Via Launches New Line of Mini-ITX Boards

An anonymous reader writes "LinuxDevices has the skinny on Via's next-generation Epia EN mini-ITX boards, which feature its relatively new C7 processors based on the Esther C5J core. The boards will be able to run passively cooled at 1.2GHz, and will clock up to 2GHz, with 800MHz FSBs." From the article: "They target thin clients, car PCs, robotics, medical equipment, kiosks, and server appliances."

Comments

What does passively cooled really mean?

[Profane MuthaFucka]

I've got a ME6000 board that isn't reliable unless there's some air flowing over the heatsink. This was supposedly passively cooled, but I had to add a little fan blowing right at the heatsink to get the temperatures down from 60C to about 38C.

It even overheated when it wasn't in the box.

Re:What does passively cooled really mean?

[claes]

I can agree. I have an VIA EPIA PD-Series, with Eden processor. It frequently hangs so I have to turn it off and on again. I don't trust VIA on this - I will mount a processor fan on it but that pretty much gets rid of the point of getting a fanless processor to begin with.

Re:What does passively cooled really mean?

[Ulrich Hobelmann]

That's bad, especially if you consider that most chips are (AFAIK) ok until they reach about 95C. I had an Eden 533 which got really hot, but it never had any real problems (except its abysmal sound, networking, and graphics chip...).

Well, I sure hope these new beasts are at least reasonable in performance, in price (up to recently the old Eden boards never really fell in price, while other CPUs ramped up speed and lowered power consumption), and of course in temperature.

Re:What does passively cooled really mean?

[Anonymous Coward]

I think they mean with no real components attatched in close proximity . . .

Re:What does passively cooled really mean?

[Bradee-oh!]

I run a PD as my firewall/router/server. I suffered frequent hang problems and replaced the cheap power supply I was using with a known-good, better quality supply and haven't had a problem. This is running passively cooled. I've also had a few other Epia boards in the past. My experience says they are very picky on their voltages, but overheating likely isn't the problem.

Re:What does passively cooled really mean?

[FIT_Entry1]

I have a similar set up with an EPIA M10000 running m0n0wall, i literally threw it together using known bad ram, an ancient 1 gig hard drive, an old 10Mb ethernet card and have had months of uptime with zero problems.

Re:What does passively cooled really mean?

[moonbender]

60C isn't particularly hot for a modern CPU. My XP 1800+ gets that hot - with active cooling. The system should still run stable at those temperatures...

Re:What does passively cooled really mean?

[Afrosheen]

Sounds like someone needs to google 'thermal paste'.

  My Athlon64 3200 idles at 32C/89F. Then again, it's watercooled, but even with the old heatsink it never got very hot.

Re:What does passively cooled really mean?

[Malor]

If that's an accurate reading, something is wrong. You can run stable at those temps, but you should NOT be seeing those on a standard XP 1800+...45C is about as hot as those chips should get, with reasonable air cooling.

Re:What does passively cooled really mean?

[level_headed_midwest]

Eh, well, mobile chips have a Tmax of about 95-100C, but most desktop chips generally are toast at 65-75C. I would imagine passively-cooled chips like the Eden probably are in the former category as fanless units tend to run hotter than actively-cooled parts.

My 2.2GHz P4-M can get up to 100C and idles at 65C at 1.2GHz, while my 2.2GHz Athlon 64 4200+ can get up to only 65-70C and idles at about 26-27C (a couple of degrees above room temp.)

Re:What does passively cooled really mean?

[AngelofDeath-02]

You wouldn't happen to have amd's clockspeed throttling software enabled would you? That'll easily get you temperatures similar/lower.

Re:What does passively cooled really mean?

[toddestan]

I have found that 60C is pretty normal for the Athlon XP line of processors, especially if you have the stock retail cooler or one like it. They run pretty hot, and have to dissipate about 60-75W. In a warm room, my old Athlon XP 2000 (1.75V) used to idle at about 65-70C. In a cool room, about 55-65C. And that was with a dual fan power supply, the stock cooler, a side case fan directly above the CPU, and two other case fans keeping the rig cool. It ran stable, and I found (by experimenting) that it would be stable until about 85C.

I replaced the XP 2000 with a Sempron 3000 (Barton core, Socket A, 1.6V) and it is running about 41C right now in a cool room (about 62F/17C). If I pegged the CPU at 100% it would get to about 50C after a while.

Of course, the Athlon XP is not to be confused with the Athlon 64 line in terms of heat and power consumption.

Re:What does passively cooled really mean?

[Al Dimond]

Yeah, those Palomino core chips run quite hot. Mine is running at 1733MHz (that's 2100 in hype numbering) and idles at 55C, during big compiles gets around 63. Recently (perhaps from the case being jostled?) the thermal paste got pushed off the area of the die and the thing began to idle at 78C.

I wish it were easy to underclock the CPU, because I really need nowhere near that much CPU and would definitely prefer lower temperatures. Unfortunately you have to physically unlock it, which carries a bit too high of a risk factor for me.

Re:What does passively cooled really mean?

[moonbender]

We might just have a different definition of what constitutes reasonable cooling. ;) If my CPU did not heat up to more than 60C under load, I'd reduce cooling to make the whole system more silent. Having it running at 30C as some do is sort of a waste in my eyes; it's very unlikely that one of my components (with the possible exception of HDs) will die from thermal stress before it's obsolete. That said, there still might be something wrong, maybe the thermal paste I used was too old or maybe the heat sink just sucks (actually, the heat sink definitely sucks). But that's all irrelevant, I'm gonna get me a new computer later this week. Here's hoping it will run cooler, or rather, more silent.

Re:What does passively cooled really mean?

[pAnkRat]

That's one of the reasons I opted in to buy a "real" athlon 64.
Not because of the 64 Bit and speed, but because of the fact that you can change it's running speed with a simple "echo 1000000 > /sys/devices /system/cpu /cpu0/cpufreq/scaling_setspeed "

Normaly my cpu runs at 1 Ghz, and has a 31C temp.
when playning games, I bump the speed to 2Ghz, and increase the cpu fan speed.

Re:What does passively cooled really mean?

[GooberToo]

While I can not saying this is the correct answer for your situation, I must add that some passively cooled devices require a complimentary designed case with passive cooling specifically part of its design. I would not consider it unreasonable to expect a case to support an intelligent convection cooling design.

For example, many piston airplanes have engine bays which are specifically designed to cause convection cooling. Without it, even while in flight, the engines may over heat. This pops up because people tend to think that raw air pressure in the front of the cowl causes cooling. In reality, for may piston planes, it's convection cooling which actually provides proper and equal cooling of various parts of the engine. On many planes, there are baffels right in front of the crowl inlets and people sometimes remove these. As a result, cooling of the engine often gets much, much worse despite having greater air pressure into the engine bay.

What does engine cooling have to do with chip cooling? Well, just because a chip is designed for passive cooling does not mean it should not overheat when outside of its case. Bluntly, it is not uncommon for many CPUs and chipsets to overheat (or at least high heat levels), when the computer is operated with its case removed...just like an engine. Long story short, even passively cooled systems often require intelligent case design to ensure proper air flow is created.

Re:What does passively cooled really mean?

[Profane MuthaFucka]

That's very helpful. My case is a reworked Sun external hard drive box with little attention paid to cooling, so that could definitely be the problem. Everything works just fine now that I added a big fan turning slowly (quiet) right over the processor.

Re:What does passively cooled really mean?

[Kirmeo]

I got one of the orginal 500MHz EPIA 5000 motherboards. I put it in a cube case with a full size dvd/cdr and 40G HD. I replace the 150W power supply (with noisy fan) with an external 60W brick and regulator board. After a few hours, I noticed that the HD was really hot. So I put in a 12v fan, run off the 5v line, slowed down to the minimum speed that would always spin up when powered on. It's quieter and cooler than my DirecTivo.

Its about time they release these boards. I think I read about them 6 months ago.

Great, but will it support Virtualization?

[luckytroll]

This is a great platform, if you dont mind the slower speed of the C3/7 processors - but the thing that I have been a little miffed about is the unsupportability to run VMware - hopefully the C7 may fix this.

Re:Great, but will it support Virtualization?

[Anonymous Coward]

One thing that gave me great joy from the article...

"a full-speed FPU (floating point unit), rather than the half-speed unit of earlier Via chips"

At last!
Also, note on the die layout that space is dedicated to SSE/MMX. So, they are not just emulated and hogging the floating point like on some chips.

I'll have to see some more benchmarks, but I should be able to stop moaning about my PIII 550 being faster than 1Ghz via parts.

Re:Great, but will it support Virtualization?

[daniel23]

Hm, my Lan server is an Epia C3-800, running SuSE 8.2 (still). It runs VMware-2.0.4 and in it another SuSE which handles the (mostly harmless, unless I link my pr0n collection) http, ftp, ssh I get here. Both host and guest OS had their uptime wrap around last summer, so I'd argue it is an not entirely unstable setup.

Oh yes, and it doesnt overheat, either, in spite of me taking all the included fans out. It has a Morex Cubid 2677 case standing on its left side, thus having convection cooling.

Re:Great, but will it support Virtualization?

[MikeFM]

For small systems that don't need a lot of power I like the form factor, low power usage, and low heat (and fan noise) of the Via proessors. They're great for shoving into a small space or for use where you just don't want noise.

I wish they could work with Nvidia to come up with a built-in GPU that had enough kick to play decent 3D games on. My 6800 card is almost as big as, and certainly louder, than my mini-itx mobos and is definately bigger than the nano-itx boards I'm waiting for. I really want a powerful system, including a killer GPU, that can fit into a size similar to the Mac Mini (which sadly also has a weak GPU).

Re:Great, but will it support Virtualization?

[default luser]

Via is the king of "rebranding" existing technologies.

Nehemiah C3 has:

* 16 stage pipeline (just like C7 Esther)
* SSE support
* Full-speed FPU

You see, the only real improvement here is that Nehemiah is getting a facelift. It is moving to a 0.09 micron SOI process (about 2 years behind AMD and 3 years behind Intel), is getting a new bus interface, double the L2 cache, and is getting SSE2 support. Other than that, it is the same old Nehemiah.

The bus and cache upgrade are the most important part. People have been stating for years that the bus speed of 133 MHz plus the pathetic 64K L2 cache has held the C3 back. All the SSE2 units in the world won't help you perform if you can't move enough data.

Think of the Pentium MMX. New process and dual voltage (2.8v) meant lower power than the previous Pentium (3.3v). Double the L1 cache and MMX support meant better real and potental performance. Now, imagine the Pentium MMX also included an upgrade to a 133 MHz bus, and you've got the equivilant of Esther. It should perform 30-40% better clock-for-clock over the C3, and should be more scalable, but it is nothing amazing.

Re:Great, but will it support Virtualization?

[petermgreen]

and is definately bigger than the nano-itx boards I'm waiting for
the via nano-itx boards sadly don't have a PCI slot.

Re:Great, but will it support Virtualization?

[MikeFM]

True, I never really used my mini-itx boards' pci slots though. Now and then for wireless but nothing else. Built-in wireless and bluetooth would be cool too. I know.. how much can they cram into that little bitty space. I'd like to find out if you can buy the boards without external connectors too as I'd rather have internal connectors so that it'd be neater when building into tiny systems.

Re:Great, but will it support Virtualization?

[petermgreen]

it seems mini-itx.com are stocking a pin headers only version of the epia N. i haven't seen any in the mini-itx size though.

btw if your building into a system you don't have to use those bulky hoods on your D connectors. Just bend the wires in the direction you wan't and tape em up to add a bit of regidity.

What does hot product really mean?

[Anonymous Coward]

"It even overheated when it wasn't in the box."

Wow! Did it burst into flames on the store shelf?

Cool but...

[Life700MB]

Looks like a very cool home server, but it lacks a second network card, like the MacMini.


--
Superb hosting 20GB Storage, 1_TB_ bandwidth, ssh, $7.95

Re:Cool but...

[Ulrich Hobelmann]

Well, if you don't need a PCI card, you can probably use the PCI slot (I hope they have one!) for another networking card (unlike the Mac mini I'm sitting at).

I ran an Eden 533 for a while, and did just that...

Re:Cool but...

[marcelC]

The previous C3 "nehemiah" line of epia boards had a model with 2 ethernet ports on it. These should come out for this model as well.

Re:Cool but...

[Anonymous Coward]

Yeah but these things are perfect for car pc's like those over at http://www.dashboardmonkey.com/ and http://www.carpcspecialist.com/

There are entire communities that use things like this.

In fact, we got sent some test units of these babys a month or two back by via and are about to get some new power supplies too. We did our tests which will be published shortly, but these puppys rock!

Re:Cool but...

[Hal_Porter]

You can connect several machines to one network port, you know.

Re:Cool but...

[petermgreen]

via do the epia CL and PD lines which have two ethernet adaptors (and a pci slot too). Its quite possible they will do something similar for this series.

and all epias have a pci slot (two with a riser card) which you can use to get more ports though you have to choose your case correctly.

Re:Cool but...

[TeknoHog]

Besides choosing another VIA EPIA model that does have two network interfaces, or using the PCI slot for a NIC, there's a somewhat neat software solution. You can actually NAT with a single NIC if you assign two IP addressess, and use a switch. It's probably not as secure as proper NAT, because you cannot separate the networks on an Ethernet level. On the other hand, local addresses like 192.168.0.0/16 are not routable, and thus not practically visible to the outside Internet.

Re:Cool but...

[Shanep]

You can connect several machines to one network port, you know.

People often like to have small machines with at least two NIC's for firewalls. External interface and Internal interface as a bare minimum. I have 5 NIC's in my Sun Ultra 10 firewall, all being used for seperate segments.

Re:Cool but...

[Jeff DeMaagd]

For server alone, a second card isn't so necessary, the mini even has gigabit. If you are using the computer as a firewall, then that is a different story, I thought miniITX boards were available with PCI slots and even had riser cards to accept as many as three cards. I think it would be better to use some old decommissioned computer than buy new computer parts for a firewall and low intensity server.

Re:Cool but...

[level_headed_midwest]

Well, it all depends on what kind of server. Consumer-grade routers can be had for $50 and their ~200MHz CPUs handle routing a WAN connection to 4 100Mbit Ethernet NICs and generally wireless too with little problem. Gigabit units are about $150-200 and generally have 8 or more GbE LAN NICs and their hardware isn't that much more powerful than the 100Mbit routers' are. I'd get a router or switch instead of a Epia unit if all you're going to do is use it as a router. Or use that old decomissioned computer to do your routing.

Now if it is a file server, that makes a little more sense, especially if you need it to be in effect a slightly-more-powerful NAS box. Old computers generally balk at using large HDDs, and the Epia can handle SATA drives.

Re:Cool but...

[Anonymous Coward]

The visibility of the 192.168 addresses would depend on how the ISP connection is made. If the ISP doesn't block broadcast between different customers, it could be visible to outsiders.

PVR

[LordSnooty]

And of course they target living-room PVR devices, but with the brouhaha over broadcast flags, maybe it's understandable that they want to keep it quiet. Do it the easy way with Knoppmyth

Re:PVR

[HaeMaker]

Cool. First thing I thought of too!

Re:PVR

[idonthack]

My dad uses some mobos like this for our Myth setup, with 1GHz processors and hardware mpeg decoding. They've got fans but they run really quiet, especially without harddrives. He had another fanless 600MHz one, but it was too slow to use as a Myth client. These faster ones will be great for the same purpose.

Re:PVR

[BooRadley]

This is for your dad:

Here's to you, Mr. "I have a kid old enough to post intelligently on Slashdot, yet I'm still geeky enough to build out my own multi-head MythTV setup."

Re:PVR

[geckofiend]

Except that the open source drivers removed support for MPEG decoding acceleration and Via's drivers suck.

Re:PVR

[wings]

Thanks. :-)

Re:PVR

[Mr. Feely]

This week's Linux Weekly News had an article on MythTV that also covered Knoppmyth. Apparently it's not as easy as it seems (or at least the LWN editor didn't find it so).

Re:PVR

[dan the person]

The openchrome drivers support XvMC for hardware MPEG decoding.

It's the MPEG4 part that doesn't have opensource drivers, but my SP8000E plays divx on the plain old CPU with 60-70% idle. (mplayer and Xv output)

They target thin clients ... hmmm.

[Anonymous Coward]

The board looks like it has more horsepower than the computer I'm using right now. Have they re-defined what a thin client is?

Re:They target thin clients ... hmmm.

[hey!]

I dunno. What exactly do you call a device that has no hard disk but boots over the net using a boot prom and mounts its storage using network file protocols like NFS? By some definitions this is not a "thin" client I suppose, but I think this is the kind of thing where a physically tiny board packing a lot of power is a good idea.

Re:They target thin clients ... hmmm.

[imsabbel]

My cell phone has more horsepower and RAM than a big iron had in the 70s...
Redefinitions happen all the time in IT, you know?

Re:They target thin clients ... hmmm.

[pimpimpim]

My home pc uses a "1 Ghz" Cyrix C3. At work we have igel fanless thin clients with exactly the same processor. After I found that out I figured I could get rid of my cpu fan, and replaced it with a zalman northbridge heatsink. Two results: 1) I cannot play dvd's anymore (but I have a workaround that works when I plug in an external DVD-player into the TV-card running Dscaler at low resolution) 2) I found that my harddrive is actually the loudest part of my pc..

Anyway, these thin client processors can run WinXP smoothlessly, but have a bit of trouble when it comes to high loads, especially for video processes. Then again, I'm using an exotic onboard trident cyberblade, I guess most of my problems would be solved when plugging in a PCI video card.

So you don't want to compile gentoo on these boxes, but I really think the current thinclient cpus would do really well as standalone pcs as well for most practical (office) purposes.

Re:They target thin clients ... hmmm.

[anagama]

these thin client processors can run WinXP smoothlessly

So that would mean "roughly" (as in "not smooth" as opposed to "approximate"). Nice word though. I'll have to use that one for its confusion value.

Re:They target thin clients ... hmmm.

[Siffy]

http://en.wikipedia.org/wiki/Thin_client
Hmm, nope. No mention of mb, ram, or mhz anywhere in that definition/description.

Re:They target thin clients ... hmmm.

[pimpimpim]

Ah yes, after submitting I figured something went terribly wrong with that word. Well then, they run winXP roughlessly!

But thank you for pointing me on this, it's realy a word with potential, might come in handy when writing reports!

Re:They target thin clients ... hmmm.

[Bing Tsher E]

Your cellphone's cpu has more raw processing power, but has pitiful I/O compared to the 'big iron in the 70's' system, which performed useful tasks for thousands. So consider it like comparing a bonfire (your cellphone) to a Ferarri engine (the 'big iron'.) The bonfire might be consuming and discharging far more energy, but almost none of that power is well-coupled to a mechanism. It's wasted cycles, for the most part.

Re:They target thin clients ... hmmm.

[imsabbel]

You seriously overestimate the IO of the 70s big iron...
They might have had supports for lots of connections... of 1200 bps terminals.
Just compare that io to several Mbyte/s via usb.
Bluetooths alone could be used to tunnel 100 terminal connections....

A prototype Epia EN board (Click to enlarge)

[urban_gorilla]

no thanks. thats the whole point.

Hardware Specs

[slick_rick]

I would have bought an EPIA years ago if VIA would open up the specs a bit more. Google around about people getting linux going on these. It is an easy thing to do, as long as you don't want to have everything on the board work (like the SVideo out, the onboard MPEG2 decoder, etc) . It can work from what I understand, just not something I wanted to spend a week trying.

So is the new line any better? If so I'd buy as I'm in the market for 2 or 3 machines like this. Question is, does VIA even care about the Linux user? Until now the answer has been no.

Re:Hardware Specs

[gid13]

Yeah. It'd also be nice if you could find more than one mini-itx case on the planet that didn't look like pure ass.

Re:Hardware Specs

[Anonymous Coward]

Considering there are linux drivers available for the currently-available boards, and that VIA link to those drivers from their homesite, I'd say, "Yes".

Re:Hardware Specs

[arivanov]

Bollocks.

I have Debian running on 30+ of them. All varieties from V onwards. In fact it has been the primary small server platform for all of my projects for 2+ years now.

Ubuntu and Knoppix also run fine.

I have heard about some problems with RHEL on the lowe end C3 which is not surprising because AFAIK RHEL nowdays by default comes with a 686 kernel which requires SSE. All you need to do is force it to use a 586 or lower kernel (if it ships with one). It should be OK with C7 and all higher end C3s (nehemia core).

Which reminds me, I have been postponing writing an article on how to run Linux on these for a while now. Need to get to work on it...

How about homemade routers?

[PingXao]

Now that Linksys is abandoning Linux completely there aren't many SOHO routers for hackers out there anymore. My electric bill here in the northeast USA is running almost $150 a month this winter - and it's been warm. Can't wait for air-conditioner season. But I digress... Passive cooling = no fan, right? If these things are quiet AND efficient they might really have something there, although 2 GHz for a router seems excessive. Maybe not once you add in network storage, a web server, a Radius server and a few CPU-hungry VPN protocols.

Re:How about homemade routers?

[michrech]

For SOHO, do you *really* need 2ghz for web server, router, storage, etc?

1.2ghz ought to be fine for that, which is what they claim it will run while being passively cooled.

Minus the vpn tunnels, I ran a web/mail/storage/router/firewall server, under linux, on a P3550mhz. I'm sure the VPN tunnels wouldn't require THAT much CPU time...

Re:How about homemade routers?

[Not_Wiggins]

If these things are quiet AND efficient they might really have something there, although 2 GHz for a router seems excessive.

Exactly correct: 2G is excessive for a router. However, if it can be passively cooled at 2G (well, the article states passively cooled at 1.2G), then it should be very cool when it is underclocked to run at a speed adequate to handle router-specific tasks. Just because the power is there doesn't mean you need to utilize it all; underclocking* is great in situations where cooling is an issue.

* - When available. Depends on if there are frequency locks on the MB and/or processor.

Re:How about homemade routers?

[ivan256]

There are *plenty* of "SOHO" routers for hackers. Linksys were always jut the "crappy but well marketed" ones. Lots of Netgear and ZyXel routers are plenty hackable, run linux, use the same processors as the linksys routers, have more features, and are signifigantly cheaper.

Release yourself from the grip of the Linksys fanboys.

Re:How about homemade routers?

[questionlp]

It may not be a retail product, but you can always get a Soekris kit that has multiple Ethernet interfaces, 128 or 256MB of RAM and supports CF for additional storage for around $300 (net4801-50, 128MB version, with a total of 5 Ethernet interfaces).

http://www.soekris.com/net4801.htm

It runs off of a 586-class processor and with all of the fixins, would only draw around 20-25W. Not bad for something that can run Linux or *BSD. I haven't messed with one yet, but they do look pretty good even for a small server that can provide: SSH, FTP, web, NTP, DNS, DHCP, etc. Heck, it may replace my Sun Blade 100 one of these days ;)

Re:How about homemade routers?

[TopSpin]

that has multiple Ethernet interfaces

I'd have bought several Epia boards by now if they had just put a useful number of Ethernet ports on board. Their 2-port boards look like one-offs and do not inspire. I want a board with 3 ports. If you're going to talk about "server appliance" you need 2+ network ports... Yet, from the photos I see here their latest stuff has

wait for it

1 port.

Sigh.

Re:How about homemade routers?

[Grelli]

Now that Linksys is abandoning Linux completely

Except for the part where they listened to the community and introduced a new model based on the V4 hardware of the WRT54G called the WRT54GL...

Re:How about homemade routers?

[tomstdenis]

You don't need a 2Ghz part for a router. My C3 1.2Ghz part runs Gentoo Linux [yes, I build stuff on the thing] and it handles traffic just fine. It can ipchains NAT traffic at full modem speed [which admitedly at 700Kbyte/sec isn't that fast].

The newer 2Ghz parts may give you more oomph but for a router/nat/dhcp/bind box they're not required.

Tom

Re:How about homemade routers?

[Trigun]

I would take as much processor as they are willing to give me, for anything. The more you have, the more you can do.

The flip side to this is the general-purpose CPU. Do we really need this for routers, firewalls, and everything else that we can think of? Maybe it's time to ask for specialized chips. Instead of having one 3 ghz chip running at 1% cpu usage, have a stripped down one running 50%.

Re:How about homemade routers?

[Anonymous Coward]

http://www.teamasa.com/home.shtml

This company sells XScale boards. I didn't see any with more than two Ethernet ports, although I saw one with Quad SATA. They're expensive, though.

Re:How about homemade routers?

[Breakfast Pants]

Linksys bailed on linux in the standard WRT54G. They took it out and used all the good press the thing had to start selling a less capable machine as the same product. But now they make the WRT54GL, which is the exact same as the old WRT54G.

Re:How about homemade routers?

[TheRaven64]

I'd also recommend the PC Engines WRAP line. I have one in a 6"x6"x1" box, and it boots happily from a compact flash card. It has two mini-PCI slots, one of which has an 802.11a/b/g card (antenna mounted on the case, four antennae supported) and the other will gain a crypto accelerator at some point.

Re:How about homemade routers?

[PingXao]

They announced it a few months ago. They killed it off already. They're not making any more of them so when the supply chain dries up it's "Adios!"

Not too exciting

[jandrese]

From the looks of it, they're just releasing a chip that is 40% less of a dog than the existing chip. Still not that great. It is a bit more power efficent than a Pentium-M, but you really pay for that in the performance. Not a terribly exciting chip IMHO, but one that will probably find some use in set-top appliances and the like.

Re:Cool but... COST

[Anonymous Coward]

The EPIA boards look nice until you actually try to spec out a system for your entertainment center. The CPU board + case + hard disk gets you into the $400+ range and makes the EPIA based system worse than buying an integrated, tested system such as the Mac Mini.

Via should concentrate on producing a $50 board with a $40 case and power supply. That would make a nice sub-$200 system for the living room

No drivers for CarPC stuff

[scsirob]

VIA so far has ignored all begging owners of other MiniITX boards to release Windows drivers that can run 800x480 resolution. This is the native resolution of nearly all 7" wide-screen displays, very popular with Car PC builders.

I sincerly hope VIA will listen this time and release a driver that fits the requests of all these CarPC project owners.

Also, there's been a MiniITX board with 12V-only power input. Unfortunately the 12V must be within +/- 5%, making it again unsuitable for Car PC usage. Why can't they release a board with wide voltage input (7V - 28V), and if at all possible with a built-in shutdown controller??

Re:No drivers for CarPC stuff

[pdbogen]

Why can't they release a board with wide voltage input (7V - 28V)

Probably because this would significantly increase the price of the parts, (being that they would need to be of much higher quality), and/or require them to incorporate a voltage-regulating power supply, which would (a) increase the size, and (b) increase the heat output. You can probably do the second part on your own, anyway...

Re:No drivers for CarPC stuff

[hackstraw]

I know I'll get moderated as a troll again because I suggest using Linux over Windows, but

Modeline "800x480" 40 800 864 928 1088 480 481 484 509 +hsync

Will do the trick for X. No drivers necessary.

Re:No drivers for CarPC stuff

[Anonymous Coward]

Wouldn't Powerstrip work?

Re:No drivers for CarPC stuff

[Anonymous Coward]

Have you tested this with a M1x000 board? The resolution is locked out by the BIOS. I've tried several hacks with no success. Also, when using a Lilliput 7", the screen itself can't lock onto a proper resolution.

Re:No drivers for CarPC stuff

[essdodson]

Besides being a troll, It's also my understanding that many of the apps car pc users enjoy aren't available in Linux nor are similar replacements. Specificly in the area of map applications that tie into GPS.

Re:No drivers for CarPC stuff

[Anonymous Coward]

You can also edit the driver INF file and the monitor INF file in windows in a similar way, but I'm too lazy to point out how to do it. :)

Re:No drivers for CarPC stuff

[eclectro]

I suggest using Linux over Windows...Will do the trick for X. No drivers necessary.

What's more, if he's hard up to run that Windows app, he should look into codeweaver's crossover office.

Re:No drivers for CarPC stuff

[Wrath0fb0b]

Powerstrip: http://www.entechtaiwan.com/ps.htm

duh . . .

Re:No drivers for CarPC stuff

[orv]

Duh. Won't work because the windows driver simply uses bios modelines to set resolutions.
With the windows driver you're limited to the resolutions added to the BIOS by VIA.
On the other hand if you're using the Linux drivers you can set any arbitary video mode due to the work of Luc Verhaegen.

Re:No drivers for CarPC stuff

[hackstraw]

Good luck in getting your display _and_ the GPS working. Seriously, good luck.

I'm also sorry

[Weaselmancer]

You seem to have mistaken VMware as a server only application.

Multicore C7 = x86 Niagara?

[Hal_Porter]

Via have been making small, cheap, low power cores for some time, whilse Intel and AMD moved to large, expensive high power one.

Now there's a move to multi core designs and blade servers, and even the slowest x86 server is probably over powered for a server, you have to wonder if they could do an x86 version of Niagara

From here

http://en.wikipedia.org/wiki/VIA_C7

"You can also see a quad-core C7, could be manufactured for the same cost as a single core P4 on 90 nm process."

Now Niagara is 8 core and each core has four threads admittedly, but there's something to be said for a four way x86 chip for blades. The power consumption wouldn't be too bad either. But you can have four C7 cores per P4 core. If I were AMD for example, I'd be playing around with an x86 Niagara.

http://groups.google.com/group/comp.arch/msg/991ff 1390b277b98?hl=en&

Hmm, and I'd find (or invent) some new benchmarks too.

Re:Multicore C7 = x86 Niagara?

[BigFootApe]

If I were AMD for example, I'd be playing around with an x86 Niagara.

I think too many existing programs in the x86 world are still course grain multithreaded (at best) for AMD or Intel to make a go of 8 core x 4 smt for an architecture. Certainly for desktop processors, multicore is still of limited utility. Look at the number of games that are not multithreaded. The speed benefits on these games when running dual core are, as yet, limited.

Even in the web area, there's at least as much of a push for blade servers as for multicore. Building a failover/load balancing scheme is at least as easy as building a fine-grain multithreaded application.

A setup like Niagara works for Sun (and IBM, with Power 5) because these customers are already accustomed to working with E10ks and large pSeries servers. The software is designed to scale to these architectures already.

Re:Multicore C7 = x86 Niagara?

[mrchaotica]

Speaking of processors, whatever happened to the "C4?" Did they just skip that number? I certainly hope not, since I was looking forward to getting a laptop with a "C4 inside" sticker (fun for airports!).

Re:Multicore C7 = x86 Niagara?

[Anonymous Coward]

whatever happened to the "C4?" Did they just skip that number?

Unfortunately yes, for exactly this sounds-like-an-explosive reason. They didn't realise at first, a "C4" was definetely on the roadmap for a while.

Re:Multicore C7 = x86 Niagara?

[Hal_Porter]

Via wanted to call a chipset the KZ, but KZ stands for "Koncentrationslager"

http://www.pcguide.com/art/rwtiu0006Chipset-c.html

So they renamed it to the KT. Luckily for them Tyrannosaur Americans haven't survived in sufficient numbers to form an effective lobby group, since they would no doubt find the acronym KT very insensitive.

Re:Multicore C7 = x86 Niagara?

[Hal_Porter]

I was thinking about using for web and database servers. There'd definitely be some scaleability issues with normal PC OSs and applications, but that's the point.

There's quite a bit of research on impoving scaleability with lots of CPU's. My idea is that you add some instructions to help e.g. spinlocks, and build them into protoypes with one OS, one webserver and one database server. The vendors will be keen on this stuff. And the net result is that you can patent the instructions, and try to license them to other CPU vendors. Then they could license them and get the speed up in the new applications when they are released, or they could wait until after and make their own version.

I'd design the new instructions by looking at cases like this

http://congreso.hispalinux.es/congreso2002/ponenci as/blanchard/talk_2.pdf

With a bit of luck, you could come up with some changes in the cache and instruction set to impove 8+ way SMP. In many ways, it's a bit of a Risc philosophy, since you don't try to build particularly complex processors. But you do spend a lot of time profiling real code running in simulations, and then build the next generation with new features, only if you can implement them efficiently, and only if they make a radical difference to the speed of the applications you are testing with.

But my guess would be that you'd some up with a few new instructions that would make 32 way SMP much more efficient. Then you cram them into your 486 class cores and put a load of them on a chip. And at this point you're software partners have something which the can release to run well on the new chip. And suddenly you have a PC compatible chip & applications to compete with big iron stuff like 32 way PPC.

It's a world away from hiring a load of microarchitecture geniuses to work in isolation on really good single thread performance. But I think if you're AMD, there's much to be said for opening up a new market, rather than relying on being able to beat Intel at their own game each chip generation.

Personally

[hey!]

I'm much more interested in what happens when you hold the processing power constant and drop the price, as the price performance ratio drops.

I understand that it's not attractive for a company to look at lower margin items, but imagine if you could retail something like the original mini ITX boards in the price range of, say, $50 (it's currently about $110). Every garage inventor in the country would be creating new embedded computing applications.

Re:Personally

[Anonymous Coward]

You are absolutely right, for right price I would be eating these things for breakfast.

I love the silence

[Anonymous Coward]

Got 3 of these now, a Nehemia and 2 Edens (566 and 800MHz fanless). I use them for making music and editing sound.
Even though they are not amazingly powerful I would never go back to some huge whirring pizza oven. No hard drives either, all boot from 4G IDE flash drives, one with a modified Dynebolic/Puppy linux crossbreed and the other I usually boot DSL (Damn small = fits on a 512M USB thumbdrive) Not a single moving part in the room! (unless you include me, and I don;t move much) All the sound files are on my file server in another part of the building.

Re:I love the silence

[rebeka thomas]

No hard drives either, all boot from 4G IDE flash drives

Remind us how stupid an idea this is when in two months you find your flash drive failing because you've used it too much just booting.

Re:I love the silence

[Anonymous Coward]

Funny, my powerbook has been booting from one for several years now. Still seems to be running.

Guess I must be imagining things. That, or you're misinformed.

Re:I love the silence

[toddestan]

Remind us how stupid an idea this is when in two months you find your flash drive failing because you've used it too much just booting.

Why? Booting usually doesn't involve very much writing, just reading (Windows is probably an exception here). Throw in enough ram so you don't have swap much, and it would probably last forever - or atleast longer than the mechanical parts in a harddisk would.

Not just the boards...

[WoTG]

Even if the boards are kept quite cheap, you have the problem of finding (cheap) appropriately sized cases. There just isn't the volume for these things, so the cost of a case is about 2x the cost of a generic mini-atx PC case. Of course you could build your own case... however, that requires a whole mess of tools and costs and time too.

Re:Not just the boards...

[Anonymous Coward]

Who needs those fancy cases anyway, these would be great for controlling those laser beams attached to my sharks.
---
Yours,
Dr. Evil

Re:Not just the boards...

[kfg]

. . .you could build your own case... however, that requires a whole mess of tools and costs and time too.

And who really wants to be an inventor if it means having to deal with a whole mess of tools and building shit, like . . .a box.

Maybe they just don't make inventors and engineers like they used to, but we used to be able to handle making a tin box pretty good.

KFG

Re:Not just the boards...

[Anonymous Coward]

What about Shuttle cases? Plenty of those around and easy to convert. I would do it :p

Re:Not just the boards...

[Breakfast Pants]

I don't think the price of the motherboards was keeping inventors from toying with embedded devices based on them. So I think we can assume he was talking about putting a hacked together product into moderate production. Sure, someone can build a tin box by hand, but is the time worth it when you are making a few hundred of the things yourself?

Re:Not just the boards...

[kfg]

Sure, someone can build a tin box by hand, but is the time worth it when you are making a few hundred of the things yourself?

Of course not. That's why God invented the local tinwork shop. You only hand build oneoffs and prototypes. The rest you farm out.

KFG

ah, more via pain

[Blymie]

I have had, and know so many Linux users that have had, problem with VIA chipsets. DMA issues, issues with lockups, VIA unwilling to communicate with Linux developers on resolving them.

Most recently an Asus board I owned last year, locked up as solid as a monkey if any heavy DMA activity occured. Worse, after doing hours of Google searches, I managed to find info stating that Windows drivers disabled various chip functions, so that the chipset could run in a stable function.

Apparently, from the slant of posts that I read, it was taken as fact that VIA often had issues with chipsets, and merely patched those issues with drivers. Typically, one buying a VIA board in Windows would end up with degradation of their chipset via drivers. Linux users were, however, not so lucky. VIA would ignore all pleas and requests about issues with their chipset, and the belief was that they did not want such issues with their chips to "make it to the press". Acknowledging that they had reduced chipset performance with drivers, would obviously not go over well. Chipsets are marketed to certain specs, and using drivers to "make it work", but not deliver those specs is clearly opening liability.

After reading this, I looked at issues I'd had over the years with graphic cards causing hardware lockups, boxes that would randomly reboot and the like. In almost all cases it tended to be with system that contained VIA chipsets. Further, I also found posts from many Myth users, complaining about DMA issues with their mini-itx boards.

VIA? I'd recommened everyone stay away.... I sure the heck do! Time isn't worth the $20 you save by walking away from an Intel or SiS chipset. Sure, these chipsets have issues, but Intel and SiS both seem a little more talkative with Linux developers.. and tend to produce a better product. VIA seems produce these flaws in almost _all_ of their chipsets.

My experience, sure. You'll have to make up your own mind. All I know is that $20 in savings is peanuts over 20 hours of debugging.. when the debugging is a useless task.

Re:ah, more via pain

[SRCShelton]

It's interesting that people report all of these problems with VIA chipsets, but that those same chips (or, at least, the southbridge) work just fine when used in desktop motherboards.

I suspect that the fact that VIA Eden processors (I'm not sure about the Nemehiahs) are not actually i686-compatible, lacking a CMOV (conditional move) instruction.

People then install super-whizzy latest Ubuntu (say) with uber-optimisations, and wonder why certian things break.

This may not always be the case, admittedly - but if you've got a chipset which just works for desktops but doesn't work when coupled to a processor with special requirements then I'd be tempted to suggest that it's software not meeting those requirements which causes problems, not the hardware directly.

Re:ah, more via pain

[synonymous]

I for one have actually found my MII1200 under Debian fairly solid.

Heres the thread dated from Fri Sep 17, 2004 19:14.

Pain/Suffering

Re:ah, more via pain

[Blymie]

I'm not sure what you're getting at, but my personal experiences were all normal motherboards, with VIA chipsets (not cpus) on board. DMA issues effect motherboards and VIA chipsets, desktop or otherwise.

My recommendation, stay away from VIA until they clean up their act a bit.

Re:ah, more via pain

[vondo]

Really? My home PC and about 4 (of 8 or so) machines at work have died over the last 18 months. Every single one of them was a VIA KT333/KT400 based motherboard (and the MB was the likely cause). We still have plenty of Intel 440BX/PII/PIII systems running, but these VIA/AMD systems are dropping like flies.

Sure makes me wary (I've been buying i875 and NVidia NForce lately).

Re:ah, more via pain

[dfries]

The diagram says 2D/3D for video. Does that include 3D acceleration under Linux? I know the SiS board I have with 315PRO doesn't have support for the 3D functions.

Is anyone else making a Mini-ITX board that has supported 3D under Linux?

Re:ah, more via pain

[proxima]

VIA? I'd recommened everyone stay away.... I sure the heck do! Time isn't worth the $20 you save by walking away from an Intel or SiS chipset. Sure, these chipsets have issues, but Intel and SiS both seem a little more talkative with Linux developers.. and tend to produce a better product. VIA seems produce these flaws in almost _all_ of their chipsets.

My experience, sure. You'll have to make up your own mind.

Not that anecdotes are all that meaningful to others, but I've had relatively good luck with VIA chipsets. Since I've had AMD processors in my last three motherboards, Intel was not an option. I haven't heard good things about SiS in general, but that was a while ago. NVIDIA is the newcomer, but to get some mb features working you need proprietary drivers (annoying enough for things like video, but I'm not interested in it for good chipset functionality).

Of four motherboards (an Epox and Abit built with KT133 chipset, and an Asus with a KT400 chipset, and a brand I don't remember with an old K6-2 450 Mhz cpu), two are still running (the Epox and the Asus). The other two experienced slow deaths (worked great for years, slowly became unstable), which I largely attribute to some improper venting of the towers causing heat damage.

In my experience, as a Linux user, it's best to seek out a chipset that is reported to work well with Linux. Then find a motherboard brand you trust. At this point all I'll go with is Asus (my experience with motherboards goes beyond my personal computers, and while I used to think Abit and Asus were equal in quality, I don't think Abit is as good anymore). I had to RMA two motherboards (with KT133 chipsets), before getting the Epox. I can't blame the chipset, though (unless there are serious quality differences in the way the chips themselves are manufactured, but I doubt it), since I've used KT133 with great luck for years since then. So while my percentages might not look great, the lesson here I think is that the motherboard manufacturer plays a more important role.

VIA, Intel, NVIDIA, SiS, whatever, you can probably find a particular chipset from each company that is reported to be stable with your operating system of choice. Once you've chosen that, don't skimp on the cost of the motherboard (and the cooling), and your experience should be a good one.

Re:ah, more via pain

[ozbird]

No problems here with an Asus A8V Deluxe (Via K8T800 Pro chipset) or Epox 8K3A+ (KT333+8233A chipset). In contrast, some friends have had problems with various nForce chipsets.

Your motherboard may vary.

Re:ah, more via pain

[Malor]

The KT133 was terrible. You must not work your computers very hard. The KT133 was a disaster in all respects, and it gave the early Athlons an undeserved bad name. The CPUs were excellent, but the boards were complete shit, no matter who made them. A quick search on Google for KT133 problems shows NINETY THOUSAND hits.

The KT266 and 333 were pretty good as Windows desktops. But I'm using one as a Linux server now, and it's... not great. I have to run it with APIC disabled. Turning on APIC (in recent 2.6 kernels) causes all kinds of problems with software RAID. And I don't get very good throughput; using an Intel gigabit network adapter, I'm lucky to push 150 megabits onto a (very fast) SCSI array. And that's with no other load _at all_. Even pulling or pushing to a tmpfs, the absolute best I can manage is about 200Mb. In theory, I ought to be able to get north of 300 on vanilla PCI, but the chipset doesn't seem to have very good throughput. (this is without jumbo frames, but with a lot of other optimizations on the Linux side; I get similar results from both Windows and Mac gigabit clients.)

NVidia chipsets are pretty good, with the exception of their IDE and Ethernet drivers. The most recent Ethernet drivers instantly bluescreen my DFI Lanparty NF3 250 machine. The older version works, but you have to disable all TCP offloading and firewalling for full stability. WoW in particular does _not_ like offloading. And the IDE driver is slow and buggy... it's best to stick with standard Windows drivers.

That sounds pretty damning, but it's really quite good, other than that. But keep in mind I've only run this board with Windows. The KT333 I mention above worked fine in Windows too, but isn't so hot in Linux.

Intel may be behind in the CPU department, but their chipsets remain the best...extremely solid. In the high end, the NForce4 Professional on Opterons is supposedly good... but for low- to mid-range servers, where I won't be around to check on a system regularly, I'd much rather do Intel.

Then again, I have a personal Intel server with an 865 chipset that's hosted somewhere in Texas. It kernel panics randomly on every version of 2.6.15 I've tried, but is absolutely solid on 2.6.14 and earlier.

Even Intel hardware is no panacea against the crappy 2.6 kernel development process.

Re:ah, more via pain

[proxima]

The KT133 was terrible. [...] A quick search on Google for KT133 problems shows NINETY THOUSAND hits.

The KT133 was immensely popular, as I recall. It's no surprise that lots of people report problems with it. On the other hand, I didn't say that the KT133 was particularly great, just that it was stable for me (while using it as an example of how the quality of the motherboard manufacturer seemed to matter a great deal). I leave beating the heck out of chipsets for review websites, which then form a decent basis of my purchases (at least since the KT400). Performance? It wasn't high on my priority list.

I don't recall noticing huge performance differences in benchmarks of articles I read, but I could certainly be mistaken. I would still take a decent performance hit if it meant greater stability (which is why I stay far away from overclocking). I tend to buy the low end (but good quality) of what's available at that time, and upgrade more often (I think it provides more consistent relative performance at a good value).

Overall, I'd jump at an Intel chipset, since Linux support is probably most consistently good there. But when the AMD options are just so much more appealing, you're left with VIA, SiS, and NVIDIA (roughly speaking), and I still have no compelling reason to move off of VIA.

Re:ah, more via pain

[ross.w]

Have to agree there. I have a similar Asus motherboard that's been running Suse Linux for about a year now with no problems.

Re:ah, more via pain

[mycall]

very true, Asus K8V Deluxe here -- super happy even though it is over 2 years old now. Zero problems with it and I bully it around.

Re:ah, more via pain

[Wiz]

The KT133 wasn't long for this world, it was replaced by the KT133a which was fine. But yeah, KT133 was a bit crap.

Which Nvidia ethernet driver are you using? nvnet was reverse engineered and was put in the kernel are forcedeth, which I believe Nvidia now maintain. I never had any problems with it. To be honest, I never touched their IDE drivers as the kernel ones seemed good enough.

It could be worse anyway, you could use ATi's Linux drivers. ;)

Re:ah, more via pain

[Malor]

My NVidia machine is my Windows gaming box, so I'm using NVidia's standard Forceware drivers. Instant bluescreen on the Ethernet driver.... intensely frustrating. The KT333 is my Linux server, and I'm annoyed with _that_ because the box is unstable in APIC mode, and I don't get very good throughput on gigabit. But those are separate issues. :)

Re:Cool but... COST

[Anonymous Coward]

Whereas, of course, any other motherboard plus processor plus power supply plus disk plus case will come in for only a tiny fraction of that.

DVI

[yet another coward]

I am disappointed to see the persistence of VGA rather than DVI. According to the page, an LVDS/DVI module will be available. All the tiny LCDs should move to the digital world, too. It would make them a little smaller and cooler.

IAWTP!

[ankarbass]

No shit.

When's the actual release?

[palfrey]

Given VIA's history of "rapid with press releases, incredibly slow with actual boards" (NanoITX anyone?), any bets on how long before we see available boards? I'm betting mid-2007 personally...

Re:When's the actual release?

[Obfuscant]

Yeah, I got stomped when I delayed a project because the NanoITX was "real soon now" ... VIA is vaporware until it's actually in my hands...

Re:When's the actual release?

[Anonymous Coward]

"NanoITX anyone?"

Yeah, those were a complete joke. I really wanted one for.... wow is it really 2 years!! since they first showed a photo of one? Talk about vapourware. And the most stupid thing of all is that now some stores claim to have them "in stock soon" they are almost twice the price of the Minis. Worst marketing VIA ever did.

I still say the Mini ITX rock though.

Re:When's the actual release?

[Anonymous Coward]

Hopefuly, we will all play the brand new Duke Nukem Forever on our shiny NanoITX boxen very soon!

Why IDE?

[kahanamoku]

With the high availability of SATA drives (HDD & CD/DVD) these days I'm wondering why they still include IDE Connectors on these boards. Surely if they want to get the size down even further they could remove the IDE ports and use only SATA.

Re:Why IDE?

[corrosive_nf]

considering a search on newegg only brings up one match for a SATA optical drive, and its the infamous plextor thats really pata with a bridge screwed on the back, I will say they left them for optical drives.

Re:Why IDE?

[pe1chl]

And don't forget IDE Flash drives!

Re:Why IDE?

[Bertie]

Have you tried installing Windows XP on a computer with an SATA hard drive? Oh man. Pain. You actually have to kick the thing into life using drivers loaded off a FLOPPY DISK. Or at least, you did the last time I tried it. I bought all these shiny new components and had to borrow a floppy drive to get it running! In the end, I binned it and went Gentoo instead.

That would explain to me why these machines don't go SATA yet - most people are building their own systems with them, I'd imagine.

Re:Why IDE?

[corrosive_nf]

Not if you slipstream the sata drivers onto a windows cd.

Re:Why IDE?

[Anonymous Coward]

ever try to install Linux on an all sata system, I just did and it took adding an IDE optical drive to get it to take anything... five linux distros later I finally got a functional OS. I hate to interupt your trashing of Microsoft with some facts but linux has sata issues too.

Re:Why IDE?

[kahanamoku]

MSI Have SATA CD/DVD ROMS

There's also the USB / Firewire options for initial setup for those who dont need optical drives in their mini PC's

Re:Why IDE?

[Anonymous Coward]

If SATA is supported natively in the BIOS, then I don't think you need a floppy drive to get the drivers setup.

Atleast thats how it was in the two systems with native SATA support that I installed XP on.

Re:Why IDE?

[bjoeg]

Slipstream your SATA driver into a new installation CD!

Yes is irritating to use floppy if you haven't slipstreamed, but not MS fault new hardware lands on market after their OS launch.

Re:Why IDE?

[fufinache]

I used to work at a small computer store doing small hardware repairs and I've ran into the same problems, this is what I found out about the issue after doing a bit of research: When SATA was still new, so motherboard manufacteurer patched on SATA controllers, which were not native, so they required seperate drivers to work. Of course, the irony of it is that the driver is usually stored on the hard drive, so in order for consumers to get SATA to work, you run a messy patch to enable OS installers to recognize the hard drive. Newer motherboards have native SATA support in the chipset, which means that you don't need an external driver.

Re:Why IDE?

[milimetric]

you're saying that you had a hard time installing windows so you went with... gentoo? Buddy, I've installed gentoo, and I've installed windows. I think it'd be easier to install windows if you had to do it while balancing two buckets of water and walking up a hill kung-fu style. I LIKE gentto better, but to say it's easier to install...

Re:Why IDE?

[Bertie]

Well, there were other reasons besides ease of installation, but Gentoo did SATA from the word go, which other Linux distributions generally didn't at the time, so I was sold.

too slow to boot

[tota]

I've toyed with these boards on more than one occasion for fun and profit, and the thing that really prevents you from using them in the embedded space is the amount of time it takes to boot the bloody thing.
Around 30s at the best of times to get to a shell with init=/bin/bash and only a little less if you use the linux bios. Disappointing to say the least, no decent set-top box can take more than 5, maybe 10 seconds to start.
Even using suspend to boot directly into a running system is not going to help since most of this time is going to be spent in the bios.

Not to mention that some boards come with a compact flash, but you can't boot from it! What's the f... point?

Re:too slow to boot

[MBCook]

Disappointing to say the least, no decent set-top box can take more than 5, maybe 10 seconds to start.

Do you have a TiVo? Do you have any idea how long it takes to turn on a TiVo? Mine (a series 2 DirecTiVo) takes over one minute to boot. My brother's (a series 2 stand alone) takes about the same time.

For something that will stay plugged in for long periods of time, boot time doesn't matter. Turn the thing on, load everything up, and when the user want to "turn it off" you let them put it in standby. Almost no power used, almost instant "boot"/"shutdown".

My TiVo can come up from suspend in 2 seconds. My Mac can do the same. If you suspend to the HD it will take longer, but if you just use RAM it can be done FAST.

Re:too slow to boot

[Anonymous Coward]

Err... My Tivo takes about 3m to boot up. Longer if it just crashed or i have been mucking around with the twiddly bits inside. So 30s is a damn-good startup time if you ask me.

Re:too slow to boot

[jawtheshark]

My DVD recorder (for TV) takes at least 60 seconds from 0 to operational. This is a problem one finds all over the place these days. It's not limited to self-made settop boxes.

Thin clients

[Anonymous Coward]

"They target thin clients.."

What about those of us who aren't so thin? Have they had problems with the not so thin using their boards?

Database benchmarks?

[failedlogic]

I'm also curious, how well would this benchmark on something like MySQL (assume its not being used as a media player). I have an iMac G5 which has constantly running background apps. I'd rather get a smaller, less power consuming system for the task. Any opinons?

I've toyed with the idea of getting one of these for a streaming audio player. I'm about to replace my stereo system and I figure why get all the fancy equipment when I'm going to be popping CD's, DVD's and MP3s into the thing .... might as well get a computer! Plus I can use it as a PVR box and control it from my TV.

somewhat disappointed

[Anonymous Coward]

I've been waiting for the C7 boards to ship for a long time. My particular application of interest doesn't need a lot of CPU time but I do need to do a fair amount of encryption work -- so the integrated PadLock stuff should suffice. Plus the C7 now supports the NX MMU bit which I consider a requirement.

HOWEVER, two problems:
    * first they're just now "about to ship" and I've been waiting for quite awhile. Oh well, I guess that's water under the bridge now. If I had known how long they would take to reach market I probably would have looked into some of the other options quite awhile ago.
    * more importantly my application needs a LOT of RAM. All of the older Via-based mini-ITX boards were limited to 1G of RAM. The biggest thing I was excited about in the CN700 chipset is that it supports 2G of RAM -- still not great but "enough" for what I need I guess. But all of the CN700-based boards I've seen announced in the last few weeks are still limited to 1G! Maddening, I tell you. Does anyone know of a CN700+C7 mini-ITX board that can actually take 2G of RAM?

...and no

[phorm]

And as somebody who had to jump through a whole lot of hoops over time to make them work. I'd have to say that the return answer is "sorta."

Yes, there were Epia drivers. They also tended towards bugginess and being a royal pain in the ass to install. Not to mention that the installer was specific for various distributions but nothing debian-based... much to my annoyance.

Mac mini and networking

[whit3]

While it isn't usually done, the Mac mini actually DOES have a second fast network
port (for a total of four: modem, Airport/802.11g, Ethernet 100baset are the others).
On the new Core minis, it's 1000baseT Ethernet and no modem...

Firewire is a capable interface, and IP traffic is one of its capabilities.
It works with a Firewire cable, peer-to-peer, between two Macs and
at 400 Mbit/sec it's sometimes an improvement on crossover-cable-Ethernet.

Re:Mac mini and networking

[ArbitraryConstant]

"Firewire is a capable interface, and IP traffic is one of its capabilities.
It works with a Firewire cable, peer-to-peer, between two Macs and
at 400 Mbit/sec it's sometimes an improvement on crossover-cable-Ethernet."

This isn't very useful for what many of us will use a VIA machine for, and that is firewalling. To provide firewall service to a LAN, a second ethernet port is required. That's fine for a home LAN or small office, but larger networks will require multiple APs even if they're wireless, and that's much easier to do with an ethernet network. It can need more than 2 network ports if there's more than one network segment or other requirements (see below), and the PCI card can handle a 4-port NIC. VIA also generally has a version of the board with two network interfaces, and that plus a single-port NIC is pretty cheap. Also, a Firewire 400 port will typically not be that much faster than a 100 megabit ethernet port, even though the bitrate is nominally faster. While Firewire can handle faster speeds, the nature of IP traffic keeps it from reaching its potential.

An example of a network setup that would need more ethernet ports would be a site that has multiple network segments. You may have different departments, have some servers that you want to isolate, or a wireless and wired network that you want to isolate. You might also want a firewall cluster with failover capability, as is possible on OpenBSD (pfsync + carp), where an additional network interface is required.

Basically, a mini can be adequate for firewalling purposes if the requirements are simple enough, but a VIA machine is much more capable in that role. It's also much cheaper, so you can get yourself a firewall cluster for what a single mini would cost you.

In summary, a mini isn't necessarily totally useless as a firewall, but a VIA machine will be much more useful for demanding situations.

Woo, im in the future!!!

[Anonymous Coward]

Right now I'm working on similar boards with a dual core via processor. Each core 1ghz.

Re:Woo, im in the future!!!

[Anonymous Coward]

my bad, those boards have a fan you cant hear 90% of the time, but they do have a fan and heatsink.

VIA released source

[metamatic]

VIA released source for all the hardware on the M10000. It's gradually being cleaned up and integrated into Linux. For example, 2.6 currently supports the hardware RNG and hardware accelerated X11, and the MPEG hardware is supported in mplayer. Sensors work, ethernet works, Firewire and USB work, all with open source drivers. They do a much better job than most other vendors at supporting Linux.

(If you know of a motherboard with SATA that'll take a CPU that can be passively cooled and has open source drivers for everything, I'd like to hear about it, as I plan to build a bigger server this year.)

Re:VIA released source

[ArbitraryConstant]

"If you know of a motherboard with SATA that'll take a CPU that can be passively cooled and has open source drivers for everything, I'd like to hear about it, as I plan to build a bigger server this year."

In the enlarged picture of the board, you can clearly see two SATA ports at the bottom right.

Re:VIA released source

[arivanov]

Via SP8000 http://www.via.com.tw/en/products/mainboards/mini_ itx/epia_sp/index.jsp

Got two departamental servers running on that. The onboard + 2 Silicon Image Adaptecs (6 disks total) using these enclosures: http://www.scan.co.uk/Products/ProductInfo.asp?Web ProductID=84948

One caveat - the enclosure does not work with newer Maxtor drives because the idiots redefined the power up SATA spec and the meaning of the LED indicator on the power pinout.

Re:VIA released source

[Anonymous Coward]

Whoosh!

(That's the sound of his point flying right over your head.)

No hardware MPEG4 decoding

[TummyX]

Damn. The SP line of ITX boards used the CN400 chipset which supported MPEG4 decoding (in linux at least). Looks like the CN700 will only do MPEG2 which is great for DVDs but not so good for DIVX/XVID.

Re:No hardware MPEG4 decoding

[orv]

Perhaps not so interesting, the CN400 only supports H263 mpeg4 'assistance' and when we last benchmarked we were actually getting better performance using software only decoding than using the hardware.
It will be interesting to see what the next 700 series chips can do since they are slated to support h264 decoding.

No floppy req'd for my Win install on no-RAID SATA

[MojoStan]

Have you tried installing Windows XP on a computer with an SATA hard drive? Oh man. Pain. You actually have to kick the thing into life using drivers loaded off a FLOPPY DISK.

I have installed WinXP and Win2000 on SATA drives (without RAID), and a floppy disk was not required. The floppy install method might still be required to install Windows on SATA RAID (I haven't tried it), which would suck. But from my experience (with Intel motherboards) a floppy is NOT required for installing Windows XP/2000 on a SATA hard disk because (from my experience) the motherboard BIOS settings (not a Windows driver) determines whether you can boot/install Windows on a SATA drive.

Some nForce motherboards I've seen have SATA RAID mode enabled in the BIOS by default, which might require a floppy. One motherboard I installed Windows XP on, the Intel D915GUX, had the SATA boot option disabled by default in the BIOS. After enabling SATA boot and choosing SATA mode without RAID, WinXP w/SP2 installed without a problem.

Maybe SP2 needs to be on the installation CD, but I doubt it because I'm pretty sure I installed Win2000 w/SP4 on a SATA drive (Intel D945GT motherboard) without using a floppy (this was a while ago).

Yes I have

[Andy Dodd]

Gentoo 2005.1 installed with no problems on the SATA-based Dell PowerEdge 850 (or was it 830???) I set up for him.

Same with my SATA-based Athlon X2 server.

Gentoo 2006.0 was flawless on my new Inspiron E1705.

interview with the CEO

[pimpimpim]

on the same website, though a bit old, an interview with the CEO, pretty interesting how they started building their humble line of CPUs amid the big-guy CPU wars:

http://linuxdevices.com/articles/AT2656883479.html

out of interest

[petermgreen]

has anyone tried manually setting that resoloution though editing the registry?

and if so did it work?

Re:Cool but... COST

[Anonymous Coward]

Furthermore, at what Apple pays people to astroturf, you can just go snag a Mac Mini.

The Mac Mini is a far better platform anyway. If Apple wants to pay people to tell folks about how great the Mac Mini is, then more power to Apple. Mac Mini Mac Mini.

Anecdote

[metamatic]

Well, my M10000 has 181 days of uptime... and it's only that low because I powered it down to rearrange some cabling. It has never locked up.

On the other hand, a friend went through hell with nForce.

Also, you start off talking about VIA under Linux, then suddenly switch to talking about Asus under Windows as if that's relevant. Huh?

10 seconds? That's a speed demon

[metamatic]

Disappointing to say the least, no decent set-top box can take more than 5, maybe 10 seconds to start.

Tell that to TiVo and DirecTV. My DirecTiVo takes well over a minute.

Re:No floppy req'd for my Win install on no-RAID S

[MrFrank]

I had a similar problem with my Dell 8400. Great machine by the way.

Anyway, the BIOS had three settings for the SATA chipset, something like: legacy (emulate PATA), non RAID and RAID (It has 4 SATA ports). With the WinXP home SP1a install disk I got with the system I had to provide a driver on a floppy or set the bios to legacy. My nieghbor who also has a 8400, but got WinXP home SP2 could set hers to non-RAID SATA.

The drivers needed for the SATA on the chipset (Intel 925, I think) were built into WinXP SP2.

Just about any 939 Athlon64 board

[Anonymous Coward]

If you put a nice big heatsink with heatpipes on it, plus undervoltage and underclock the CPU you can run it hard without overheating even in a hot room. At least it has worked for me. (Though I'm using a power supply with a fan -- but I have the speed turned down as low as it can go -- I can't guarantee it will work with a no-fan power supply.) The CPU doesn't even get very hot. The video card and chipset both run hotter than it does.

Re:Just about any 939 Athlon64 board

[Anonymous Coward]

okay, I'm not sure I believe you, but you know, you don't have to have the power supply be inside, you could use a brick and deliver dc voltage, then it doesn't matter how hot your brick gets (it's outside and you know won't stop running or anything just because it gets a little hot) -- plus, how hot the power supply gets depends on its efficiency, and since a brick is less crammed for space, you can find higher-efficiency ones. (The heat in a power supply is conversion loss).

So, tell me about your Athlon64, which you undervoltaged and underclocked, and only passively cool with a heatsink with heatpipes on it: what model CPU is it? (speed), how much did you undervoltage/underclock it? (Did you follow directions from somewhere? where?) what model of heatsink with heatpipes on it are you using, and how warm is it where you live? (So I can get an idea of "hot room.").

proper passive cooling

[r00t]

Choose a heat sink with a thick base and widely spaced fins that all run the same direction. The fins should be about 1/4 inch apart. Do not orient the fins or CPU horizontally. Air should be able to rise (hot air rises) through the fins from bottom to top without being blocked by components above or below. As always, a massive copper heatsink is best.

Guide the air. There should be a smooth tube running from the bottom of the PC to the top of the PC, with the heat sink embedded in the middle. The heat sink fills the tube, and is not open except to the tube. Continue the tube up and out of the top of the PC as a chimney, going up at least 1 foot. The bottom of the PC, where the lower end of the tube is open, needs to be lifted off your desk or floor. Air flow needs to be directed vertically but otherwise very free of restrictions.

Re:proper passive cooling

[Anonymous Coward]

Don't worry about having to cut holes in your peripheral cards to achieve this. There's a lot of spare silicon in those that never gets used and is just there to justify the high price.

Re:proper passive cooling

[arivanov]

I will second that. I look after 20-30+ EPIA systems. Mostly M, but some V, MII, TC, SP - so nearly all varieties are represented. I use them for all odds-n-sods servers (DNS, News, SMTP, firewalls, VPN using the Via AES accel, archiving, alerts, even some slower file services). The only times I had thermal problems with them was when there was obstruction from cables.

it's wicked fast at crypto

[r00t]

The chip has special inctructions for AES, SHA, RSA, and true (non-algorithmic) random number generation.

You can do a round of AES crypto in 1 or 2 clock cycles. This chip does AES about 8x faster than the fastest Intel and AMD CPUs, and much much faster clock-for-clock.

So if you run a web server or ssh server that gets bogged down by crypto, get this CPU.

Re:it's wicked fast at crypto

[jandrese]

Unless you have a cycle hungry database to run on the same server of course.

Re:it's wicked fast at crypto

[jpc]

Yes I am really interested in trying this out, and seeing what its really like.

Whens it shipping?

Re:it's wicked fast at crypto

[Anonymous Coward]

Yeah. I want to try the next generation of freenet (not out yet) on one of these for the same reason, as you would expect it spends a lot of time doing crypto and hashing. They released an open source JCP driver some time ago, so Java apps can pretty much use it transparently.

Flex-ATX

[yoprst]

Flex-ATX mobos with the same specs as for M10000 are sold for $60 (and you've got 2 ddr sockets and more pci slots), and flex isn't much bigger. The only problem is to find one...

Buy a better power supply

[Anonymous Coward]

Your problem is not the board - it's the crappy way you are trying to power it. Get a board that provides a regulated 12VDC rail - Opus Solutions have some nice boards, Mini Box also have some nice stuff which is aimed at the Car PC enthusiast. If you are using one of the really cheap Morex boards that require a regulated 12VDC input then you have the wrong PSU for the job.

Re:Buy a better power supply

[scsirob]

I'm aware of the external power options, both good and bad. I was refering to the Epia-TC boards that can be powered by 12V. My point is that today's boards don't need 12V, so all on-board voltages are transformed and regulated down anyway. Why can't they make the input range a bit wider?

PCI add-in

[KlaymenDK]

My M10000-based m0n0wall has a 4-port PCI ethernet card in it -- instant built-in hub in the same box as the firewall. :) But I grant you this: an additional ethernet port on the motherboard could be used for a DMZ.

Price? What's the price? Anybody know the price?

[Qbertino]

Price.

function checkOutCheapAndSmallComp()
{
  if (newMiniITXLine.notablyCheaperThan(MacMini))
  {
  return interesting = true;
  } else {
  return pointless = utterly;
  }
}

Since the MacMini all the small component, low power PC solutions have missed out. Only if MiniITX board and CPU+casing+powersupply+drive+ram+hdd is cheaper than the MacMini, only then will it stand a chance.

Re:Price List

[RockDoctor]

For certain product categories, it has to be some of the most effective advertising available,
I take it this means that you've, errr, succumbed to the evil advertising genius of the Slashdot mafiosi then. What did you buy?

[MAXIMS, with elements of truth]

There's no such thing as BAD publicity.

There's no good advert like a satisfied customer.

and, the scary corollary that could make Slashdont the worst advertising medium in the world ...

There's no bad advert like a dissatisfied sustomer.

With Slashdot being (at most) lightly moderated, the prospect of getting bad reviews from people with real experiences should be enough to give any intelligent marketing 'droid the dry heavings.

Yeah, I know : "Intelligent marketing 'droid" = "contradiction in terms".

Aaaah, good ole XF86Config. 'Miss that. :-)

[Qbertino]

X Configuration. Yeah, those were the days.
Setting Color, V-Frequency and Resolution to a non-sucking condition on Suse 6.4. Getting the newest NVidia drivers to run on Suse 7.2 only to watch Sax wreck havoc on the XF86Config. Sweating bulltets while trying to recover X into runable condition. Finding the right setting on Debian to run Loki's Tribes 2 in hardware mode.
Nothing like hand-cofigging your XF86Config. All you miss is the dirt, heat and steam and having to shovel coals into a hatch below your PSU every 10 minutes.

I'll still don't know what half of those options mean and I'll probably never know :-)

*patts his iBook w. OS X Tiger*

perfect firewall/router proc

[gosand]

I run a Pentium90 as my firewall/router. (yes, you read that right) It is truly fanless, except for the power supply, and does a great job. It was also quite cheap (free). The only drawback is that it takes up a bit of space. But you don't really need even that much processing power for a firewall/router. Michael

Re:No floppy req'd for my Win install on no-RAID S

[Wiz]

Well it isn't like that anymore let me assure you!

On the Intel 945 chipset, Windows installs fine if SATA is set to IDE mode which is the default. If you set it to AHCI *OR* RAID mode, then it requires drivers from Intel http://www.intel.com/design/chipsets/matrixstorage _sb.htm. This was with a clean XP SP2, and I'd imagine it'd be the same for Win2k also.

Generally SATA controllers do require drivers, unless they are in an IDE compatabile mode. It appears that the Intel 945 has a newer ICH which is likely the reason why this is the case. Going forward, I can't see the situation changing.

RAID mode isn't the issue, it is the IDE compatability mode.

Console redirection?

[erice]

Does recovering from network failure or intial configuration involve attaching a notebook computer via serial cable or do I have to to find space/power to plug in a monitor and keyboard?

Not really

[Anonymous Coward]

More like Vista is redefining what "fat" is.

OT: Tell me more about specifying ModeLine in X

[KWTm]

Hello. Can you give me some guidace as to how you got the numbers for the ModeLine command? I need something similar, but for two other types of screens. One is for a new 1440x900 LCD monitor I plan to buy; the odd dimensions are because it doubles as a widescreen TV. The other is for my ATI Rage Fury Pro TV output; I can't seem to get the TV output to work with the newer X.org arrangements like Ubuntu, and am stuck using the XFree86 in Mandrake 10.0-o.

I've tried digging around for safe numbers to use on my screens that won't fry them, but I haven't come up with anything practical.

Re:OT: Tell me more about specifying ModeLine in X

[Anonymous Coward]

I recommend you to look at the X.conf that get posted at linuxlaptops for 17 inch laptop screens.
I own one with a 1280x800 and another with 1440x900 and is a bit of a pain to get those resolution. In newer X servers (and newer distributions) that can be picked up by the configurator. In SuSE 9.3 and 10.0 that works great.

you can also have a look at www.amiloforum.de and look for a X.conf for an Fujitsu-Siemens Amilo M3438G (1440x900). Is it in german, or look for some DULL model or HP in that web sire. I hope it helps !

Number of /. uids in a generation

[alienmole]

That's pretty sobering: a Slashdot uid difference of 856730 represents an entire human generation...

Via Mini-Itx boards have lockup issues

[nihkee]

I've got an EPIA 800 board which is supposed to work as a NAS/print/web server.

It locks up every now and them. Even when it's not under any particular load. My impression is that it's got a hardware bug and VIA are incompetent since they've been unable to fix it as the problem exists in various EPIA models.

I've been looking for an alternative since no way I'm going to buy another EPIA ever again. Pathetic, low-performance boards with issues.

See http://forums.viaarena.com/categories.aspx?catid=2 8&entercat=y if you care. Look for "hang", "dma" and "lock"