He once said he would invest in both INTC and ARMH if he had the trading capital available. You are a colossal jerk.
Indeed, and I can STILL see the case for owning both stocks.
Even with the new clash around servers, networking, iot etc.
Was it when you still thought RT was going to be a success?
churlish, childish and content-free?
RT was doomed to fail as soon as MS announce the locked down Win32 environment. I wasn't one of the RT fan boys:)
As a side, you're going to love the rumours that MS is working on a JIT x86 to ARM translator. Wonder what they could use one of those for?
"Good for you, so why don't you put your money where you mouth is?:)"
[So, in addition to being churlish, childish and content-free, you are now giving advice that you never, ever take?]
Question, are you blind to your own postings? This whole thread is "churlish, childish and content-free"...But it's OK, when you do it, and not others eh?
Cause I'm seeing a few tiny signs. LOL
Good for you, so why don't you put your money where you mouth is?:)
[Awwwwwwwwwww, Monk was really hopeful about this. LOL]
No not at all. If the ARMy are to make the inroads into the server market, it'll come from a variety of SoC designs targeting different sections of the market. I've no idea on how good or bad this QCOM server SoC is.
purpose did it serve? Right or wrong, it makes no difference, plus it was mostly innuendo. I never see him bashing ARMH or any of Intel's competitors. I wonder who pays him via his transfer agent, the Motley Fool.
But beyond that, a good chunk of what he wrote is correct (when talking about density of past nodes). Many Intel bulls used to argue about it on this forum, until Intel started to show the slides that Ashraf used in his latest click bait piece. Why bash ARMH? Who would click?
You must have had way too many strawberry muffins and fried bread with your greasy english breakfast this AM. Maybe even a snort of sherry?
This is what you wrote:
(also a hoot is this QCOM-Google sideshow which gives the Ash's yet something more to muck rake and self-contradict (aka bald faced lying) themselves about)
Fancy explaining that?
[And yet another content-free post in the same childish, churlish style. ]
I can only assume you don't know what he's talking about either.
[Monk, this is the kind of content-free posting that has become your new standard. ]
Perhaps you could explain what the poser was going on about?
Stay the course and INTC will be back in the high 30's if noit low 40's
in no time (also a hoot is this QCOM-Google sideshow which gives the Ash's
yet something more to muck rake and self-contradict (aka bald faced lying)
What are you going on about?
Well, they were doing great in smartphone business, but Apple and Samsung pretty much dropped them.
can put a dent in Intel's server business. AMD has been at it for years and look where they ended up eventually. At least with AMD, they have or had a much better chance of competing against Intel.
Qualcomm investors are dreaming if they think Qualcomm can actually produce server chips that can serve anywhere near what Intel's can. If anyone talks non Intel's servers, it's only for the reason of using it as a bargaining chip.
FYI, Apple have never used QCOM in it's application processors, only for it's modems.
I don't think you've got a handle on the capability's of what a smart phone can do. Just use your smart phone and turn off connectivity and play a game, all the horse power is still there. There is no dependencies on 'servers' as you put it. The servers allow communication, ie, your facebook app would be useless without servers.
Your example of AMD is actually a good question. Why could QCOM succeed when AMD failed?
AMD is (or rather was) competing with Intel on a socket by socket and form factor basis, much like in PC's you 'just switch' between AMD and Intel. But this isn't what QCOM and the rest of the ARMy are doing. They are going after specific workload's and sections of the server market that aren't dominated by high CPU load. The server SoC's will look very different.
As a simple example, take google's youtube. When users are watching a video, the server is simply moving data from storage or memory to the network, and as such you aren't bound by the CPU, but the speed of your network and io. To increase the 'performance' of this workload, you don't increase the CPU, you add more io IP blocks directly to the server SoC (which reduces the cost of the server and reduces it's power consumption). Basically, the ARMy's server SoC's are light on CPU (in terms of single threaded performance), but heavy on network, storage and ram.
do you understand now
I never stated that Intel is fabbing for Altera @ 22nm -
what I said was that Intel is fabbing for several small FPGA customers at 22 nm FinFet - these are ultrahigh margin products with small quantities - it includes Tabula, Anchronix, Microsemi (bought Actel)
IOW Intel is very capable to put FPGA's into silicon - they do have the expertise and IP -
do you understand? Intel has several FPGA customers and your comments about lacking IP are rubbish -
wasn't that A.E. point - no I did not even read it but that's probably what he implied beside his idiotic comments about loosing process lead
TSMC does not break down its revenue into 20 nm and FinFet - so nobody really knows
Anchronix signed up with Intel in 2000 - so it takes time and Intel has a very different understanding than TSMC about the term "ramp"
I wasn't replying to you, but to Wallis.
Wow. I suggest you read what A.E is talking about, then we can talk more. We're talking about Altera and Xilinx, not some startup. In regards to 3rd party IP, A.E's point was that Intel has had problems integrating 3rd party IP into it's process before (Infineon).
You should know that it's not all about process...
[The whole presentation was based on a comparison assuming that Xilinx 14/16nm is equivalent to Intel's 14nm process. So, quit trying to change the game. And Intel's 22 nm products are what you should be comparing to. Are you saying they aren't in the market yet? ]
You really don't understand do you? Intel isn't producing any Altera products at 22nm! For that generation, Altera is using TSMC.
As I have said several times, Intel's 14nm is superior to TSMC's 16nm, but that superiority means NOTHING if you don't have a product in the market (in this case a FPGA). Also, while Intel's node is superior, ultimately it doesn't mean that products using that node will be superior to the competition as the design of the SoC is extremely important.
Do you understand now?
"And the former will commit to using the latter's chips IF they meet performance goals". Google's experiments with ARM chips likely have little to do with performance and everything to do with strengthening their power in price negotiations with Intel.
Question, you are the Exec at Intel in charge of the Google account. Do you give them a price discount or call their bluff?
In other words, the Xilinx 14/16-nanometer FPGA is an inferior product. End of story.
It doesn't work like that. Intel's transistor technology is better (and denser), but that doesn't necessary lead to a better product. There are a whole set of other factors involved. Besides, according to Ash, Intel's product isn't even in the market yet.
This is hardly more than a small scale experiment with 1,000 cores. The biggest grand challenge associated with this effort is simply building it.
This post was really about teasing Wallis, given his MU and ARMH rants (hence the OT).
In terms of Exascale, there are various ARMy based projects afoot and like all Exascale projects, it's all early days. This project is interesting as they are claiming that the bottleneck (from a power point of view) isn't in the processing units, but the interconnect.
One for you Wallis:
"Europe's ExaNeSt project is looking behind the lounge for ten million ARM processors, to support its exascale supercomputing project.
As well as a bucket of processing, what it's calling its “straw-man prototype” will use liquid cooling, Flash memory in the processor fabric, and will use “innovative, fast interconnects” to avoid congestion.
The prototype won't get anywhere near millions of processors: it will have 1,000 cores using 64-bit ARM Xilinx Zynq Ultrascale+ processors using 16 Gbps interconnect, and 16 GB of low power memory on each board.
Other contributors to the project include EuroServer (inter-processor communications design), Ecoscale (programmable hardware accelerators), Xilinx (FPGAs and communications), Micron (low power memory and storage), and Kaleao (productisation).
Over on the register.
[They never have. You can't really expect me to accept all those half-steps with the ugly back-ends to be true FinFETs, can you? ]
I don't expect you accept anything. A FinFET is a transistor. Ugly back-ends or not, you have 2 foundries churning out tens of millions of chips using FinFET transistors.