Via: The PC's The Thing Wherein You'll Catch The Processor King

Continued from ‘Via: How Do My Readers See Ya?‘…

You’ve heard the expression ‘Man cannot live by bread alone’? Well, it’s my contention that x86 suppliers cannot live by non-PC design wins alone, either, especially when they’re competing against folks like AMD and Intel who can play equally well in both the PC and embedded spaces. To that point, here’s an irony for you; although Via’s boards-based focus is a critical piece of its short-term success plan, within it may lie the seeds of the company’s long-term demise. Those of you who use PC market-developed building blocks in your embedded designs do so because of the rapid pace of technology development and volume-driven low prices…but that same rapid pace applies equally well to obsolescence, of course. As a result, you build flexibility into your designs; so what if a newer replacement component runs faster than its predecessor and faster than you need, as long as it still fits within your bill-of-materials cost and power consumption envelopes?

As my 2004 study showed, and as more recent AMD and Intel data confirms, you can go mini-ITX without having to also go Via. Given that Via wants to make nano-ITX and pico-ITX boards industry standards, the same logic will inevitably also apply to those form factors once their market size grows big enough to draw AMD, Intel and their respective partners’ attentions. So what if your Intel-based (and lower-priced than Via’s alternative) ITX board goes EOL? Swap it out for a newer model, and forge on. I’m being intentionally simplistic in the interest of brevity, but I think you catch my drift.

The other thing that drives me batty about Via, I confess, is the company’s poor execution even in markets that it vigorously claims as its own. Take mini-ITX. Yes, Via standardized the board dimensions early on and (to the best of my knowledge, at least) hasn’t varied them since then. But the locations of various connectors (power, fan, memory, IDE and SATA, USB and IEEE-1394, etc) on the board have ‘evolved’ over time and multiple board generations. All of which drives partner companies like Casetronic (enclosures) and Mini-Box (power supplies) absolutely insane.

Or take the Via-based NAS that’s sitting un-powered in my living room right now. Why’s it un-powered? Well, I’ll tell you. Back at Christmas, I worked with Via on the assembly of a Windows Home Server-based system leveraging a fan-less C7 CPU variant of the EPIA-SN motherboard (whose documentation, you’ll note, indicates that it’s explicitly targeted at network storage appliances) and a Travla case that Caselogic modified (since I had no need for PCI expansion) in order to shoehorn two 3.5" HDDs inside for O/S-level mirroring of critical files. The system arrived in mid-January, soon after I returned from CES, and almost immediately I encountered glitches; missing software, along with insufficient documentation that (for example) didn’t identify which of the board’s two CAT5 ports was GbE-enhanced. Those speed bumps got resolved in fairly short order.

I’d noticed right away that the NAS was noisy; the system fan continuously ran at full speed. In a closet, this might not be a problem, aside from the fact that mechanical subsystems like fans (and HDDs) tend to be among a system’s weakest Achilles’ Heels from a long-term reliability standpoint. But with the NAS in a corner of my living room, the din was unacceptable. I immediately alerted Via to the problem, along with relaying the contents of a CES conversation I’d had with a Via manager, who’d admitted to me that the motherboard’s BIOS support for power management was at the moment immature. Three weeks later, after multiple proactive reminders on my part regarding my request for a report on BIOS support and plans regarding system fan control, I finally got back:

Currently, the option is not available in the BIOS. So you can read the RPM but you cannot control the speed.

Followed shortly thereafter by (after another inquiry for clarification by me):

Yes, there is a plan to address the noise issue for EPIA-SN by BIOS update. I think we can resolve it in mid March. I will follow up with an updated BIOS when it’s available.

Then I waited. And waited. Finally, in late March, I ‘pinged’ my Via contacts for an update. It took three more weeks, and several shuffles to new Via contacts (accompanied by requests to re-tell my tale from scratch), but finally on Tuesday came the following (with typos that I’ve corrected):

We don’t have any plan to include the system fan control function to our board. We would let system integrators solve the temperature issue since we don’t know the whole system collection. Therefore, the system fan will run at full speed.

I reminded my new contact that her predecessors had promised a BIOS update to add system fan control support, and asked if I at least there were temperature sensors on the board that an O/S-level utility like SpeedFan could tap into. One day later, I got this (typo-corrected, again) follow-up:

We don’t have temperature sensors on the boards since our customers don’t need that. We don’t know what customers’ solutions are, and the shape and the size of the chassis would affect the temperature, too. That’s why our customers don’t need this function from our boards. As far as I know, other vendors don’t provide this function on their boards, either. BTW, our CPU can provide fan control manually. You can set it up to stop/turn on the fan at a specific temperature.

Here’s the truly ironic bit. As I was sorting through my email archive in researching this writeup, I came across correspondence from mid-January. Check this out (once again typo-corrected by yours truly):

Continue reading with ‘Via: Hands-On Frustrations, Conclusions And A Request‘…