Thread: Don't You Hate It When A Plan Doesn't Come Together?

  1. #1
    Registered /usr
    Join Date
    Aug 2001
    Location
    Newport, South Wales, UK
    Posts
    1,273

    Unhappy Don't You Hate It When A Plan Doesn't Come Together?

    Hello,

    Well I've spent a good few hours toiling away at this problem and have reached the point where I'm calling it a day (it's 2:58 AM where I am at the moment)...

    A few months ago I had an idea to build a mini-ITX based PC that featured a single PCI slot. I have a few cards lying around that I would like to use from time to time, so having a small box to plug them into instead of having to disconnect and haul open the larger tower would be a godsend.

    I also "believed" that if I could get one of the last PCI graphics cards to be produced (in this case, a PowerColor Radeon HD 2400 Pro) I could also get it to play back Blu-Ray titles (this budget card doesn't perform H.264 acceleration but does have HDCP). This would mean that it could also serve as a HTPC also.

    Of course, I wasn't exactly awash with cash at the time, so I just made a shopping list and saved up. Which brings us nicely to today...

    I currently have sat in front of me a Intel D945GSEJT mobo featuring the Atom N270 laptop CPU and 945GSE mobile chipset. It has DVI, SATA and gigabit LAN onboard. It also has a PCI slot. And is completely fanless.

    Acting much the same way as I usually do on Christmas Day, I eagerly tore open the pack containing a 2GB SO-DIMM and slotted the memory in, wired up the case connections and plugged in the 75W external brick power supply. So far so good. I plugged in a monitor and USB keyboard and hit the power button. It's alive!

    After having a flick through the famously sparse BIOS menus, I then turned the PC off and whipped out the PowerColor. I carefully pressed the card into the PCI slot, changed the monitor connection to the card and flipped the power again...

    This time, nothing.

    I turned it back off, removed the card and tried again. Works fine?

    I tested the card in my main tower. There was some havoc when Windows XP loaded and found a really slow graphics card in addition to the PCI Express one, but other than that it worked fine.

    I scratched my head. The mobo works, the graphics card works... but why not together?

    Thinking the board may have a dodgy PCI slot, I tried inserting a video capture card. Booted fine.

    Then I remembered that there is an option in the BIOS to select which graphics device to use when booting... but this was missing from this particular BIOS(!)

    I frantically tried updating the BIOS to the latest version... no dice.

    It seems that the mobo won't boot with the PowerColor installed because Intel have kindly hardwired the BIOS to initialize the onboard graphics first, then it has a look at the PCI slot, tries to initialise the device and locks up because it was not expecting another graphics device to be there.

    I know that this is something that Intel have done because a similar mobo with the same chipset but the manufacturer's own BIOS (the MSI IM-945GSE) indicates that you can change the primary video adapter to PCI in the manual. It's just a shame that that board is not widely available (and costs twice as much).

    Now, you might be thinking "Don't be sad, cos 1 out of 2 ain't bad", but I just find it really annoying that people still go around crippling things for no good reason and spoiling what would otherwise have been a pretty decent little box.

    So, for the record then: you cannot use a PCI graphics card with the D945GSEJT. I hope that this info helps anyone else who had a similarly bright idea, and maybe prods Intel into doing the decent thing and repealing this senseless edict with an update.

    Now I have a £30 PCI paper weight. Let's hope that someone else out there is using a different board and would like the convenience!

    Right, rant over. I'm off to bed.

  2. #2
    Malum in se abachler's Avatar
    Join Date
    Apr 2007
    Posts
    3,195
    Well, in intel's defence most commercial applications of ITX systems sort of don't use add on graphics and in fact rarely use the PCI slot at all. The point of these systems isn't to be a small form factor desktop system, but to be a small form factor single board computer for use in space constrained systems. They also didn't cripple it, as the functionality you mention must be explicitly added to the BIOS in the first place. I highly doubt they added it only to remove it.

    Oh and welcome to the reality of capitalism, products with more functionality cost more to buy, even if they cost less to produce.
    Last edited by abachler; 11-14-2009 at 09:28 PM.

  3. #3
    Registered /usr
    Join Date
    Aug 2001
    Location
    Newport, South Wales, UK
    Posts
    1,273
    But Intel's BIOSes are largely unified, the previous model in the line (D945GCLF2) having the option available. The only difference between this board and that one is that this has the mobile version of that chipset. I bought it based on the logical conclusion that if it looks similar to the previous model then it should have the same level of hardware support.

    I can understand that because this is effectively a laptop without the screen and keyboard, the BIOS may as well be the same as if it were one. But this was sold to me on the basis that it has a PCI slot which you can use any 32-bit PCI card in, which is false.

    This would have been avoided if Intel had put a note in the manual in the PCI slot section indicating that "PCI graphics cards are not supported". But I guess sinking £100+ on hardware to suck it and see is all part of the learning process, eh?

  4. #4
    Malum in se abachler's Avatar
    Join Date
    Apr 2007
    Posts
    3,195
    Well, again, these boards are NOT essentially a laptop, they are for space constrained applications, generally industrial or mobile (not laptop) applications. Specifically applications that do not generally require any video whatsoever once deployed, and only need to support it for installation, field service or troubleshooting. I've also seen them used in cash registers, and other POS equipment. The mobile chipset may not support switching the video. I am pretty certain that whatever the reason it is technical, not marketing.

    Have you tried calling their tech support, there may be a workaround.

  5. #5
    spurious conceit MK27's Avatar
    Join Date
    Jul 2008
    Location
    segmentation fault
    Posts
    8,300
    Quote Originally Posted by abachler View Post
    Have you tried calling their tech support, there may be a workaround.
    Or at least try and trade it in to who you bought it from. If you can't do that, you bought it from the WRONG place. I've ordered parts before, gotten them shipped, then realized they're wrong and sent them back in the package for a refund or exchange. Generally, that is how it is supposed to work...esp. since it doesn't sound like you got a low low price or anything.
    C programming resources:
    GNU C Function and Macro Index -- glibc reference manual
    The C Book -- nice online learner guide
    Current ISO draft standard
    CCAN -- new CPAN like open source library repository
    3 (different) GNU debugger tutorials: #1 -- #2 -- #3
    cpwiki -- our wiki on sourceforge

  6. #6
    Malum in se abachler's Avatar
    Join Date
    Apr 2007
    Posts
    3,195
    Actually I just got off the phone with a friend of mine at Intel, she said you are probably overloading the power supply.

  7. #7
    Registered /usr
    Join Date
    Aug 2001
    Location
    Newport, South Wales, UK
    Posts
    1,273
    Although I can send it back, that doesn't leave me any better off. I can't buy a board that does exactly what I want it to do, seemingly...

    Quote Originally Posted by abachler View Post
    Actually I just got off the phone with a friend of mine at Intel, she said you are probably overloading the power supply.
    So... 75W brick pushes 12V with maximum current 6.25A.
    At startup the other chips on the board consume about 20W (no drives currently attached).
    That means this card pulls a relatively huge amount of juice, even at startup.

    As ever, the manufacturer does not provide power information for the card. Can I test for this using a multimeter?

  8. #8
    Malum in se abachler's Avatar
    Join Date
    Apr 2007
    Posts
    3,195
    I suppose you could inline the ammeter with the power supply, or if you have an oscilloscope you can check the voltage level during startup. If the voltage drops, its a definite sign that the PSU is overloaded, the ammeter will also tell you what the actual current draw is, although it may not be accurate, as if the PSU is overloaded it may limit the amps to its safety limit. You really need to test the voltage level. You can use the voltmeter, but it wont give as accurate a picture as the o-scope.

  9. #9
    Registered User
    Join Date
    Dec 2006
    Location
    Canada
    Posts
    3,229
    That means this card pulls a relatively huge amount of juice, even at startup.
    If you look at the heatsink ... space heaters take a lot of power.

    Wouldn't it be easier to measure the current at the wall? Assuming efficiency is more or less constant.

  10. #10
    Registered User
    Join Date
    Dec 2006
    Location
    Canada
    Posts
    3,229
    BTW, I have been wanting to get an oscilloscope for a while now... but they cost quite a few arms and legs.

    There are programs that will turn the soundcard into low frequency (4.4kHz) oscilloscopes, but apparently they don't work for low frequency/DC stuff, because soundcards do AC coupling.

  11. #11
    Registered User
    Join Date
    Aug 2006
    Posts
    100
    Quote Originally Posted by cyberfish View Post
    BTW, I have been wanting to get an oscilloscope for a while now... but they cost quite a few arms and legs.

    There are programs that will turn the soundcard into low frequency (4.4kHz) oscilloscopes, but apparently they don't work for low frequency/DC stuff, because soundcards do AC coupling.
    Check out some of the USB ones...
    usb oscilloscope - Google Product Search

  12. #12
    Internet Superhero
    Join Date
    Sep 2006
    Location
    Denmark
    Posts
    964
    It's not the power supply, a v3.0 PCI-slot can supply a maximum of 25watts, assuming that the graphics card have no extra external power plugs connected and that your estimated 20 watts for the board itself is correct, there is no way the power supply is close to 75 watts.

    But why don't you check by plugging up a normal 300watt+ power supply rather than the tiny ITX one? Then you could be certain without shelling out for an oscilloscope.
    How I need a drink, alcoholic in nature, after the heavy lectures involving quantum mechanics.

  13. #13
    Malum in se abachler's Avatar
    Join Date
    Apr 2007
    Posts
    3,195
    Quote Originally Posted by Neo1 View Post
    It's not the power supply, a v3.0 PCI-slot can supply a maximum of 25watts,
    That is highly dependent upon the implementation. I haven't looked at the standard, but I'm pretty sure that is only a minimum, not maximum.

    And if the board is limited, ten it may be the card is drawing too much current for the PCI bus.

    Ultimately of course it could simply be that the card doesn't support add on graphics and their internal documentation simply doesn't mention it.

  14. #14
    Registered /usr
    Join Date
    Aug 2001
    Location
    Newport, South Wales, UK
    Posts
    1,273
    I've just put everything back after opening up the tower, disconnecting the 380W PSU from my main board, attaching the P4 connector to the mini-ITX board and shorting the power status on the ATX connector with the ubiquitous paperclip.

    Tried it without the card: board starts up fine, hardware monitoring shows more fluctuations in the rails than from the brick, but then I guess it is a different beast.

    Turned it off, plugged the card in, turned it back on: the power LED comes on but there's no one home.

    I know the card works because as I said before, I managed to squeeze it in alongside a PCI Express Radeon within my tower and Windows booted and detected it straightaway (not CrossFire capable, natch, but it would've been a laugh).

    The best explanation I have heard so far on the Internet is that because the board uses the AMI Aptio EFI firmware instead of a conventional BIOS that it somehow needs a driver to be able to use it as primary graphics.

    P.S. Manual says that the PCI slot is version 2.3 compliant. Not quite top of the shop.
    Last edited by SMurf; 11-16-2009 at 05:28 PM. Reason: Manual insight

Popular pages Recent additions subscribe to a feed

Similar Threads

  1. Problem
    By gtfoh143 in forum C++ Programming
    Replies: 7
    Last Post: 08-18-2009, 10:44 PM
  2. Program Plan
    By Programmer_P in forum C++ Programming
    Replies: 0
    Last Post: 05-11-2009, 01:42 AM
  3. Tabbed Windows with MDI?
    By willc0de4food in forum Windows Programming
    Replies: 25
    Last Post: 05-19-2005, 10:58 PM
  4. Pet Peeves
    By Srg Pepper in forum A Brief History of Cprogramming.com
    Replies: 29
    Last Post: 10-03-2002, 11:34 AM
  5. I hate string parsing with a passion
    By DavidP in forum A Brief History of Cprogramming.com
    Replies: 2
    Last Post: 03-19-2002, 07:30 PM