Intel D510M0 Motherboard: Atom 2.0

Want to talk about one of the articles in SPCR? Here's the forum for you.
HFat
Posts: 1753
Joined: Thu Jul 03, 2008 4:27 am
Location: Switzerland

Post by HFat » Sat Feb 13, 2010 7:14 am

Another development in the D510MO & T3310 story...
HFat wrote:After laying it horizontally ...
That seems to have been a mistake. Intel said one would get better cooling by having airflow along the largest dimension of the heatsink but I didn't think convection would create much airflow there. Apparently, I was wrong: temperatures have dropped shortly after I positioned the case vertically on its stand. The case is poorly suited for that because the power supply must be heated by the airflow from the board and because the ATX cable is obstructing airflow (perhaps I could tie it out of the way somehow). Still, reported temperatures have dropped about 13 and 10 degrees to 37C and 54C (21C ambient). At least a hard drive would be out of the way of the warm air in that position.

I suppose a good case for the D510MO would have nothing obstructing airflow between the heasink and a large vent when the board is positionned vertically. Which cases would qualify besides the T3500 sent to SPCR?

For what that's worth, I'm currently running LXDE on the current i386 Rawhide and I've connected a USB optical drive, a mouse and 100M ethernet. I'm not seeing significant (>2C) temperature differences between BIOS screens, memtest or Linux (maybe the drivers are not idling all the components properly).
HFat wrote:Does anyone know where the "remote" sensor is located?
To answer another of my questions, according to Intel's technical product specification, it is between the heatsink and the edge of the board, in the "processor voltage regulator area". Intel says its temperature can rise up to 85C in an open chassis. Interestingly, it also gives 99C for the processor (perhaps that can be approached when the GPU as well as the CPU are stressed but I'm skeptical as this heatsink is doing a good job) and 113C for the NM10 chip (the one which hurt me when I pressed my finger against it yesterday).

And I'll close this update with another question for the experts (should one show up): is there some rule of thumb to estimate how much higher the board's temperatures would get if the ambient temperature was higher. I guess it's a bit less than +1C per +1C but maybe not.
Last edited by HFat on Sun Feb 14, 2010 1:35 am, edited 2 times in total.

MoJo
Posts: 773
Joined: Mon Jan 13, 2003 9:20 am
Location: UK

Post by MoJo » Sat Feb 13, 2010 10:32 am

HFat wrote:The case is poorly suited for that because the power supply must be heated by the airflow from the board
I wouldn't worry about it. A modern low wattage DC-DC PSU like that won't have any heat issues.
HFat wrote:Intel says its temperature can rise up to 85C in an open chassis. Interestingly, it also gives 99C for the processor (perhaps that can be approached when the GPU as well as the CPU are stressed but I'm skeptical as this heatsink is doing a good job) and 113C for the NM10 chip (the one which hurt me when I pressed my finger against it yesterday).
Not sure about the CPU, 99C seems a bit high but then again nVidia GPUs can reach 110C without problems. You would think that such a low power CPU would need extremely bad cooling to get anywhere near 99C though. I know the thermal paste Intel use is sometimes terrible (more like chewing gum) but even so...

For the northbridge 113C is in line with out Intel chipsets. I have a G33 based Intel board and the supplied monitoring software shows temperature levels only reaching the yellow zone at about 110C and red at 115C. It idles at about 95C and doesn't come with a heatsink.

HFat
Posts: 1753
Joined: Thu Jul 03, 2008 4:27 am
Location: Switzerland

Post by HFat » Sat Feb 13, 2010 11:22 am

MoJo wrote:For the northbridge 113C is in line with out Intel chipsets.
It's the southbridge. It's supposed to be a 2W part. There's no northbridge.

crisspy
Friend of SPCR
Posts: 228
Joined: Sun Sep 29, 2002 9:05 pm
Location: Powell River, BC, Canada

Post by crisspy » Sat Feb 13, 2010 8:00 pm

More heatsink ideas: Zalman, Thermalright, Enzotech and Arctic all have numerous heatsinks and products that would be a good idea to bump up the cooling on at least the SB chip. Arctic Silver also does thermal epoxy adhesive, which would presumably be better performing than the "thermal tape" sold with various RAM sink products. The little RAM sink packs look like an ideal candidate for the SB, they are about the right size, and cost $8 to $16 per 8-pack, thermal tape included. So you even have spares for other projects.

I also just popped the big stock heatsink off my board, and I can only say it would be hard not to do better with any choice of alternative heatsink. Lots of cheap options out there for bigger badder sinks, including all the heatpipe stuff, including VGA targeted ones. The D510MO has a nice big open area there too, with four handy dandy holes to solidly bolt something down.

I know the Intel stock sink is OK for passive in the right orientation, but it's still a pretty wimpy candidate, in all honesty. It looks like the kind of thing we used to replace 10 years back when the whole CPU cooling thing actually started to take off. And I think I need only mention the little generic square of thermal wax to illustrate how easy it will be to do much much better than stock. Without hardly even trying. It's not even 25 watts.

I think I will take a good look at the heatpipe options, since you can get almost anything you would want for under $40. I can see putting the radiator side outside the case area, thus sidestepping at least half of the heat buildup issue altogether. Holes in cases are easy to make. I'll see what coagulates in my mind over the next while.

HFat
Posts: 1753
Joined: Thu Jul 03, 2008 4:27 am
Location: Switzerland

Post by HFat » Sun Feb 14, 2010 12:25 am

crisspy wrote:Arctic Silver also does thermal epoxy adhesive, which would presumably be better performing than the "thermal tape" sold with various RAM sink products. The little RAM sink packs look like an ideal candidate for the SB, they are about the right size, and cost $8 to $16 per 8-pack, thermal tape included.
Good idea, but now that I see that:
-its maximum temperature is supposed to be 28C higher than the voltage regulator aera
-it's only a 2W part
-the "voltage regulator area" is not directly cooled by any heatsink yet its heat is dissipating
... unless I've got at least some of this wrong, I see no point in cooling the NM10 chip in a vented case.
crisspy wrote:The D510MO has a nice big open area there too, with four handy dandy holes to solidly bolt something down.
Between the RAM, the ATX connector, the caps and the cables, I'm not seeing such a big open area (at least not in this small case). And, seeing that you're not supposed to ever touch the CPU, I don't know that we can safely make assumptions about how the sink is attached to the board until someone actually removes the thing.
crisspy wrote:I know the Intel stock sink is OK for passive in the right orientation, but it's still a pretty wimpy candidate, in all honesty.
It looks more than just OK. I haven't tried to stress the GPU yet but stressing the memory controller isn't heating it much (2C at most) as far as I can see and the Atom cores are supposed to have a higher TDP than everything else on the CPU anyway. I acknowledge I don't know which operations would heat them up most but calculating primes for hours on end is only raising the CPU's temperature by 8C when the board is positionned vertically. In a vented case, it looks like it's the "voltage regulator area" which would be most prone to overheating and that only heated up about 4.5C under that same load.
Now, this is in a vented case that I assume is unsuited for your application. In a closed case, yeah: you'd need a much larger heatsink. In fact, you'd probably need to connect the heatsink to the case and use that as a heatsink.

EDIT: the heatsink is not that efficient horizontally, particularly for indirect cooling so that temperatures rose by about 11C and 9C under CPU load in my latest test. The "voltage regulator" reached 73C. This is bad: in the summer without AC or if the GPU was being stressed as well, I can see how one could get close to or even surpass Intel's maximum temperature for that area. Such temperatures can't be good for the other components either. A hard drive would be close to the heat pathway in this position for instance. So I can only recommend the T3310 for the D510MO if you're going to have the board vertical or if you're going to be careful with the load.
crisspy wrote:I can see putting the radiator side outside the case area, thus sidestepping at least half of the heat buildup issue altogether. Holes in cases are easy to make.
Wouldn't one of these closed cases designed to be used as a heatsink from the ground up (the ones which come with external radiators) make more sense?

Something easy which I've been thinking about and which might help with the cooling of the parts which are not directly cooled by the heatsink whether the case is closed or vented would be one or several large thermal pads such as the one used in SPCR's D945GSEJT & T1610 review. Does anyone know where to source good ones and whether they actually help?

crisspy
Friend of SPCR
Posts: 228
Joined: Sun Sep 29, 2002 9:05 pm
Location: Powell River, BC, Canada

Post by crisspy » Sun Feb 14, 2010 5:38 am

crisspy wrote:I also just popped the big stock heatsink off my board...
Hfat wrote:And, seeing that you're not supposed to ever touch the CPU, I don't know that we can safely make assumptions about how the sink is attached to the board until someone actually removes the thing.
In that order. And "not supposed to ever touch the CPU"... HA.

The stock heat sink is held on by 4 push-lock pins. Reach in between the fins with some small pliers and pull out the small center locking pins by their heads (the topmost knob). They pop out, which allows space for the fat locking bottom ends of the bigger outer pins to squeeze inwards where they lock into the holes in the circuit board. Then you just give a little squeeze + push motion from the bottom side of the board, and the pins pop through the holes.

Now that I've rubbed it in, I will say again that the stock HS is pretty pathetic. The most basic starting point would be a copper heat spreader, not just a slightly thicker mid section on a thin all-aluminum unit. Or better yet, a real all copper low profile unit like they use in thin rack mount servers. The only thing I see good about the stock unit is it's wide fin spacing to allow for passive airflow to not be a total flop.

And I will also repeat that the generic thermal wax pad they use in factory installs like this can only be mediocre at best.

Now I would like some clarification: I can't ratify
the one which hurt me when I pressed my finger against it yesterday
with...
-its maximum temperature is supposed to be 28C higher than the voltage regulator aera
-it's only a 2W part
... and still come up feeling good about those kinds of temps. The bottom line is that cooler is better, and even 2W in a small BGA that get so hot as to hurt fingers, is not a good situation.

I vote for installing both a RAM heat sink kit, and a MOSFET heat sink kit. If I spend $30CAD between those two, I will still have some left over for other projects. There are 7 MOSFETS on the D510MO, plus the SB, plus maybe a couple of other smaller chips where it wouldn't hurt to add extra cooling if the leftovers would otherwise just sit in your drawer gathering dust.

There is no such thing as "supposed to be 28C higher" than anything. It's an engineer who is willing to risk reliability over shaving a few pennies of the price of a product they sell by the tens of thousands to OEM's. We can do better.

HFat
Posts: 1753
Joined: Thu Jul 03, 2008 4:27 am
Location: Switzerland

Post by HFat » Sun Feb 14, 2010 6:38 am

I thought you had another board you were comparing to this one, sorry. The D510MO just became available here but I forgot they've been available in other locales for some time.

Since you've got another one and that no one else posted temps, could you say whether your think mine are in line with yours or if one of us might have an outlier? I assume you haven't got my Morex case but I also posted open case equilibrium temperatures at the BIOS screen.
My main purpose in posting those temps is to allow us to compare cases but it's no good if my board is an outlier.
Now I would like some clarification: I can't ratify
...
I'll try to clarify by paraphrasing myself but I think we've simply got different perspectives because we've got different goals.
When I was only looking at horizontal temperatures without a guideline, the SB seemed too hot. Now that I know it's supposed to be hot and that I discovered that the vertical temperatures were much lower, it seems acceptable.
Also, based on what Intel claims, the priority (at least for open cases) as far as cooling is concerned is the "regulator area", not the SB chip. It seems to me that improving the heatsink or its connection with the CPU has more potential.
Yes, cooler is better. But there's such a thing as good enough in my book, especially for cheap, crippled systems. The temperatures are not the worst thing about this board (depending on your application).

Of course I would prefer a cooler SB and of course we can do better but the bottom line for me is whether this guy made the right call:
It's an engineer who is willing to risk reliability over shaving a few pennies of the price of a product they sell by the tens of thousands to OEM's.
How much worse is the reliability going to be for the hot SB? If it would result in 2% of board failures per year for instance (this is by no strech of the imagination an estimate, just a random example), I'd rather leave the SB alone for the time being. It it was more like 10% per year then yeah, I'd want to fit some kind of heatsink on it. I kind of trust Intel's people by default and I realize I may make a fool of myself by doing so but it's no big deal really: these boards are cheap parts and the downtime cost in case of a failure for my application would be very small as well. So I'm inclined to risk it.
Now, if you happen to know that Intel made a bad call and if you have data to back your claim, I'd be very interested in your reasoning. I'm willing to believe Intel messed up on this one and I'm open to tweaking these boards. I just need a more rational grounds than simple perfectionism. Can you actually estimate failure rates for these components (I sure can't)?



Since I can't post, I'm EDITing this in: D510MO, T3310 and a hard-drive
This'll give one more argument to crisspy. I've mounted an old mechanical hard-drive (single.platter 5400.2) in the "stock position". This location is not ideal in that it obstructs airflow somewhat. I didn't get out of my way to use short cables and to position them so as to preserve airflow but the preliminary result I get is bad: vertical idle temps have gone up by slightly less than 4 and 3 degrees which is a lot compared to the increase I got for computing primes, especially when one looks at the "regulator area" temperature.
Worse: when the hard-drive heats up because it's stressed, it seems to be heating the board a bit (it's in the way of the air intake). I doubt this can be fully explained by the load disk access puts on the board because it also happens with SMART self-tests. At CPU load, with a hotter baseline, the effect seems smaller but it still there. The effect is barely significant (with this drive) so I don't want to overstate it.
The good news is that the drive itself remains fairly cool at idle even though it's very close to hot components and hard-monted to a metal case (35C according to SMART or 14C above ambient). Suprisingly, the temperature reported by SMART barely responds to CPU load when the drive remains idle.
Due to the reduced case airflow, the boards warms significantly more in response to CPU load when a hard drive is mounted however: +9.5C and +6.5C. My main concern with the use of a 2.5'' drive is therefore its effect on the already hot "voltage regulator" area (+5.5C at CPU load compared to the diskless temperature).

When the board is laid horizontally, the hard drive seems to have the same effect on board temperatures but the drive also gets cooked: 50C idle.
Note: I haven't allowed as much time for the temperatures to stabilize in this particular case as I did in the others and it may be that the system would have heated up one or two degrees more if I had. I doubt it but I can't rule it out.
Last edited by HFat on Mon Feb 15, 2010 12:09 am, edited 1 time in total.

crisspy
Friend of SPCR
Posts: 228
Joined: Sun Sep 29, 2002 9:05 pm
Location: Powell River, BC, Canada

Post by crisspy » Sun Feb 14, 2010 4:50 pm

Hey Hfat,

I just got my D510MO recently too. I ordered it from www.mini-box.com, because they seemed to be the first place to get the thing in stock this January, and they are also of course the main source for the PicoPSU. I waited patiently for about 6 months for this board to come out, knowing that a build around an older 330 + 945GC (such as Intel D945GCLF2D) would burn around 2x the power, and very low wattage is what will actually get me through the day. Of course I could have gone with a 270 + GSE or a Via or somesuch, but I need better performance, as in dual core to reduce system stutters that could interfere with real time data collection.

So, having only had mine a few weeks, and being very busy with work, I have only posted the board and played with BIOS, board open on my desk, to see that it is at least alive and happy with the older 2GB RAM that I put in and the PicoPSU.

For preliminary thoughts, I will liken the situation to this imagined experiment: put a 25W light bulb in a small closed box and watch the temps go ever up. It's bad news. Now cut a few holes in the box so that air can drift through. It's slightly less bad news, maybe even good enough to live with, but dicey any way I juggle it in my mind. Now IF that whole box was a reasonable heat sink, and reasonably well coupled to the heat source, maybe the whole thing would only get warm, not hot. Or if the holes are really big and really open, and the air can really move freely, then maybe just maybe there wouldn't be a problem.

That's my intuitive take on it. I imagine that an ultra quiet fan to force the airflow, almost no matter how minimal, would VASTLY improve the situation. An 80mm or 120mm, a low speed /quiet model to start with, then under-volted into pure silence, would be good insurance that you would have to see or put your ear against to even know it was still spinning. But that draft would do the job.

I am not an expert on maximum allowable temps. But I know that in higher temperature ranges, there is usually a definite lifespan vs. temperature curve, that pits service years vs. tens of degrees C. So add an extra 20+ C, and you have moved significantly along that curve, to the worse. ALL these chips use the same plastics, the same silicon, the same materials in general. I can't see the wisdom of saying that 75C is high for a typical CPU, but over 100C is fine for the SB. There is simply nothing to lose by running cooler, and $30 worth of extra heat sinks is too cheap for me to argue for the insurance it will provide. And like I said, I don't need to push that curve. $30 is nearly nothing to me compared to my time / effort / expense for a MOBO replacement. Comparatively, $15 (half retail), added to the board price by Intel, would completely screw up the market price point of the product, so they have every reason to use their superior engineering and play MTBF percentage games and cut everything as close to the wire as they can. It's how they make the big bucks. I prefer to improve on their margins when I can do it so easily.

HFat
Posts: 1753
Joined: Thu Jul 03, 2008 4:27 am
Location: Switzerland

Post by HFat » Mon Feb 15, 2010 1:13 am

crisspy wrote:ALL these chips use the same plastics, the same silicon, the same materials in general. I can't see the wisdom of saying that 75C is high for a typical CPU, but over 100C is fine for the SB.
Many CPUs have a maximum temperature under 75C. Mobile CPUs tend to have 95C or 100C. Do you figure these temps are BS? I'd have no issues with running a mobile CPU at 75C if that did not heat the other components. They often run quite hot when people stress their laptop. Have you seen many dead laptop CPUs? Laptop boards tend to be a whole lot more fragile for instance.
Now, Intel says the maximum temp for the SB is well over 100C and my guess is that it's too small to heat up the whole board which I think is real concern here. The RAM I'm using is not rated to run these kinds of temperatures for instance.
crisspy wrote:$30 is nearly nothing to me compared to my time / effort / expense for a MOBO replacement.
You also need to figure the time/effort to select and install the stuff. In any case, that's actually mighty expensive as insurance against an unquantified risk (compare with other types of insurance). I figure you'd need to lower the yearly failure rate by something like 15% to justify it. I doubt it's that high to begin with and you're certainly not going to drop it to zero.
We'll be in a better position to figure this out once D510MOs start failing in the wild.

HFat
Posts: 1753
Joined: Thu Jul 03, 2008 4:27 am
Location: Switzerland

Post by HFat » Mon Feb 15, 2010 11:00 am

croddie wrote:This is not a board you'd want to connect a display to except for troubleshooting
I connected my test D510MO to a nice Samsung 1680x1050 monitor and, while I could tell there were issues with the analog connection by using test patterns, regular images and text were displayed without any flaw I could perceive. I haven't tried videos. I have seen genuinely bad analog outputs and, at least paired the Morex PSU I'm using, I think it's fair to say this board's is decent. I'll try different screens and higher resolutions some time.

For all you MS-heads out there, I also tried the Windows 7 RC which runs pretty smoothly without Aero (I didn't try it). The NIC is recognized out of the box so there's no need to futz around when installing. Intel has Windows video drivers and stuff on their website.
By the way, since Linux support for the hardware is rather bleeding-edge right now, this might also be of interest to people for whom MS isn't necessarily the first choice. I did not test VM performance with a Windows host yet though.

What software do y'all use for temperature monitoring on Windows desktops/servers?
EDIT: it would be helpful if someone could answer this because I seem to be getting slightly lower temperatures in Windows (with Intel's drivers) and I'd like to verify that.

Jay_S
*Lifetime Patron*
Posts: 715
Joined: Fri Feb 10, 2006 2:50 pm
Location: Milwaukee, WI

Post by Jay_S » Fri Feb 19, 2010 11:03 am

Just bought one of these boards along with 1GB G.Skill PC2-5300. I immediately flashed it to the latest 0154 bios. Noticed mem voltage defaulted to 1.84V. I lowered it to 1.8V and the board wouldn't post. Had to set the "config" jumper to even get into system setup.

Anyone else unable to runV-mem @ 1.8V? I didn't check mem voltage prior to flashing to 0154...

Currently waiting for my picoPSU-80 and 60W brick to arrive. I'm temporarily using an old Zalman 300W PSU. With a 2.5" Scorpio Blue and a 80mm 0.09A fan my Kill-a-Watt reports 32W idle. I'm going to guess at the Zalman's efficiency ... at this load, 65%? 70%? Maybe less?

MoJo
Posts: 773
Joined: Mon Jan 13, 2003 9:20 am
Location: UK

Post by MoJo » Fri Feb 19, 2010 11:13 am

HFat wrote:What software do y'all use for temperature monitoring on Windows desktops/servers?
EDIT: it would be helpful if someone could answer this because I seem to be getting slightly lower temperatures in Windows (with Intel's drivers) and I'd like to verify that.
The Intel monitoring software is usually definitive. Temps in Windows will usually be lower than the BIOS due to power saving features. The BIOS does not issue enhanced HALT instructions or power down the PCI-E bus when not in use.

CoreTemp is also pretty good. Not sure if the Atom supports it but I would expect it to.

As for video quality, I don't think Intel are really aiming this board at HTPC users. It seems more oriented towards Windows Home Server or low end workstations and RDP clients. That means either lowish resolutions or no video output at all.

It does seem like an odd decision to only have an analogue output when there could be a combined digital/analogue DVI connector, but perhaps the board is designed to be very low cost. It doesn't even have parallel or serial port headers which is usually a sign of extreme cheapness. It's a pain for me because I want to use the serial ports and most small cases don't have any PCI slots where a header could be mounted.

HFat
Posts: 1753
Joined: Thu Jul 03, 2008 4:27 am
Location: Switzerland

Post by HFat » Sat Feb 20, 2010 2:33 am

MoJo wrote:The Intel monitoring software is usually definitive.
What's the name? Or where can I get it?
MoJo wrote:CoreTemp is also pretty good.
It looks like this would only show CPU temperature. Or am I mistaken?
MoJo wrote:It does seem like an odd decision to only have an analogue output when there could be a combined digital/analogue DVI connector, but perhaps the board is designed to be very low cost.
It looks like there's some kind of shortage. It was only in stock for a few days around here and retail prices are being jacked up so it's not so low-cost right now from my perspective. It's more of a high-profit board. ;-)
It looks very much as if Intel crippled the board in order to enforce market segmentation. Maybe they're trying to nick Ion in the bud as well. In any case, it would certainly have cost very little money to make this board suitable for more uses but I don't think the monopolists at Intel are going to unleash Atom's potential any time soon.
As it is, it's already a boon for silent computing so I find it strange that this board isn't getting more attention here. Sure, you might have the skills and the budget to build or to buy a better silent PC for yourself... but what about regular folks who are annoyed by the noise their PC makes? Most people don't need better performance than this board can provide.

crisspy
Friend of SPCR
Posts: 228
Joined: Sun Sep 29, 2002 9:05 pm
Location: Powell River, BC, Canada

Post by crisspy » Sat Feb 20, 2010 3:43 am

HFat wrote:It looks very much as if Intel crippled the board in order to enforce market segmentation.
...and...
... I don't think the monopolists at Intel are going to unleash Atom's potential any time soon.
I agree wholeheartedly. I think they released this early and limited to appease the hard core / hungry reviewers / reference platform market needs, but don't want to compete with other MOBO companies that they want to have building and selling new fancy ATOM boards in real bulk instead of ION. After all, Intel is first and foremost a chip company. Expect some of the Asus, Gigabyte, MSI versions to come fully loaded, with extra things like the Broadcom and Wifi on board, plus DVI/HDMI/DisplayPort/whatever. That and the industrials / verticals, who will build out specialties such as many serial ports, many LAN, 12v only, LVDS, weird form factors, etc., and who need some very clear market space to justify moving to new Atom from the old Atom, considering the small markets they serve and the cost of re-development they have to swallow to play at all.
HFat wrote:As it is, it's already a boon for silent computing so I find it strange that this board isn't getting more attention here. Sure, you might have the skills and the budget to build or to buy a better silent PC for yourself... but what about regular folks who are annoyed by the noise their PC makes? Most people don't need better performance than this board can provide.
The problem is that for an extra $100, you can get a 330 + Ion that plays media and actually appeals to the bulk of the enthusiast market who might want HTPC ability (even in principle), or at least a bloody DVI at minimum. And the normal folks, well, The Boss buys Dell or HP for wee office appliances, normal schmoes buy at Best Buy or Staples or even Mall Wort. 50% don't know a damn thing about computers, except it's got a power button, their ears don't BLEED any more (most computer are much quieter than a few years back), and doesn't their new monitor use a DVI or HDMI thingy, not that old blue thingy?

Face it, we ultra low power / noise at any cost freaks are, well, actually kinda... uhhhmmm... freaks. Everyone else will just wait till real boards or systems come out. Intel threaded a fine needle on this one, lucky it works for my weirdo purpose, so I get to be kinda happy sooner than most.

HFat
Posts: 1753
Joined: Thu Jul 03, 2008 4:27 am
Location: Switzerland

Post by HFat » Sat Feb 20, 2010 4:31 am

Well, the Intel board seems to be selling fast here, especially the bulk version. Meanwhile, the flashy Asus D510 board is in stock pretty much everywhere. It's got the legacy ports MoJo wants but it's more expensive. Maybe I should buy one and see how passive cooling works with that heatsink.

HFat
Posts: 1753
Joined: Thu Jul 03, 2008 4:27 am
Location: Switzerland

Post by HFat » Sat Feb 20, 2010 4:44 am

Jay_S wrote:Anyone else unable to runV-mem @ 1.8V? I didn't check mem voltage prior to flashing to 0154...
I was pretty sure I would have to play with the jumper afterwards if I tried but I figured: why not? And indeed it wouldn't work. There's something weird though: now I can't find the voltage setting anymore and it looks like manual settings are looked at 1.8V because manual settings won't work at all. Not that I tested them before...
Are you still able to change the voltage? It's right under the manual timer settings, right?

Now, an update about the thermal situation: while I was playing with the jumper I figured I'd try opening the case AND laying the board vertically... and I got much higher temperatures than when I laid the board horizontally. So it looks like convection actually works best at cooling this thing when it's horizontal as I originally suspected but that the meshed cover of the Morex 3310 is enough to hamper convection so much that vertical positioning is an improvement.
I didn't unmount the hard drive so the test is not totally fair but the board is only about 2C cooler than it was with the case cover rather than a hard drive obstructing airflow. It makes some sense because opening the case doesn't get rid of the PSU, of the power cables and of the side mesh which are above the board when it's vertical but still: compare to the dramatic difference when it was horizontal. I'd say it's weird again but the problem is evidently with my broken thermal intuition.
Last edited by HFat on Sat Feb 20, 2010 5:18 am, edited 2 times in total.

HFat
Posts: 1753
Joined: Thu Jul 03, 2008 4:27 am
Location: Switzerland

Post by HFat » Sat Feb 20, 2010 5:10 am

oops!

Jay_S
*Lifetime Patron*
Posts: 715
Joined: Fri Feb 10, 2006 2:50 pm
Location: Milwaukee, WI

Post by Jay_S » Sat Feb 20, 2010 9:46 am

HFat wrote:I was pretty sure I would have to play with the jumper afterwards if I tried but I figured: why not? And indeed it wouldn't work. There's something weird though: now I can't find the voltage setting anymore and it looks like manual settings are looked at 1.8V because manual settings won't work at all. Not that I tested them before...
Are you still able to change the voltage? It's right under the manual timer settings, right?
In my case it was a little stranger than that. I didn't initially suspect V-mem. I had adjusted a whole bunch of settings at the same time, and I thought 1.8V ram should be fine at 1.8V (silly me!). To fix all this, I pulled the jumper (boots into recovery mode) and re-flashed with 0154 again. Then I installed the jumper in 'config' mode (pins 2 and 3) and reset factory defaults. Finally, rebooted with the jumper on 'normal' mode.

I don't remember if the V-mem setting was present after restoring factory defaults.

I currently have this in a Norco RPC-230 rack case with a single scythe 80mm 0.9A fan aimed at it, plugged into the D510MO's fan header. I initially had the D510MO dynamically adjusting the fan's speed, but it's too eager to go to 100% upon boot & wake. So I fixed it @ 50% (as low as it goes). I'm using ubuntu 9.10; using lm-sensors CPU temp has never gone above than 30C. It's been at 28C for the past hour or so. This board is very easy to cool, even considering the RPC-230 has possibly the stupidest fan arrangement ever (will be modded soon - I'll post a gallery thread about this in about a week).

So far, I'm loving this board. Ubuntu 9.10 works perfectly out of the box (still loads the r8169 module by default - which works OK with the 8111 nic - I replaced it with the proper r8168 anyway). I'm using this as a music server, and installed Ubuntu a-la-carte style from the mini.iso adding packages as necessary. I didn't realise the Intel Atoms don't have any frequency scaling - the D510 is fixed at 1.66GHz. Took me a while to figure this out too ("why isn't cpufreq / powerenowd / etc working") - hardware doesn't support it :)!

HFat
Posts: 1753
Joined: Thu Jul 03, 2008 4:27 am
Location: Switzerland

Post by HFat » Sat Feb 20, 2010 10:47 am

Jay_S wrote:Ubuntu 9.10 works perfectly out of the box
There's more drivers in the upcoming kernel though it might not matter for your purposes.
Jay_S wrote:I didn't realise the Intel Atoms don't have any frequency scaling - the D510 is fixed at 1.66GHz.
With 2.6.33, cpufreq seemed to be able lower speeds but I don't know if that did anything. The cores don't use much power anyway.

MoJo
Posts: 773
Joined: Mon Jan 13, 2003 9:20 am
Location: UK

Post by MoJo » Sat Feb 20, 2010 11:04 am

HFat wrote:
MoJo wrote:The Intel monitoring software is usually definitive.
What's the name? Or where can I get it?
It's called Intel Desktop Utilities. If the board supports it you can find it by going to the Intel web site and searching the downloads for that mobo.
MoJo wrote:CoreTemp is also pretty good.
It looks like this would only show CPU temperature. Or am I mistaken?
No, that's right, only CPU.
As it is, it's already a boon for silent computing so I find it strange that this board isn't getting more attention here. Sure, you might have the skills and the budget to build or to buy a better silent PC for yourself... but what about regular folks who are annoyed by the noise their PC makes? Most people don't need better performance than this board can provide.
I agree. It has more than enough CPU power for most tasks, and being dual core will also be quite responsive. The on-board graphics, while basic, do offer plenty of hardware acceleration for web browsing and the like in Windows 7. I'm not talking about Aero performance, I mean smooth scrolling and video in Firefox etc. People don't seem to realise that now Windows composites everything with the GPU and can accelerate most stuff if the hardware supports it newer GPUs make a significant difference even in office apps.

HFat
Posts: 1753
Joined: Thu Jul 03, 2008 4:27 am
Location: Switzerland

Post by HFat » Sat Feb 20, 2010 12:02 pm

MoJo wrote:It's called Intel Desktop Utilities. If the board supports it you can find it by going to the Intel web site and searching the downloads for that mobo.
It's on the D510MO page but the program doesn't display temperatures.

Is there really no easy way to get this information from within Windows? I doubt it but still: how do people figure Windows is user-friendly?

MoJo
Posts: 773
Joined: Mon Jan 13, 2003 9:20 am
Location: UK

Post by MoJo » Sat Feb 20, 2010 4:45 pm

Strange, it shows temperatures for the CPU, NB and one other mobo sensor on my system.

Jay_S
*Lifetime Patron*
Posts: 715
Joined: Fri Feb 10, 2006 2:50 pm
Location: Milwaukee, WI

Post by Jay_S » Mon Mar 01, 2010 7:27 am

Jay_S wrote: ...I'll post a gallery thread about this in about a week.
viewtopic.php?t=57818

namaku0
Posts: 5
Joined: Wed Mar 10, 2010 7:50 pm
Location: Indonesia

Post by namaku0 » Wed Mar 10, 2010 7:58 pm

If I have a PCI display adapter, is it possible to use dual-monitor with
this board (using both the integrated and the add-on display adapter)?

MoJo
Posts: 773
Joined: Mon Jan 13, 2003 9:20 am
Location: UK

Post by MoJo » Thu Mar 11, 2010 8:43 am

Should be possible, yes. Check the BIOS, see if there is a setting for Primary Graphics Adaptor or a way to force the on-board video to remain on even with a PCI video card plugged in.

I have one of these boards now so can probably answer your question for certain next month if you want to PM me.

Jay_S
*Lifetime Patron*
Posts: 715
Joined: Fri Feb 10, 2006 2:50 pm
Location: Milwaukee, WI

Post by Jay_S » Thu Mar 11, 2010 2:29 pm

Any one running the latest (March 8 2010) BIOS rev 0175? Intel download link. Any issues?

namaku0
Posts: 5
Joined: Wed Mar 10, 2010 7:50 pm
Location: Indonesia

Post by namaku0 » Fri Mar 12, 2010 12:40 am

MoJo wrote:Should be possible, yes. Check the BIOS, see if there is a setting for Primary Graphics Adaptor or a way to force the on-board video to remain on even with a PCI video card plugged in.

I have one of these boards now so can probably answer your question for certain next month if you want to PM me.
Thanks for the reply MoJo.

Based on my little experience adding another video adapter to a
mainboard (AGP or PCI-E based (non SLI/CrossFire)) that already
has an on-board video; my understanding is that the one that will
be used is depends on the BIOS setting which one to use or take
priority. The other video adapter on the same bus will be disabled.

But I have never had the chance to try to mix it with a PCI (not
PCI-E) video adapter, so I don't know and asked the question. I
am affraid if I must spent my money on this board just to know
it doesn't work. Especially after reading page 6 of Intel D510MO's
BIOS glossary PDF (biosglossarybymenu_vera02.pdf, I can't post
link yet)

MoJo
Posts: 773
Joined: Mon Jan 13, 2003 9:20 am
Location: UK

Post by MoJo » Sat Mar 13, 2010 6:14 am

namaku0 wrote:Based on my little experience adding another video adapter to a
mainboard (AGP or PCI-E based (non SLI/CrossFire)) that already
has an on-board video; my understanding is that the one that will
be used is depends on the BIOS setting which one to use or take
priority. The other video adapter on the same bus will be disabled.
That is often the default behaviour as it is the least confusing for people when they install a graphics card. However, you can usually override it in the BIOS on modern boards. The exact settings can be a bit cryptic, but generally speaking if the IGP is set to "auto" it will turn off if a card is installed so you need to set it to "enabled".

I'm afraid the manual is rubbish and does not say if it is possible or not. I'll let you know next month when mine is set up.

namaku0
Posts: 5
Joined: Wed Mar 10, 2010 7:50 pm
Location: Indonesia

Post by namaku0 » Mon Mar 15, 2010 9:03 am

MoJo wrote:
namaku0 wrote:Based on my little experience adding another video adapter to a
mainboard (AGP or PCI-E based (non SLI/CrossFire)) that already
has an on-board video; my understanding is that the one that will
be used is depends on the BIOS setting which one to use or take
priority. The other video adapter on the same bus will be disabled.
That is often the default behaviour as it is the least confusing for people when they install a graphics card. However, you can usually override it in the BIOS on modern boards. The exact settings can be a bit cryptic, but generally speaking if the IGP is set to "auto" it will turn off if a card is installed so you need to set it to "enabled".

I'm afraid the manual is rubbish and does not say if it is possible or not. I'll let you know next month when mine is set up.
Thanks, I'll wait for the result here. :o

DonOregano
Posts: 1
Joined: Mon Apr 12, 2010 10:31 pm
Location: Sweden

Post by DonOregano » Mon Apr 12, 2010 10:37 pm

Has anyone tried the Morex 3320 case?
What are the differences compared to the 3310?

The only info I've found is on the mini-itx website.
Note: case includes larger venting panels than the original Thin Client case to assist fanless operation - any CPU fans attached to boards are still required however.
Which leads me to believe that it is exactly like the 3310, but with a little more ventilation.

Post Reply