First and foremost – Atom represents a significant departure from what we’ve come to expect from Intel.  It would almost be like Microsoft releasing a build of Linux (well, OK, maybe not that radical).  Atom is all about “just enough” – just enough speed, just enough performance, just enough of everything.

Remember – we’re talking about Intel here.  Intel – the core king.  Intel – the horsepower junkie’s enabler.  Multi-core, multi-socket, HUGE memory footprint Intel.

So – what is Atom?  Pulling from Atom’s Product Brief –

·         Up to 1.86GHz (core), 533 MHz FSB

·         2.5 watts TDP

·         45nm process

·         Hyper-Threading

·         Enhanced low power sleep states

·         Dynamic L2 cache sizing

·         SSE2 and SSE3 support

·         Embedded lifecycle support

So what’s the big deal?  Take a close look at bullets 2, 5, and 8.

Power

Consuming roughly 2 and a half watts for a 1.86GHz clock is unheard of.  Think of this in terms of previous (and current) mobile CPU power consumption.  Think new lows in low power.  Think solar panel low.  Think portable radio low.  Think small batteries and a decent lifespan on that battery.

Sleep States

Does 2.5 watts get you interested?  How about new sleep states that flush the cache during periods of inactivity, further reducing power use and extending battery life?  If you flush the cache, you don’t need to power the cache to maintain data.

Embedded Lifecycle Support

If you’re relatively new to embedded, this won’t mean much at first blush – but does an expected lifespan of 7 years appeal to you?  7 years for a core component in the PC market?  Yes, 7 years of expected availability, without change.

Where does Atom work best?  Remember my “just enough” comments?  Take a look at the Windows Task Manager on an average desktop PC and move over to the Performance tab.  See the graphs?  Those graphs show you two important system status on two key functions: memory and CPU usage.

Now, take a look at your average CPU usage (disregard the peaks you see when you start an application – just watch the graph and mentally look at the average).  What do you see?  I’m looking at mine as I write this and see the following:

When I took this cap I was running the following applications:

Microsoft Word – 3 documents open

Microsoft Outlook – main application and 3 message open

Two copies of a browser

One LARGE spreadsheet

Corporate AV running real time protection is running as well

1085 threads and 61 processes running

CPU usage?  Hovering around 8%.  Yup, 8% is all I’m using.  I’ve done this a couple of times over the last week or so and have seen consistent results around the 8% to 10% mark.  So by that measure, my current CPU is overkill for my usage.  I’d venture that if you do this yourself, you will see similar results.

So – how does Atom help?  Atom certainly isn’t a barn burner by current standards.  At “only” 1.86GHz at the top speed bin, you may be thinking “that is so slow it would barely run GoogleChrome” but in the case above, Atom would be enough.

Just enough for a new way of thinking

Where Atom will shine first is in low power, extended life mobile devices.  It has enough horsepower to run XP and a host of applications.  Can you compile a few hundred thousand lines of code while you watch a DVD movie and crunch a long SQL query AND render a 3D image?  No.  But when was the last time you did that?  No, Atom isn’t going to replace multi-core, high performance processors in the mainstream computing market, but it will be an ideal fit in a new generation of ultra mobility designs and has enough “oomph” to run modern operating systems.

Only the market will tell if Atom is a hit, or I’m nuts.

I’m hoping the former. 

·         Could be a desktop refresh. 

·         Could be a server migration project (retiring old, migrating to virtual). 

·         Could be that a department is growing and needs more storage.

·         Could be an M&A expansion. 

·         Could be…….

………..well, you get the idea

Next question – what exactly do you need, and where will you find it?

There are certainly a lot of choices – local vendors, whitebox, multi-national corporations (MNCs), regional VARs, large and small OEMs, build it yourself, and a host of others (I’m sure if we did a raise of hands we’d see scores of other options).  What’s the right choice?

The easiest choice is to buy from an MNC – nobody ever got fired for buying from them, right?  They have a huge presence – web, television, print.  They have a huge sales force (well, if you do enough business for them to notice you anyway).  Their prices are almost unreal.

But is the MNC the right fit for your project?  Do they offer the peripherals you want from the manufacturers you prefer?  And who, exactly, built their “stuff”?  Does the product selection from the MNC look like this:

Look at stock product lines on the web or in the latest ad slick/line card.

Pour through pages of products and somehow not find what you are looking for (how is that possible?)’

Move on to find a product that is “close enough” that you can tweak a bit to be a good fit

Check what, if any, onsite help you can get

Is there another choice?

I’d argue there is, and in many cases those other choices are worth exploring – their names might not always be top of mind, but when you can find a supplier/partner who operates in the same business mode you do, who offers a wide variety of robust and well-engineered projects, who also offer quick, capable support – take a second look at them.  You may find that not only are their products competitive and designed/manufactured by some really big names in the IT business, but that same supplier is more than willing to work with you on your projects, can offer to pre-test and pre-load your software image or a specific driver version you need, can even source ingredient components for your system which are not on their standard component lineup.

So, how do you make the choice? 

Having spent a lot of years in R&D, and talking to more IT managers than I can recall in that time, there are five questions I would ask anyone, regardless of size, before I signed on the dotted line:

Q: “Who actually makes your stuff?”  Or, said another way, “Who is your ODM (Original Device Manufacturer) partner?”  Very few IT suppliers really solder chips and fab boards – most rely on ODM partners to do that.  If you get a direct answer like “well, we do of course” try again – “no, who actually solders the chips on the boards and who designed those boards?”  Expect a simple, straightforward answer.  A front line sales person might not know, but they should be able to get the answer for you and should have no qualms about doing so.

Q: “How do I get parts replaced?” or “Can I get a failed part replaced without having to talk to your support staff?”  Your time is valuable, look for a supplier/partner who understands this and proves it with service offerings. 

Q: “What sort of customization is available?” – no, this doesn’t mean “what color do you want it?”  Suppose you need a server with, say, an Infiniband card pre-integrated and pre-tested.  If the specific card you need isn’t on the partner/supplier’s existing option list, can they add it?

Q: “I need a single image installed on every system – can you do that?”  If you provide a stock image with your operating system, driver set, applications, and configuration – can they preload your new systems with that exact image?  This can save a huge amount of time in large deployments, especially if systems are installed at multiple sites.

And last, but far from least:

Q: “Why should I buy from you?”  The answer here can be a deal maker, or a deal breaker.  Answers like “because we are the cheapest price” are OK if you are just looking at acquisition costs, but if that were true you probably would not be reading this.  A clear, concise, and consistent statement of the overall value backed up with concrete examples of exactly how that value is fulfilled is the minimum I would accept.

There’s an old expression: “You spend your money and you make your choice” and it is nowhere more true than when shopping for products for a new IT project.  Money is tight, spend carefully.  Ensure you get the best value.  Look beyond the obvious choices. 

Well if you’re reading this first entry, let me start by saying “thanks.”  Although blogging is pervasive, getting started with one’s own blog can be a bit daunting.  So bear with me for a bit while I get the hang of it.

I suppose I should start with some background about me – a subject which I well qualified to discuss.

I’m a middle aged guy, a bit round in the middle and sparse on top.  I’ve been in the computing industry for about 15 years now.

I started my professional career in Sales, working my way through high school and college at a local retail shop in small town America.  I was there for about a dozen years – starting as a counter rep and work my way through shipping/receiving, finally ending up as a buyer.

Sometime in the mid-90s I decided I needed a career change. I’d tinkered with computers during my college days (this was way back in the green screen / text display days) so I decided to update my skills a bit and spent a year or so in public service in California as a PC tech.  I learned a lot – both about technology and about people.  I learned that you never want to lose the “Configuration Diskette” when working on IBM PS/2 systems and I learned that a double beep at boot was OK, so long as it came from a Compaq.  I learned that many users “just want it to work” and don’t care how, that some users really do care “how,” and there are varying degrees between.

I left California and moved to Idaho to work for Micron Electronics (now MPC).  I started out in Tech Support (I’m sure I’ll be sharing anecdotes from those days in future entries) and survived the introduction of Windows 95 (no dig at Microsoft here).  I also lived through supporting Windows NT 3.5.1 and 4.0 (often at the same time) as well as WFWG 3.11.  Ahh….those were some heady times.

I later moved on to managing special projects in the support organization.  I launched and managed an expert system diagnostic program (resident on the end user’s PC) to try and make support more efficient and also evaluated several knowledge base systems (and helped launched the final winner) – all to help provide better support.

As PCs got easier to use and standards really became standards, I moved over to the qualification team and helped develop test protocols for ensuring interoperability.  I was at the forefront of the transition to soft modems and developed a benchmark that convinced the “doubting Thomases” of the world that, yes, there were sufficient CPU cycles to give a soft modem good performance and, no, they would not spontaneously quit working.

Subsequently I moved to Server platform testing and finally to managing the team that did it.  I spent about 20 minutes leading all of Engineering, but that is a story I’ll save for another entry.

Which brings me to today – Director of R&D for MPC.  I love this job (and no, I didn’t write that in case my boss was reading this).  I get to speak with quite a few of our customers, I get to hear what makes their lives difficult (at work) and try and collaborate on ways to make that better.  I get to work with partners and suppliers and regularly evaluate new technologies, build prototypes, and still convince the “doubting Thomases” that we can really do what I propose.
-Doug

Purpose: Provide a resource to MPC customers through…·  Insights into MPC Technologies·  Best Practices in Deployment of IT Solutions (based on customer experiences)·  Expert Commentary on IT Trends·  Managing IT Costs