For Servers, HP Bets On ARM (Again)

[Guest]

ARM-based chips should finally be ready for primetime when it comes to the server market, and HP is betting strongly on them with the announcement of the Proliant m400 and m800 ARM-based servers.

For Servers, HP Bets On ARM (Again) : Read more

 

[IInuyasha74]

I don't really understand this. Why use ARM to do this and have to get around all the barries and troubles with making an ARM server like this when MIPS servers based on the same idea have been built for over a decade now?

 

[anthony8989]

Isn't it obvious? ARM chips cost less. Plus it's built on a more malleable ecosystem which will be beneficial later on.

But mostly it's about the bottom line.

And it's only natural that ARM starts branching off into auxiliary micro processor markets .

 

[NotExpertAtAll]

yeah I agree. with ARM the bottom line is cost of ownership and data center's power use is becoming the thorn on every companies overhead expenses. If they can reduce expenses on data center operations, they would have invested on MIPS and ARM due to lower power usage per cpu core.

But I think Intel Atom/Bay Trail is one of the competition on this environment too.

 

[ethanolson]

ARM makes more sense than MIPS because the programmers available as well as the ported OS infrastructure. It's comfortable and commoditized.

 

[IInuyasha74]

MIPS sells complete CPUs for just as cheap as ARM.
They are fully compatible with both Windows, Android, Linux, and several other operating systems.

 

[anthony8989]

The key factor that will drive adoption of the ProLiant m400 cartridge for Moonshot or any other ARM-based servers will be whether or not they can deliver measureable value for customers when compared to their existing server solutions and for that matter, what other processor manufacturers brings to the table.   For the majority of large scale datacenters, this value is measured in total cost of ownership (TCO) which includes both the acquisition cost of the equipment and the operating cost of using this equipment during its lifetime.
http://www.forbes.com/sites/patrickmoorhead/2014/09/29/hewlett-packard-launches-the-first-64-bit-arm-server-with-applied-micro-circuits-x-gene/

 

[ldo]

I was going to suggest MIPS, too. As the world’s second-most popular CPU architecture (after ARM), they should be worth a shot.

 

[IInuyasha74]

That and MIPS was originally designed for severs. In fact some of the most powerful super computers in the world ever built use MIPS.

 

[anthony8989]

"Servers " is too vague a term in this context. That's probably what is confusing you. This Proliant m400 is geared specifically toward "simple data transactions (such as web requests), that can scale well and that use little power."

 

[IInuyasha74]

"Servers " is too vague a term in this context. That's probably what is confusing you. This Proliant m400 is geared specifically toward "simple data transactions (such as web requests), that can scale well and that use little power."

I'm not confused at all.
http://en.wikipedia.org/wiki/SiCortex

SiCortex for a long time made scalable MIPS based servers. The most powerful of these could do 8.2 teraflops of computations. It would take roughly 87 Intel Core i7-4770k CPUs to do the same amount of work, while consuming more power. These systems could be scaled all the way down to 10w 6 Gflops systems or anywhere in between, and this was back in 2007. In that amount of time MIPS has significantly reduced power consumption and increased performance. They sell for roughly the same amount as ARM, were designed for servers, have a history of being used in servers, work with all major OS types with the exception of Apple, and offers higher performance per watt.

While I can understand while building desktops like this is a bad idea, since Single-threaded performance would be terrible and applications don't have near the ability to make use of so many threads, it makes no sense not to make severs out of them.

 

[anthony8989]

You are far more confused than I initially assumed. You don't understand the purpose of this device.

 

[anthony8989]

I'll reiterate, you said:
"I don't really understand this. Why use ARM to do this and have to get around all the barries and troubles with making an ARM server like this when MIPS servers based on the same idea have been built for over a decade now?"

Answer:
"The key factor that will drive adoption of the ProLiant m400 cartridge for Moonshot or any other ARM-based servers will be whether or not they can deliver measureable value for customers when compared to their existing server solutions and for that matter, what other processor manufacturers brings to the table.   For the majority of large scale datacenters, this value is measured in total cost of ownership (TCO) which includes both the acquisition cost of the equipment and the operating cost of using this equipment during its lifetime.
http://www.forbes.com/sites/ "

This is a new device. It has yet to be proven in application . But as the title of this article reflects " For servers , HP bets on ARM (again)" .

Big Ole HP thinks ARM's got something worth investing in. HP's opinion > your opinion. womp womp 🙁

 

[IInuyasha74]

Well suppose can't really argue with the last bit, they must for some reason think this worth investing in. I just seriously question the idea of using the RISC chips designed for tablets and smartphones inside of servers as opposed to the RISC chips designed for servers.