This Amazon Web Services engineer has nice things to say about AMD/Sea Micro at the end of an interesting presentation. While it is a leap to say "AMD/Sea Micro is getting Amazon cloud server business" because somebody with an Amazon shirt on was talking favorably about AMD technologies...it does highlight the relevancy of AMD's ARM/Sea Micro CUSTOMIZABLE strategy as evidenced by the Verizon deal.

http://www.datacenterknowledge.com/...-seamicro-backed-infrastructure-as-a-service/
"Cloud service launches are a dime a dozen these days. But upon its launch last week, the Verizon Cloud was dubbed “technically innovative” by Gartner analyst Lydia Leong, an influential voice among cloud analysts.
“The Verizon Cloud architecture is actually very interesting, and, as far as I know, unique amongst cloud IaaS providers,” Leong wrote on her CloudPundit blog. “It is almost purely a software-defined data center.

“It’s an enormously ambitious undertaking,” she added. “It is, assuming it all works as promised, a technical triumph — it’s the kind of engineering you expect out of an organization like AWS or Google, or a software company like Microsoft or VMware, not a staid, slow-moving carrier (the mere fact that Verizon managed to launch this is a minor miracle unto itself).”"

http://gigaom.com/2013/11/15/how-am...ly-doing-everything-to-keep-cloud-costs-down/
"Like Google and Facebook, Amazon is designing its own servers, and they’re all specialized for the particular service they’re running. Back in the day, Hamilton used to lobby for just having one or two SKUs from a server vendor in order to minimize complexity, but times have changed. Once you master the process, going straight to server manufacturers with custom designs can lop 30 percent off the price right away, not to mention the improved performance and faster turnaround time.

Today, “You’d be stealing from your customers not to optimize your hardware,” he said."

AMD is quickly becoming a custom shop with APUs and ARM products...
 
For the moment AMD isn't really trying to make its way into this business; unless you count Windows 8 tablets, but here the competition is just Intel.
 
For the moment AMD isn't really trying to make its way into this business; unless you count Windows 8 tablets, but here the competition is just Intel.
so where they are going ?
So called next gen consoles, done.
and after ?
 
Status
Not open for further replies.
Back
Top