When you put it like that, it really makes me wonder how VIA stays in business! I guess eventually R&D costs on x86 chips will drop, but that'll also reduce their performance advantage over ARM.
There is more to cost than the silicon. ARM chips carry on average a $0.2 license cost. Perhaps its a bit more if you use ARMs hardmacro's and mali gpu, but it wont be much more. That pretty much gives you a soc design you can take to the fab. Thats why you see small startups like Nufront coming out of nowhere with what looks to be a very competitive chip:
http://www.itproportal.com/2011/01/09/nufront-preparing-arm-based-server-chip-august-2011/
They just used ARMs off the shelve, high performance implementation. Their R&D cost is close to zero, at least when compared to what intel and amd need to spend on designing a new chip.
Lets assume AMD spent something like $500M on bobcat. IF they manage to sell 50 millions of those, each will carry a $10 R&D cost. A figure thats almost as high as the sales price of highend ARM chips. Pretty tough for them to compete. And it gets worse if AMD actually starts competing on price with ARM chips, since the availability of ~$20 bobcats would surely canalize its higher margin (and higher R&D cost) chips.
When you put it like that, it really makes me wonder how VIA stays in business! I guess eventually R&D costs on x86 chips will drop, but that'll also reduce their performance advantage over ARM.
http://arstechnica.com/business/new...ook-for-nvidia-gpu-on-intel-processor-die.ars
Looks like Intel will getting Nvidia GPU IP for Sandy bridge and Nvidia will get 1.5 Billion USD plus access to Intel's CPU IP for project denver.
How will this affect AMD's decisions to design ARM APUs?
AMD is one of the most IP rich companies anywhere in IT, but the company in itself is not set up to compete with ARM design houses that have been working on such platforms for 5-10 years.
How about Notebook ARM? Wouldn't AMD be capable of building something in that class? (Higher Power platform with less integration)
Well that's the thing Intel's next move is 22nm. If Intel feels threatened they can pump out a 22nm Atom with x64 capabilities & even hyper threading in a duel & quad core design if they want to. Throw in Lucid Virtu software and Intel can have their own version of Optimus with an AMD discrete mobil GPU. That also isn't counting their advantage with SSDs. Intel can easily sign some sweet deals to companies for netbooks & tablets.
AMD had a MIPS based SOC called Xilleon and ARM based SOC called Imageon as a result of the ATI acquisition. Both were sold off because both were money losers..
I dont want to underestimate the difficulty of building a proper SoC, but if a tiny company like Nufront can come up with 1.6-2 GHz dual cortex a9 out of nowhere, by just implementing licensed ARMs hardmacro's, I have a hard time imagining why AMD could not, and pair it with their GPU. From there to getting in to acutal mobile phones and tablets is something else, as nvidia found out (and Im sure nufront will), but AMD already had Motorola and Nokia as customers for Imageon.The ARM world is fiercely competitive world filled with powerhouses like TI and Qualcomm and many other strong contenders most of whom are architectural licensees. Hell look at nVidia they spent 3 years and hundreds of millions to finally break into the high end ARM market
Tom's hardware said:For netbooks, AMD's configuration combines a SB750 southbridge with a Ontario. Since the wireless device is directly hooked up to the APU's single PCIe link, there are improvements in power management, as the southbridge can go into an idle state without sacrificing connectivity.
For notebooks, the plan is to pair a SB750 southbridge with an Ontario or Zacate, with the additional use of a four-lane PCIe connection to a discrete GPU. However, this sacrifices some power management savings by hooking the wireless device to the southbridge. AMD's logic is that in a netbook, users would be less likely to need a discrete graphic solution. By moving wireless connectivity up to the APU, the southbridge only needs to deal with I/O devices like the keyboard, touchpad, USB devices, and flash media. Given that mobile users are less likely to use USB devices and flash media while on the road, the SB750 only has to transfer small bits of data from the keyboard and touchpad, which translates into higher power savings.
Everyone can benefit from power savings, and in a world where we leave our wireless connection active, the ideal situation would be to always have the wireless device hooked directly into the APU. However, this is not possible if you are using a discrete graphic solution in a x4 configuration. Remember, there are five PCIe controllers off the APU, and one is reserved for the UMI link. The other four are intended for peripherals. For discrete graphics, you can either use a single x1, x2, or x4. So in practice, it is possible to connect a discrete graphics chip with x2 and still simultaneously have two x1 connections available. Meanwhile, all "Hudson-M1" southbridges come with a UMI connection to the APU (Ontario/Zacate) that is based on a single x4 PCIe connection; probably with some aspect of proprietary signal handling.
I dont know if they actually lost money (or more money than AMD tends to lose anyway, even with superior products), but these were strategic assets on which AMD could and should have built. Imageon now lives on as Adreno, the GPU of a gazillion mobile phones and tablets powered by qualcomm. The CPU part of Xileon was useless (other than for its intended markets, where AFAIK, it actually did pretty well), but pairing it with a better CPU core, even an off the shelve ARM one, would have given it an excellent shot at IPTV.
I dont want to underestimate the difficulty of building a proper SoC, but if a tiny company like Nufront can come up with 1.6-2 GHz dual cortex a9 out of nowhere, by just implementing licensed ARMs hardmacro's, I have a hard time imagining why AMD could not, and pair it with their GPU. From there to getting in to acutal mobile phones and tablets is something else, as nvidia found out (and Im sure nufront will), but AMD already had Motorola and Nokia as customers for Imageon.
They seriously blew this one, and any chance at this market. They'd have to start from scratch, assuming they would even be allowed, as I suspect qualcomm demanded some guarantees. I wouldnt be surprised one bit, if this is one of the key reasons they got rid of Dirk Meyers.
Here is what Tom's hardware had to say about the power management scheme of Brazos Hudson PCH.
Of course I don't understand the tech details to any appreciable degree. I am just a layman. But it sounds like a quality part and I am wondering if AMD could use in combination with a 2.0-2.5 Ghz ARM A15 or AMD designed ARM custom core?
It's certainly not a cake walk; Nufront has been working on this since late 2004.
At this point, time may have run out for any semiconductor company to start from scratch, if they have yet to enter the game.
Actually their website claims: "Nufront since 2006 started its R&D focus on new generation of highly integrated, full function and lower power consumption computer system chip."
Moreover, they had zero experience in this field. From what I can tell, they only did software and some broadcast tech. They seem to be doing a lot of apparently totally unrelated stuff with just 200 people total. I dont know how many of those worked on their SoC or what their background is, but this seems to be the semico equivalent of a handyman building cars in his garage.
I disagree. This market is only just beginning, its not nearly mature.
Moreover, the current players do not overly impress me. Texas Instruments is a big player, but frankly, hasnt ever done very well outside embedded markets and seems to be dropping the ball pretty hard (omap4 was announced in early 2009). Samsung has the right pedigree and might be the one to beat, but it has reportedly placed a 2-$300M order for Tegra's which cant spell much good about Orion. Both Marvell and Qualcom seem to struggle to keep up on both the CPU and GPU front. Apple doesnt exactly have a long history as semico and intel chose to bring a truck to a kart race.
If I had been in charge of AMD, this is the market I would have concentrated my efforts on. Id even go for it now, better late than never, rather than spending billions on highend x86 which long term future seems destined to shrink significantly, and where you are not gonna beat Intel anyhow.
And I wouldn't count them in as a major player yet.
If this samsung/tegra rumor turns out to be correct, I agree honeycomb for tablets is part of the reason. Although I wonder how long it would take for samsung to port android to orion, I doubt they expect to sell ~20 million tablets in the meanwhile. Moreover, the rumor also points to samsung using tegra for mobile phones, where honeycomb really isnt a factor. We will know more next month, but my guess is that Samsungs choice of Mali is backfiring.The bit about Orion and tegra doesn't really make sense, as a company size of Samsung certainly can source certain designs from external, if it fits their particular product requirement; this actually makes sense as Tegra 2 is the devel target for Honeycomb, so no surprise there. It's like saying that IBM should not be using intel chips since they have Power, or Sun would never use x86.
Yeah. Talking about a rock and a hard place. Whoever takes over at AMD is in for a serious challenge, thats for sure.Depending on the forecast, AMD may or may not still have an opportunity here; and even if there is some chance, the odds of AMD surviving long term as an independent company either against the ARM onslaught, or be an also ran in the space, is not good.
I Laptops, small desktops and over time, server appliances. Not that many non x86 players there, excellent TAM, far easier to get in to as mobile phones or cars. Power consumption is not so much of an issue and all you need to get started is a few Chinese white box vendors and ubuntu. Im surprised no one else jumped on this earlier.
Sounds like Nvidia would see good opportunity then. Maybe AMD will join the competition as a way of providing an CUDA/Physx alternative for these Higher powered devices you are referring to?
P.S. I am also looking forward to see how Google develops Android? Maybe a 4.x versions will become SFF desktop/laptop OS? (I am having a hard time buying into ChromeOS appealing to everyone at this time)
NV has really played their cards right in the last 3-4 years. Now provided that they can compete with some of the more established entities like Samsung, TI, Marvell, Qualcomm, etc, they will be in very good shape to take advantage of the trends in IT. I remembering some ridiculing their Tegra strategy a couple of years back, and now their direction has been validated. Although their discrete GPU products are really inferior to the competition in terms of the efficiency of the microarchitecture and profitability per wafer, it does not really matter in the big picture any more. They basically changed how the graphics game is being played.
AMD, on the other hand, I don't see a likely way forward as an independent company. They are probably going to be shut out of the markets where we see most of the hardware growth in the industry, and in embedded systems in general. A few minor miracles would be needed in order for them to still be afloat by 2020.
AMD heading in the right direction on ARM but too slow, say Taiwan Notebook makers
Yenting Chen, Taipei; Adam Hwang, DIGITIMES [Wednesday 12 January 2011]
AMD starting to develop display processors to support ARM architectures is correct in terms of marketability and market potential, but its pace is too slow compared with Nvidia's success with Tegra 2, according to Taiwan-based notebook makers.
The global CPU market in 2011 will feature two important things: Wintel will no longer be the only platform and will be increasingly replaced by ARM which tends to dominate in tablet PCs, and AMD will be able to change the ecosystem and score points in the global notebook market if it moves faster; and Nvidia's Tegra 2 has been widely adopted by vendors for mainstream notebooks expected to be launched in June 2011, the sources pointed out.
These vendors are satisfied at Tegra 2's lower costs and that it allows lighter and slimmer notebooks than Wintel as well as shorter boot times.
These vendors are satisfied at Tegra 2's lower costs and that it allows lighter and slimmer notebooks than Wintel as well as shorter boot times.
It's saying something about "Display processors" to support ARM. Does that mean AMD will develop video cards supporting ARM?
But a 14" ARM based notebook running windows, ubuntu or chromeOS, is not a smartphone, its a 14" ARM based notebookSmartphone operating system and processor with a 14 inch screen isn't a Notebook... It's a smartphone with a 14 inch screen.
But a 14" ARM based notebook running windows, ubuntu or chromeOS, is not a smartphone, its a 14" ARM based notebook