Typically well-written review. I wasn't surprised when a system built with this board took first place in the low-power shootout, but it does surprisingly well in my media-encoding-efficiency shootout as well (see Lowest power consumption for DivX encoding).

Well written review and the board is good for a low-end HTPC / kioks type machine.

Had wanted to use such a solution (ie VIA C3/Eden/C7 + mini-ITX combination) for a 24x7 home file server (with terabytes of SATA RAID 5 based storage). The main draw for me towards this is the low power consumption. However, with only 1 PCI slot, it's really not a viable solution for me. I'll like to start with a 4-port SATA PCI controller. Then, add more SATA PCI controllers and drives later when my needs increases. For now, the AMD Turion 64 on Socket 754 seems more like a better solution for me. Anyway, I'm still researching.

It would be great if the mainstream motherboard makers can put the VIA Eden / C7 processors into microATX / ATX motherboards.

Or better still, VIA package their processors into Socket 478/754/939/LGA775 packages that can be dropped into a mainstream motherboard (like they did with the VIA C3 for Socket370).

Cheers

And Devon is, of course, quite kind in his comments about the vendor.

Possibly the problems with Photoshop and the video are due to more than just software drivers - maybe CPU or chipset bugs?? This is not a widely used nor tested products (with Win XP / Apps ).

That had occurred to me. The C7 implements SSE, SSE2, and SSE3, and I could see those implementations being incomplete/error prone.

Still, I would also expect the hardware testing process to be much more rigorous than for drivers. I wanted to give them the benefit of the doubt.

I also tried swapping out the power supply, and would have tried the RAM if I had been able to find another working stick (DDR2 800 wouldn't boot).

Any idea on the cost of these boards? The performance per watt when used with the picoPSU is staggering.

The review stated:
“It also cannot handle HD video properly — at least not when it is encoded in Windows Media format. It is possible that an HD clip encoded in MPEG 2 — as is likely to be found on upcoming HD DVD and Blue-Ray discs — would have been able to take advantage of the hardware decoder and played back properly.â€

Ehhh...IF you did a little more research, you would have realised that you can buy active PCI-riser cards that allow you to install up to 3 PCI cards into a single PCI slot. You can also get flexible PCI extenders too.

Personally, it sounds like you're better off considering Serial Attach SCSI...With its ability to handle heaps of SAS and SATA drives.

Actually, I believe that HD-DVD and Blu-Ray both support MPEG2 and AVC codecs. From what I understand, the few HD-DVDs that have been released in Japan have been done in MPEG2 format because mastering facilities for AVC video do not yet exist. Obviously, things will change over time, but I don't think MPEG2 is dead by a long shot. There's simply too much infrastructure for it.

http://www.logicsupply.com/product_info ... cts_id/544

logic supply shows this at $272

Ran across this article through [H]ard OCP. From my perspective, the deal killer on a board like this is the cost, even with the fact that it includes the processor. If the board included a 16x pci-e slot and was $100 cheaper, I could see it as a good deal. You can buy a Sempron 3400+ AM2 and a Biostar T-Force motherboard for ~$180. Of course, what kills the mainstream offering is no mini-itx format.

I know Commell makes a mini-itx board with a pci-e 16x slot, but that runs over $350!

Is there a happy medium?

Best Regards.

Cheers,

Peter

MPEG2 support is there primarily for legacy DVD playback I assume; for the VIA platform to support ‘typical’ HD format discs they will need to add H.264 acceleration at some point. From what you stated it seems as if HD-DVD players are pretty flexible.

interesting review and a worthy addition to the SPCR knowledge base. I do get the feeling that the compromises in performance needed to get power consumption lower and lower are just a bit too severe. Given that the 17in iMac you reviewed draws little over the Via set up at idle and ~30W at load including a monitor but boasts a dual core CPU and discrete graphics. I know I'm not comparing like with like and that the Via should have lower consumption whatever, but the tradeoffs for those last few watts seems too high.

If it's for a dedicated low power set up (file server, media center perhaps), particularly one that's on 24/7, then I'd imagine the higher cost and tradeoffs would make it worth it, more so if power savings are taken into account. Other than that, especially for peeps that can only afford 1 PC, it's just too focused IMO.

Nice review, I really like the fact that you don't bore us with graphs and such. You tell us what we need to know which is all that matters.

More than 1 raid card on desktop PCI kills performance anyway, since the PCI bandwidth is shared between all ports.

When I started to read this article, I was thinking, "This would be perfect for a HTPC...I wonder if I could use it for my Warcraft 3 back up machine." My brothers and I are hooked on WC3 and it would be nice to have a second machine to run it when one of my brothers is over. This article answered my question...kinda. I was wondering what the difference of adding a PCI GFX card would be to WC3 playability. I realize PCI is not the best interface for GFX, but theres no other choice, eh? Any chance you could pop in an old PCI GFX card and run WC3 through it paces?

Here's hopin'

its great to see some Mini-itx reviews on this site.

on the bit about performance. i'd love to see you test out a Mini-Itx board with a pci-express 16x slot, and slap in somthin like the 7600gs graphics card. (which comes stock passive in a single slot configuration)

here are 2 boards i think have alot to offer. both have pci-express 16x, and can run either a pentium-m or the new core duo.

http://www.logicsupply.com/product_info ... cts_id/552

http://www.logicsupply.com/product_info ... cts_id/410

one thing im sure a reviewer would point out is the fact that you cant use "mainstream" standard cpu coolers on these. Though the coolers they use are a standard 50x50mm often listed under "industrial" cpu coolers. you can get quite a few fairly beefy all copper coolers in this standard. no heatpipes that ive seen so far, but they should do alot better than the stock aluminum coolers that come with these motherboards.

theirs definately a change in the winds. you can see it even from intel who is claiming "ghz isnt everything" now. I think even the most hard core gammer would be satisfied on a low power cpu system with a top of the line video card.

I don't think there is any point in a VIA CPU + PCIe video card. Even the C7 has very poor floating point (FP) performance. GPU T&L accelerations has removed a lot of the tradional FP calculations from the CPU, but most games still require quite a bit of FP performance from the CPU.

A nice comparison to this ITX board would be an undervolted 90nm Sempron 2500+ running on a VIA IGP motherboard using the same power supply. I think the K8 can run cooler and faster than a C7.

This board is fine. but it is at least $100 too expensive.

Devonavar, something is very wrong with your review sample.

Why is the CPU temp LOWER than the System temp?!

Even your BIOS screen capture confirms this! Did VIA screw up with the labels in the BIOS again?!

Did any of you folks notice this discrepency?

Do you have one of those electronic thermometers to confirm the BIOS and Speedfan numbers?


OT: I'm not gonna take credit for noticing this, as orF from OCAU (Overclockers Australia) forums, pointed this out.

Why not? Do you know where the "system" sensor is located?

No, do you? That's what I'm curious about.

Why is the CPU temp lower? Is it a bad reading? Mis-label in BIOS? Or is this a genuine improvement? Has the reviewer explored or verified this in any way via other means?


OT: That's the problem with this generation...No one questions anymore, just take in what you're given!

What discrepency?????

"System" is not an ambient air sensor. It may be a sensor on the IGP or one of the VRMs.

Stress testing my Sempron 2800+ 1.4 GHz @ 0.88 V in my Tforce 6100, the "System" temperature is much higher than the CPU internal diode reading. It *should* be. The undervolted K8 runs very, very cool (especially with a Ninja on it), and the IGP (which I assume is where the system sensor is) is still working hard. It is even worse if I am using the IGP for video.

BTW: the only thing special about this C7 is that VIA actually set the default voltage at a proper value. AMD has most of the slower K8's running on way too much voltage by default.

Also AMD has the 1.1 V Vcore limit problem on some of the newer K8s (and/or motherboards). 1.1 V is still too much voltage for the sub 2.0 Ghz K8s, and SPCR hasn't tested any sub 1.1 V K8s either.

Of course I noticed it. But, what do you want me to make of it? I can question it, say "hmm, that's funny", and that's about it. As others have pointed out, we have no idea what it is supposed to be monitoring. I don't buy that the sensors are mixed up; the CPU temperature increased much more than the "system" temperature under load, just like it should have.

You will notice that in our other reviews we don't generally include system temperatures. The reason is precisely the difficulty that has been raised here: We don't know what's being reported. I would have omitted it in this review too but for the fact that it did read considerably higher than the CPU temperature.

Either that, or they just have a system of their own where the "system" sensor routinely reads higher than the CPU sensor, and came up with reasonable explanation for the phenomenon.

Trust me, I do plenty of questioning in general.

Why shouldn't the CPU temp be lower than the system temp?
My boards 'system' sensor is actually located down near the floppy drive connector & ITE controller chip at the front of the board (I asked Foxconn).
Idling, the CPU temp is currently 18C and system is 25C and both are above ambient. What's wrong with that?
If I run CPU Burn for a while then things will be different and the CPU will then be hotter.

I (have) own(ed) several VIA EPIA boards. On the early models (with C3 Samuel II CPUs) they had separate heatsinks for the CPU and the Northbridge. It was quite common for the heatsink on the northbridge to be hotter than the one on the CPU. That is why VIA moved to a single heatsink for first both chips and now for all three of them.

Don't forget that the VIA Eden CPUs are very lower power, so there is nothing strange about them not being the hottest part of the PC.

Identifying which sensor is the CPU by looking at which one gives the highest reading is not a smart thing to do with modern low power CPUs. You should look at a graph of all sensors when you start loading the CPU. The one that has the steepest jump is the CPU, not the one that ends the highest. In my undervolted A64 3000+ system even the HD is hotter than the CPU.

You should also note that the person pointing this out to you is an overclocker. Some overclocked and overvolted CPU is much more likely to be hot than the newest low power x86 CPU.

For what it's worth, I work for a company that makes appliance-based firewalls, among other things. Some of our newer boxes use VIA CPUs, either Eden or C3, and recently we've been noticing that the "system" temperature sensor sometimes reports very high temps (up to 90 C in one case). I'd speculate that the sensor for that reading is integrated into the VIA chipset itself, but I don't know for sure... in any case, it's at least possible that the sensor isn't very well calibrated with respect to "real" temperature.

couldn't one use an IR thermometer to find the temperatures of various parts of the board, and then see which one best correlates with what this "system" sensor is reporting? IMO it's most likely to be the power MOSFETs.

PS. I do a lot of questioning too. No old person "your generation" stereotyping please.

There is that big heatsink covering most of the board. How will you point that IR thermometer at the CPU and the northbridge?

From the pictures in the review it looks like the main heatsink is secured by pushpins. They say it is not easily removable, but it should come off; the low thermal dissipation means that the board can be run long enough in this configuration to take measurements. It is not rocket science.

It's not rocket science. Just a $272 "mistake" if you manage to break one or two of those plastic pins. (Murphy says plastic pins always break.)

While I certainly don't want to discourage people from asking questions and digging for answers, devices like this all-embedded board are often best treated as black boxes, from a review point of view. Why? Because you can't change a dang thing anyway... and there's always decisions to be made about best use of time & energy -- it doesn't make sense for us to spend beyond a certain amount with most items we review. There's just too much pressure to generate new content with this always-growing list of samples. Our reviews are meant to give a concise and thorough analysis and appraisal, mostly from our silencentric and efficiency-focused PoV; beyond that, unless it's a truly exceptional item, we're not likely to report on it again, so exchanging experiences w/other users in these forums is your best bet.

In this case, whether any of the system/board misbehavior can be attributed to too-high temps anywhere would be very difficult to ascertain. It's unlikely anyway.

We had no desire to remove the HS. Still don't.

We don't believe any instability or negative consequence arises from the system temp being higher than CPU temp -- as others have chimed it, it's not unusual for this to be the case in many modern systems.