ARM upstart Calxeda pours $55m into server chip war chest

ARM chip upstart Calxeda is lining its coffers as it prepares to do battle with its 32-bit EnergyCore ECX-1000 processors, and two more cores in its roadmap, to conquer some corner of the server world. Calxeda now has more than 100 employees, who work in its Austin, Texas headquarters as well as in development labs in Silicon …

[Anonymous Coward]

64-bit vs 32-bit

does a web server really need to be 64-bit to work? Have seen at least one web development contract where customer requirement dictates x86_64 but no reason whatsoever for that. Mindset is probably a bigger hurdle than availability of an O/S.

Better still, if a mainstream server vendor - or better still, a major vendor with a server business but other revenue lines so a little freedom to experiment (Cisco, Oracle) - could see the value in doing mixed-architecture blades, the way Sun had briefly (you could at one point have a 6000 chassis with Opteron, Xeon, and SPARC all in same enclosure, with one enclosure meeting whatever mix of workload) then that would be interesting. Focus on quality not low price (so no "low sticker price" servers that then break the bank when you add normal things like LOM at £200, or rack rails at £100 a pair). Maybe not as blades, maybe more as smaller units, and with pizza-box equivalents, then people could decide "actually, my database needs to be 64-bit because of software version, but my image cache server can be whatever's cheap". Let customers decide, not manufacturer;s marketing departments.

If they get this right - and not just Calxeda but the other players, obviously - then things could become far more interesting than they have been for a while.

[druck]

Re: 64-bit vs 32-bit

AC@17:42 writes: Have seen at least one web development contract where customer requirement dictates x86_64 but no reason whatsoever for that.

There are a number of reasons for specifying x86_64 instead of x86_32 over and above the 64bit memory handling, such as more registers and additional instructions.

[Anonymous Coward]

Re: 64-bit vs 32-bit (@druck)

apologies druck, i should have been clearer: the specific customer requirement did not include any reason whatsoever for that.

[Alan Denman]

a bitty article.

"32-bit processors have limited appeal to programmers".

Programmers do not give a stuff. 64 bit actually has many pros and cons so is not that important.

If we ever get to 128 bit it will be a bit of a case of 'the dial goes up to 11'

[Steve Davies 3]

How will Intel and AMD react?

Quote

All of the ARM server chip players – Calxeda, Marvell, Applied Micro Circuits, Cavium, and possibly Samsung Electronics – will also be gated by another factor. How Intel and AMD react to the ARM uprising and what they do to bring computing and networking closer together in future systems.

Easy.

1) Sue the blighters into oblivion. The VC's will soon run for cover.

or

2) Buy them out and bury the technology for 5 years plus put golden handcuff on the honchos so they do leave and statup all over again prompting a rinse/repeat cycle.

2) is probably what the VC's want all along so they can get a nice fat profit in quick time.

[mkihailp]

ARM is not there, but will be soon

Basically, APM.com presented the first ARM64 jump into the pool of Enterprise Computing (EC).

But..., X-gene is only a 64 bit CPU with RAS and someone has to proof that is is EC capable systems.

it might be not so easy to build something large in SMP or cc:NUMA and to be competitive against current EC architectures, like POWER, x86, SPARC, but it is the right direction to go. It right just because the power consumption of ARM is much lower than anything else in the 64 bit world. BTW, SeaMicro did almost the same with Intel Atoms, so for ARM64 is seems as a promising future.