Intel sneaks out low-power microserver chip Intel has not made a big fuss about it, but the chip maker has kicked out a promised 15 watt processor aimed at the fledgling and sometimes cloudy microserver market. The chip, which is paradoxically called the Pentium 350, was promised back in March when Intel pre-launched the "Sandy Bridge-DT" Xeon E3-1200 series of chips …
"....or I can buy a quad core ARM 1.3GHz chip...." You missed the bit where you try to convince your customers that it's more fun to code new software, with no guarantee of success, rather than take advantage of the massive amount of proven commercial and freeware x86 apps that will already run just fine on the new "Pentium". Good luck with that!
"Why on earth would I try to convince my customers of that?...." Because the generic, everyday apps used by most businesses just don't exist for ARM. Where's the M$ SQL for ARM? How about Oracle DB for ARM? Not every business is ready to try roll-your-own Linux builds on ARM with OSS apps, otherwise they'd already be doing it on x64 and M$, Oracle and a raft of other companies would have already curled up and died. Businesses buy proven, trusted software, they don't buy risks. Until there are proven commercial stacks on ARM to match those for x64 it will be niche at best (NAS servers, webservers maybe).
".....the software that I write is completely processor agnostic...." Yes, but I'm betting you don't make generic software applications for the general market. In commercial apps, different versions for different processors will be tuned and tweaked to try and gain the best market advantage. A simple example is that the Oracle DB for Windows is heavily tweaked for x64 (as is the Oracle DB for Linux), whilst the version for SPARC-Slowaris is probably going to get a lot more attention than the versions for AIX or hp-ux. Despite the last three being commercial UNIX versions, all C-based and BSD-compliant, I can't run the release of Oracle DB for one on any of the others, and that's not just because of OS differences but because they have all been tweaked and recompiled to get the better performance on the different CPUs.
"......Are you from the past?" No, I just happen to think about realities a lot more than you obviously do.
"I'm betting you don't make generic software applications for the general market."
The "general market"?
Didn't you just use the lack of Oracle DB on ARM as an argument?
Is Oracle a big player in the "general market"?
Also, did you miss the part that said these chips are for micro SERVERS?
Good to see you haven't given up on slipping in the odd "Slowaris" or two into your posts at any and every opportunity though Matt, nicely done there.
Hmmm, FAIL or WTF? icon?
Bugger it, I'll just go with Paris
"......Didn't you just use the lack of Oracle DB on ARM as an argument?...." Sorry, I was trying to think of a generic app that was to be found in many businesses on current x64 servers that you insist ARM will replace.. Obviously, I didn't aim low enough to reach the bottom-of-the-barrel type business you work in.
"......Also, did you miss the part that said these chips are for micro SERVERS?...." Erm, M$ SQL and Oracle DB are both generic server apps, they make a lot of money for both companies and are to be found in many businesses. What was your point, beyond the small fact that you know nothing about the software market? Oracle and M$ SQL are often sold embedded as the database in many of the apps used on even microservers. Unless you find a market which needs the advantages of low-power and yet is willing to write their own stack (POS or maybe HPC), you will need commercial app support.
".....Good to see you haven't given up on slipping in the odd "Slowaris" or two into your posts...." Ah, now I see the reason for your lack of insight, it's just that old Sunshiner bitterness. I'm not surprised a Sunshiner would have zero visibility of either the hosted environments or small businesses that would run microservers, seeing as Slowaris is simply not prevalent in those niches (well, not in any niche nowadays). The one generic area which is truly platform agnostic - webserving - in which ARM might find a foothold, was surrendered by Sun to Windoze and Linux many years ago. Businesses still mainly use Windoze Server by a massive margin for the simple reason it is easy to buy, easy and cheap to hire Windoze admins, and easy to utilise existing software stacks already proven in the marketplace.
ARM might make a hit in niche areas like POS devices or big HPC clusters, where proprietary Linux apps can be written for each customer and the power-savings make sense, but in the general server market it will need a swathe of generic appliations to succeed.
You're not seriously trying to convince readers that Intel will let this be a fair competition based on price, are you? There are far better ways than cheap chips for Intel to keep ARM-based systems at bay, at least where mid-tier Intel customers are concerned (Dell are Intel-dependent and HP have allegedly already taken the ARM bait).
C'mon, do you think people can't (a) read (b) remember ?
Actually, you could be right Bruce.
small systems where the gui isn't relevant... if they are cloudy, were are probably talking lots of them so cost is an issue and we are probably looking at companies with skills.
Intel may have a go at the home market (windows home server) but at those prices they'll have difficulty competing with the arm-linux-web-gui appliance market.
The likelihood is you'll just be using independently threaded workloads (web farms etc) if you are buying lots of small systems and those are probably already on linux.
Intel probably needs to bank on small MSSQLServer apps which don't need much oomph but do need an intel chip.
Sorry Intel (and Matt) but processor architecture lock-in just doesn't work in the real world when performance matters.
Have you completely forgotten that Apple changed architectures and made all their die-hard fans buy new software not once but TWICE!?
Some grumbling, a brief learning period, then great happiness with the new and better technology.
Your memory obviously doesn't stretch back far enough to remember the complete failure of Apple's server bizz, and their inability to take marketshare from M$ or even Linux. Selling pretty toys to the sheeple is a far different game to selling real computers to businesses.
"Businesses buy proven, trusted software, they don't buy risks"
That is truly hilarious.
Tell that to anyone who's observed the tens of millions of corporate ERP victims around the world (SAP, Oracle Apps, etc), and they will rightly fall about laughing.
Tell that to anyone who's observed the hundreds of millions of corporate MS desktop + email rollout victims around the world (you forgot Vista already? Office 2010?) and they too will rightly fall about laughing.
Many corporate IT departments (and their SME equivalents) are fashion victims, with little interest in maximising the business benefit and lots of interest in maximising the size of the empire.
Occasionally a sensible idea gets to be fashionable. Is ARM going to ride that wave?
"How about Oracle DB for ARM? "
Why would Oracle bother until there's money to be made, which there hasn't been (and still isn't, at least till some high-volume ARM hardware and common OS flavour arrives).
Of course, now Oracle are rather attached to SPARC, they may not bother at all.
Of course also, Oracle devs may have some free time now they've dropped IA64 support, maybe they'll use it for ARM developments (maybe it'll be a bigger market than zero).
You're response simply underlines the argument I presented.
"....Tell that to anyone who's observed the hundreds of millions of corporate MS desktop + email rollout victims around the world...." Yes, and how many of them dropped Windoze and went for Linux desktops? They're all queuing up for Windoze 7. The current hype in desktops seems to be around VDI and thin-client desktops (oooh, bad flashbacks!), but still around Windoze. Yet fully-featured OSS desktop offerings have been available for years. A new offering has to offer both monetray savings AND business performance advantages. The PC-Windoze combo did that, freeing up a lot of time and money that was tied up in mainframe-client style setups. Simply saying "you should buy my solution because I say it is technolgically better and because I don't like Intel" is not going to get many businessmen jumping on your bandwagon, especially in a recession.
"....Many corporate IT departments (and their SME equivalents) are fashion victims...." Call it what you like, but those "fashion victims" are the ones spending the money, and so they are the ones the vendors will make products to please. That's simple economics, only it seems not quite simple enough for you to comprehend.
".....Why would Oracle bother until there's money to be made....." EXACTLY! It's a chicken-and-the-egg scenario - a buisness will look at an innovative offering and check that it at least offers what they can do now. For ARM to succeed in the server biz will require it offering at least a comprehensive core application portfolio of those popular business applications as x64. Companies like Oracle will not invest in the porting work unless they can see a profit or are incentivised by a platform vendor, and all the current platform vendors have too much invested in x64. Businesses will not invest unless they see the apps up front. No chicken, no egg, no apps ecosystem to push ARM's boat into the water.
".....Oracle devs may have some free time now they've dropped IA64 support...." Apart from the small matter of a courtcase that Oracle has to win for that to happen, you are only underlining my argument. Oracle's attempt to kill the Itanium bizz is by threatening to remove a popular application from availability on Itanium. ARM doesn't have that availability now, and (as far as I can see in Oracle's public roadmaps) won't have it. So funny that your own ranting shows the gaping hole in your argument!
/SP&L
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
