I agree with you regarding Intel and profit opportunities, if they can make money on it they will. I was wondering what all the hoopla was about the ARM offering in the server market, but given the lower power requirements for ARM cores it makes a lot of sense. There are huge server farms that use an incredible amount of power and a machine / family that can save 5% or more would be worth the re-investment in the long run.
Intel putting pressure on ARM? That's a joke right? There's no significant way in which Intel is putting pressure on ARM right now.
ARM is the one attacking Intel's markets, and they may not be terribly successful YET, but Intel is the one feeling the pressure! And their profits will drop dramatically over the next 2 years because of it. Because they have to compete with ARM chips which are a lot cheaper.
A lot of server software will support the 64 bit ARM architecture from DAY ONE. In fact many already do support it, and it's still a year to go.
This is not like Windows and Windows RT. Most servers use Linux, and it has already been adapted for ARM.
People may recall that Intel once offered a line of ARM processors under the StrongARM label. Intel got the line from Digital Equipment as part of the settlement of a suit. StrongARM got rebranded as XScale, and XScale was eventually sold to Marvell in an Intel reorganization.
ARM is on a roll, due to perceived power advantages over Intel. ARM chips just use less power, and power requirements and associated costs are an increasingly large concern for data centers. With a line of 64 bit ARM cores in the offing, ARM based servers in the data center become a possibility. It's no great wonder an assortment of companies are looking to address the market.
I'm unconvinced by worries about software. What software actually runs on the server side? Offhand, the OS itself, and software that supports the server role. Linux is a very popular server OS, and it already runs on ARM. Windows 8 has been ported to ARM, so if MS sees a sufficient market, we can expect a Windows 8 Server implementation for ARM processors. In the current environment, other things that run server side are databases and web servers. Oracle is anyone's guess (though I'd be surprised if there wasn't an active porting effort), but MySQL has been built for the Raspberry Pi. Dell contributed a sample ARM port of Apache to the Apache Software Foundation.
Bottom line, if this takes off, I expect Intel be in the market, too. They were an ARM licensee, may still be, and want to sell silicon. If that means ARM because there's a market big enough to profitably address, I don't think "not invented here" will stop them.
In 10 years there will likely be three survivors, but who knows who they will be right now? As we reach the end of the road on semiconductor scaling, whole new areas of innovation must be found. 10 years from now, I think the market will look very different than it does now. Ten years ago, who would have guessed Apple approaching $200B in annual sales?
If AppliedMicro and Calxeda gain traction early they will be acquisition targets. Neither has the scale to compete in true server markets and they won't be able to grow fast enough on their own to fill the space. The market won't tolerate Intel as the only source of server/data center chips. There will clearly be an alternative ARM ecosystem that competes with Intel.
Market for ARM-derivative server on total Sandy Bridge 32nm displacement mkt is 113,594,361 units at revenue potential of $72,395,536,747 on Intel 1K price; Average Weighed Price $637, Average Marginal Revenue $293, Average Marginal Cost $344.
Intel AMC for Sandy Bridge E3 ramp is $117 & thru Ivy Bridge E3 run down drops to $83. Intel Price to Cost ratio for Ivy E3 v Sandy E3 increases 6.85%. Where Price to Cost ratio between SB full run & Ivy ramp jumps 38% indicating after flooding channels with SBridge at 32nm, Intel monopoly raisex price thru Ivy ramp; naughty Intel.
Market for ARM server on per component basis by Intel Price Structure thru 24 month full run is 25 mil ATOM priced sub $100; 7,466,263 EN/EP/E3 priced sub $250; 24,451,308 EN/EP/MP priced sub $625. 56,917,571 units is roughly one traditional Intel desktop short run.
Considering multiple ARMs on blade to reach Xeon perf parity; on ratio 4 ARM 32 bit Quads to 1 Xeon 64 bit Hexa possibly a mkt for 225,670,285 ARM units. Analyst suspects at 64 bit ratio can reach 1:1 perf for specific workloads in realms that are today LAMP Stack, moving into HADOOP Clusters, HipHop, OpenStack.
The key issue for ARM-derivative success in server will be foundry capacity, growth in capacity, stability of supply. Lacking growth in capacity with allocation order of fabless design producerís established winners can be anticipated before the race even starts.
Into 2013 analyst suspects ARM server will be flattened by Intel monopoly production technique sans owning the foundry advantage. There will be merchant and captive vertical producer examples. There will be design producerís trading off mobile supply for server supply. Analyst suspects a reconfiguration by charter players to survive through ARMís Dunkirk, followed by second wave of assault into Intel land lead by fast followers into 2014.
APM, Calxeda, Cavium, HiSilicon/Huawei, Marvell, Nvidia, Qualcomm, Samsung, TI are addressing ARM-derivative Server.
Three original contenders; Calxeda, Marvell, APM were well categorized; low, mid, high perf segments.
Calxeda & Marvell initial efforts have been shielded from Intel on unique 32 bit engines which Intel does not address in portfolio. Switch in SOC this analyst suspects from Intel during 26 nm gate length run end before true 22 nm shrink; exposes Intel to political issues among NIC, switch & route compliments in Intel server environment.
Nvidia & TI appear uniquely segmented into HPC, where TI w/DSP accelerator and Nvidia w/GPU ala CUDA on ARM control plane address FLOPs instead of MIPS and IOPS.
Analyst suspects Nvidia large multicores v Intel Atom Octa and bridge power reduction between discrete CPUs on blades.
Needless to say end market is pressing for ARM 64-bit, and as usual, ARM willingly licensees all comers to see what sticks.
ARM under financial pressure in slow market seeks a royalty hedge; insurance. Where Intel has won 3 out of first 4 ARM offensive moves into notebook, Smart TV & Signage where Intel won without a fight, where in server Calxeda does reach first base, only A15 in phones and tablets brought in ARM base holders.
Intel has ARM cluster surrounded in its strongholds and ARM Ltd. is feeling the squeeze.
Intelís crush campaign targeting ARM and AMD has been especially effective dumping 24,993,201 Sandy Bridge Celeron at cost last & current quarter.
Celeron dumping will be followed by 25 million dumped ATOMs & Ivy Bridge sludge through first half of 2013.
Finally, Intel confidence agents have been effective curbing ARM ability to compete by disabling ARM Ltdís strategic marketing functions on all the old hazards of Intel relations and the politics of Intel Imperial Court.
I am guessing he is about 3 too high on the count of survivors. What exactly can an ARM server core do than an undervolted x64 can't? And all that software that needs to be rewritten, retested, and recertified. It's not exactly that Intel will allow itself to be seriously underpriced, and this removes the only possible way these things can make inroads. Plus finfets should let them drop supply voltage even further. I smell an Itanium.
ARM's A9 & A15 cores will do just fine for :
HTC = High Throughput Computing
and not really for :
HPC = High Performance Computing
A15 need to put a cap on its peak power consumption in the range of less than 2 watts/core should be the target in the longer run.
Many-many, too-many cores will help to build next-gen data centre ARM server processor.
Data centre server processor should handle maximum small threads in the range of 16/32/64/128 simultaneously.
Data centre software threads are not compute (math) intensive and hence having many-many cores to handle many-many threads on a single chip will lead to increased energy efficiency and lower operating cost.
Blog Make a Frequency Plan Tom Burke 17 comments When designing a printed circuit board, you should develop a frequency plan, something that can be easily overlooked. A frequency plan should be one of your first steps ...