• Status: Solved
  • Priority: Medium
  • Security: Public
  • Views: 5606
  • Last Modified:

Mobo suggs: uATX, DDR2, IDE133, PCI-X


I have been tasked with building a computer for a customer and would really appreciate hearing recommendations, and good and bad experiences in selecting a new motherboard.  

Some background:

The current computer was a custom build about 2 years ago.  

The mobo is a K7SEM, being replaced because the keyboard controller failed, causing a *huge* volume of low priority inerrupts that are pegging Windows XP at 100% utilization sustained.  Here's a link to the vendor website:
http://www.ecsusa.com/products/k7sem.html   So far, I am unable to find a new replacement K7SEM, which introduces the opportunity to consider modern hardware ...

The mobo is a Micro-Atx form factor.  I don't know whether a different size board will fit properly within the current case, although it is a pretty tall tower.  Consider the uATX form factor a constraint.

The case is a full height big and tall box, designed to sit erect on the floor, not a lan party special that fits in a backpack.

The computer has a new Maxtor 6y250p0 hard drive, 7200 rpm, 8M onboard cache, 250G storage, interface is Ultra ATA 133 mhz.  This is a really nice disk that is going to stay.  RAID and SATA are nice to have built onto the mobo, but are not required.  This 250G IDE drive *MUST* be supported well, and strong preference at it's maximum speed.

There are two IDE cd drives, one of type cd burner, the other is a dvd-rom drive.

The rating on the power supply is currently unknown.

The prime function of the computer is to translate speech into text using Dragon Naturally Speaking, which inserts the text into MSFT Word.  This person dictates alot, and produces prodigious compelling verbiage for a living.  Doing text to speech conversion well (read: fast) is a key factor.  The current lag time is about 5-30 seconds on a 1.6ghz processor (, that is unfortunately pegged by a continuous and inescapable onslaught of deferred procedure calls!).  An impressive result would be near realtime conversion!

During normal text to speech operation, the overall volume of data movement will be relatively low, except for OS startup and program launching.  This will not be a database machine that is being continually hammered by a webserver farm 7x24.  On the contrary, the machine will be a deskside workstation, used interactively.  But when it is used, the text to speech process will consume the processor 100% sustained for the duration of the document.  

OS = XP Professional.

High performance graphics is not a requirement, or even particularly interesting.  This computer will probably never be used to play games like Doom or Half Life, but will spend it's life within the bounds of MSFT Windows XP Professional.  However, the basic nature of the design for a gamer machine might be an extremely good match for doing near realtime text to speech conversion well.  The presence of killer display engine would not be deal breaker by any means.

There must be a parallel port, ps2 style keyboard and rodentia, a(n onboard) nic, usb2 (there are 4 usb devices to be plugged in, but a usb hub could be gotten). There must be at least 1 expansion slot, to hold a modem card.  But the ability to expand is a nice thing.

I'm interested in DDR2, but not married to the idea.  Am I barking up this tree just because it's new, or is DDR2 really that much faster than DDR memory?  Note: reasonable cost is an issue.  I'd rather sink the cash into a great mobo, then processor, then memory.  But I understand the performance constraint that memory bandwidth has, so the ability to support really fast memory is a key factor.  Presume the cost of memory will not exceed $200, for at least .5G.  

The ability to support > 1G of memory is a good thing, but not a requirement.  However, based on the general historical trend that things tend to get bigger not smaller, and more and bigger are better, the ability to support 2G+ would be nice.

This cpu should be biased toward excelling at computation and data movement.  Multiple processors would be a waste.  The nature of text to speech is unthreaded, so it will run best on a single processor that is fast and has super memory bandwidth.  I don't think Dragon is like PhotoShop, which is written to be threaded, and will not take nice advantage of multiple processors.  Think single processor for this one.

The processor type is open.  Presume a max price to be in the $200 ballpark.

The motherboard has to be a uATX form factor.

The processor will need to be something Intel compatible.  How much does this rule out cool low power technology like Efficeon?

Goals, hopes, and dreams:

Tell me what mobo to get!  I've been doing research on this for a while now, and there are some really goor/ great ones out there.  But I have come to the inescapable conclusion that others might also have encountered this kind of project before, and might have a thing or two to say on this.

Comments, suggestions, guidance, red flag warnings please!

  • 28
  • 9
  • 7
  • +2
12 Solutions
Right now, when FSB is 800 MHz, dual channel DDR2 is overkill.

If you still want motherboards with i9x5 chipset, you should know, thay they usually support only one IDE channel (while you need to connect 3 devices - 2 IDE channels necessary). Some motherboards have additional IDE controlers (like Asus P5QD1, mentioned by Nobus). There are 2 problems with them:
  1. I was unable to find i915/i925 motherboard with additional IDE controler and micro-ATX form factor (all of them were ATX)
  2. At least Asus recommends to connect CD-ROM to primary IDE controller, and HDD - to additional. So you will need to provide additional drivers during OS instalation (press F6). You won't be able to do that without FDD :)

The only microATX i915 motherboard is Intel BLKD915GAGL(ddr)/BLKD915GUXL(ddr2), but it supports only 2 IDE devices.

I was unable to find micro-ATX motherboards with i875 chipset (that would be my choice).

Now you need to choose between:
* uATX or ATX (so you can use motherboard with i915 chipset)
* replace IDE HDD with SATA (so you can choose Intel i915 motherboard; i suggest choosing DDR, not DDR2 now)

If you still need 3 IDE devices and need uATX form factor, you need to look at i865 motherboards.
Possible choices:

You can also look at Intel, Abit motherboards (that is true for all motherboards - choose between Asus, Intel, Abit).

If you've got a full tower case, I wouldn't limit yourself to the uATX form factor.  It should also accept a standard ATX form factor, so you'll have more features to choose from and better cooling.

I concur that DD2 isn't necessary right now.  The best bang for your buck are the 800MHz Intel procs with 512MB - 1GB DDR.

Here is what I recommend:

Motherboard:  Asus P4C800-E Deluxe
$176 from newegg.com
It supports 6 IDE devices, has 4 USB ports (expandable to 8), gigabit ethernet, and supports up to 4GB DDR400 PC3200 memory.  It also has onboard sound, firewire, SATA, RAID, and a Wi-Fi slot.

Processor:  Pentium 4 3.0E-800
$189 from newegg.com
A good fast processor with 1MB L2 cache.  One caveat:  these processors are known for running a bit warm, so be sure that the case is properly ventilated.  If there is any question or if you plan to stress the processor or overclock, I'd suggest buying the OEM version of this proc and installing a high quality fan with an oversized copper heatsink.

Memory:  Mushkin Dual Channel Kit (1GB PC3200 DDR400)
$203 from newegg.com
Good quality, FAST memory.

Total price:  $568 + $1.89 shipping, plus newegg will throw in a free pen, watch and a beachtowel.  Awesome!
Protect Your Employees from Wi-Fi Threats

As Wi-Fi growth and popularity continues to climb, not everyone understands the risks that come with connecting to public Wi-Fi or even offering Wi-Fi to employees, visitors and guests. Download the resource kit to make sure your safe wherever business takes you!

ChrisEddyAuthor Commented:

Awsome smart responses!  Thank you!!

Ack on DDR2 being overkill.  Thank you for preventing me from rathole-ing down that path!

On the Intel P5GD1, this is one nice board!  Modern processor socket, high end onboard audio, 800 Mhz FSB, sockets for 4 memory sticks.  One down side: it's an ATX form factor.   Although it won't support all four current devices (IDE disk, ATA cd burner, ATA DVD-rom, floppy), a solution would be to get a single CD+DVD combo drive which could become SATA.  The specs don't list a floppy controller, but it does list a floppy cable, and I think I see a floppy connector in the (small!) picture toward the front (bottom).  

On the Asus P4C800-E Deluxe, it's also a sweet board, but also an ATX board.  I don't yet know whether ATX and uATX are (reasonably) symmetrical.  Your thoughts???

One of the things I'm seeing in the marketplace is that, on average, ATX boards tend to run +$100 over uATX boards.  Not sure how apples:apples the comparison is yet.  !equivalence = price disparity.  I will check on this further and report back findings.

So far, I've got a load of data to slog through, so I'm in the process of making a spreadsheet to compare and contrast the different mobos vs features.  I'm out to the W column and I think I can easily double that.  (this has the makings for a Toms Hardware article ...)

Thank you for sending actual links + opinions!  They really help navigate though the data.  Thank you!!

Question: what are the salient differences between the Intel 9XX series of controller chips vs the 8XX series.  

Question: How can I tell whether the mounting holes for the ATX board will line up with the uATX holes?  If the uATX vs ATX mounting holes are a non-issue, then yes this does open up opportunities for more choices.  

Question: What's this South and North bridge business?

P5GD1 is made by Asus, not Intel :)  Yes, it has support for "legacy" FDD.

Mounting holes for ATX and uATX not all are the same, but if computer case is "generic", there should be enough mounting holes for all format motherboards. IMHO unless you have Small Tower or less, mounting is not an issue.

For full format motherboards, there are more PCI ports, more integrated devices, more complex wiring and so on. AFAIK if you need basic functionality (utilize the same functions of ATX and uATX) - there is no performance difference. Since you don't need a lot of PCI cards, there is no difference between ATX and uATX boards (BTW, look at integrated audio - there might be differences).

BTW, if you will choose i865 ATX motherboard, i suggest to choose Asus P4P800 series:

About new things in i9x5, you can read at http://www.digit-life.com/articles2/roundupmobo/i915-i925-chipsets.html
From user's point of view - new CPU socket, new video card slot, better SATA support (with more RAID options), less PATA controllers, new PCI-X slots.
ChrisEddyAuthor Commented:

Thank your for your guidance and support!  I have reviewed everything you have provided, and then some...  

I have pulled together information from many sources, made a spreadsheet of AE columns and 20 rows (mae west: too much of a good thing is wonderful!).  I assigned values to the different attributes (uATX?, LGA775?, Chipset=9xx series?, Onboard good quality audio?, S/PDIF?, Onboard Video?, USB2?, 1394?, Serial and Parallel ports?, Sata and Pata connections?, PCIx x16?, AGP?) to assist in scoring the motherboards, as a method of slogging through an otherwise gorgeous collection.  Btw: Please comment on whether these attributes are reasonable.

These are the two that came up on top:  Intel D915GAG, and Asus P5GD1-VM.

1) Intel  D915GAG

2) Asus P5GD1-VM

Cost of Intel D915GAG = $121 avg:
   $117 NewEgg
   $123 ComputerBrain
   $123 eXcaliberPX.com

Cost of Asus P5GD1-VM = $118 avg
    $105 ZipZoomFly.com
    $109 Ajump.com
    $139 Monarchcomputer.com

Memory constraints for D915GAG mobo (which I assume also constrain the Asus board):

Nutshell (cut and pase from INTC website):
       2.5 V (only) DDR SDRAM DIMMs with gold-plated contacts
      Unbuffered, single-sided or double-sided DIMMs with the following restriction:
           --> Double-sided DIMMs with x16 organization are not supported.
      4 GB maximum total system memory
      Minimum total system memory: 128 MB
      Non-ECC DIMMs
      Serial Presence Detect
      DDR2 400 MHz and DDR 333 MHz SDRAM DIMMs

Memory: .5G ddr400
   $100 Crucial


  P4 3.0Ghz 800fsb L2=1mb = $185 avg
    $183 @NewEgg
    $186 @ComputerBrain

  P4 3.2Ghz 800fsb L2=1mb = $220 avg
     $215 @GameVE.com
     $220 @ComputerBrain
     $223 @NewEgg

Average cost range for this project:
     $440 P4 3.2Ghz
     $405 P4 3.0Ghz

My big concern: insufficient support for enough of the existing devices.  The 6Y250P0 ide disk *must* be supported as an IDE drive, and will rule out the usage of the mobo if it can't.  A floppy drive *should* be supported natively.  The CD burner and DVD-rom could be replaced using a SATA combo drive.  

Am I missing anything here ...

I have read some scary reviews about the new 915 chipsets.  Great promise but big bugs.  They may have the bugs out by now but hey why take that  chance?  I would be careful to not put a buggy machine on someones desk.  The 865 and 875 chipsets work well and are super stable.  If you need high end video it may be worth the risk but you said this is for heavy naturally speaking.  The 865/875 chipsets will take the latest prescott cpu's - the 2.8-3.2's are cheap at $175-$220.

I personally have seen really crappy asus boards- they performed well below their specs.  I was lucky to figure out they were the problem- it was kind of an educated guess.  After changing the sytsem board to an Intel - everything ran much faster.  I would stay away from them...

I agree that you should be able to put a smaller board in the case and that you will get a better selection and better value with an ATX.

Also you better get a good Powersupply from Enermax, Antec or the like. Don't rely on an idontknow brand.  If you put more than 1 HD I would get a 400w or higher.  I usually put to much rather than to little.  I don't like complaints though.  I would suggest the extra cost rather than have it screw up later.
ChrisEddyAuthor Commented:

I've finally dove into the technical documentation for the Intel D915GAG board, and now have some clarity on things that I was missing.

One of the missing things that I've recently realized is this board takes a 24 pin power connector, where the standard ATX power connection, and this mobo in particular, takes a 20 pin connector.  However, I downloaded the technical documentation in pdf, and it says:

"The board has 3 power connectors:

1) Main Power - a 2 x 12 connector.  This connector is compatible with 2 x 10 connectors previously used on Intel Desktop  boards.  The board supports the use of ATX12V power supplies with either 2 x 10 or 2 x 12 main power cables.  When using a power supply with a 2 x 10 main power cable, attach the cable on the right most pins of the main power connector, leaving pins 11, 12, 23,24 unconnected.

2) ATX12V power - a 2 x 2 connector.  This connector provides power directly to the processor voltage regulator and must always be used.  Failure to do so will prevent the board from booting.

3) Alternate power - a 1 x 4 connector.  This connector provides additional power when using high wattage PCI Express x16 graphics cards.  "

Sounds benign so far.  However, the next note (just below) suggests otherwise:

"Integrator's Note

When using high wattage PCI Express x16 graphics cards, use one of the following power supply connections to avoid system instability:

1) The preferred method of power delivery is to use a power supply with a 2 x 12 main power cable.  In this configuration, use two connectors to provide power to the board:
a) The main power connector
b) The ATX12V connector
In this configuration, the alternate power connector is not required.  The 2 x 12 main power cable can provide up to 144W of power from the +12V rail.

2) An alternate method of power delivery is to use a power supply has a 2 x 10 main power cable.  In this configuration, use three connectors to provide power to the board:
a) The main power connector
b) The ATX12V connector
c) The alternate power connector."

Another Integrators note:
"Use only ATX12V-compliant power supplies"

Power supplies:
  330 watt
     $54 Atacom  Antec True330

  350 watt
     $30 NewEgg   Foxconn Allied AL-A350ATX

  380 watt
     $67 NewEgg  Antec True380

  400 watt
     $30 NewEgg  Allied AL-A400ATX

  430 watt
      $70 NeuTronUSA.com Antec True430

Goofy question, but how will I know if these power supplies will fit in the case?  I don't have the computer with me so I can't take measurements.  On the Antec website, there is mention of a uATX power supply under "Specialty Power Supplies".  

Open question: How can I calculate the power consumption of the devices plus a fudge factor (pc speak: Margin of Safety), rather than go with a 400 watt power supply and being secure in the knowledge that that will definitely cover all now and future power needs.  Operating assumptions: D915GAG, .5G memory, PCIx x16 video, Maxtor 6Y250P0 disk, floppy drive, CD drive, DVD drive, 4 USB devices, overage to support +2 of the same hard drive.

There is mention of a slightly different board, called: D915GAV.  The difference noted is that the GAV is an ATX form factor board has 4 PCI connectors, 2 PCIx x1 connectors; the GAG is auATX form factor board with 2 PCI connectors, and 1 PCIx connector.  "The x1 interface supports simultaneous transfer speeds up to 500 MBps"  Both boards have a PCIx x16 "connector supporting simultaneous transfer speeds of up to 8GBps."   All those who feel the need for speed, run forward now!

According to the rechnical documentation, there are "Manufacturing Options".  
"Table 3 describes the options.  Not every manufacturing option is available in all marketing channels.  Please contact your Intel representative to determine which manufacturing options are available to you."
ATAPI CD-ROM Connector - a connector for attaching an internal CD-ROM drive to the inboard audio subsustem.
ATX fan connector - only on the GAV board
IEEE-1394a interface - what you'd expect
Intel 82801FBW I/O Controller Hub - hub support of Intel Wireless Connect Technology
SCSI Hard Drive Activity LED Connector - what you'd expect
Serial Port B - what you'd expect
S/PDIF Connector - a 1 x 3 connector for what you'd expect
Trusted Platform Module  - "A component that enhances platform security".  
Interesting options, but the first one was the one I was most hopeful about, but turns out to be an audio-only connection, not a data connection.  Sigh!

I have confirmed the presence of a "Diskette Drive Connector".   I have been on a slow boil on this for a while, but no longer.

Memory interleaving is another missing thing.  In a nutshell, two .25G sticks will be faster than one .5G stick.  The first pair of memory slots is for channel A, the second pair is for channel B.  If only one .5G memory stick is placed in the first channel A slot, then memory interleaving is disabled.  If .25G of memory is placed into the first Channel A slot and the first Channel B slot, then memory interleaving is enabled.  If the amount of memory is different between channels A and B, memory interleaving is disabled.  

Again from the technical documentation:

"Dual channel (Interleaved) mode.  This mode offers the highest throughput for real world applications.  Dual channel mode is enabled when the installed memory capacities of both DIMM channels are equal.  ..."

"Single channel (Asymmetric" mode.  This mode is equivalent to single channel bandwidth operation for real world operations.  This mode is used when only a single DIMM is installed, or the memory capacities are unequal. ..."

Another missing item is on the topic of power for SATA devices.  Again, from the technical documentation:
"Note.  Many Serial ATA drives use new low-voltage power connectors and require adaptors or power supplies with low-voltage power connectors.  For more information, see: www.serialata.org"
I can reuse the fast IDE disk drive, and the "legacy" floppy drive, but the functionality of a CD burner and DVD reader is currently present and should be supported - but apparently can only be done via SATA.  I read this as putting the last nail into the coffin of keeping the old power supply.  A new PS+CD/DVD gets added to the design now.

Open question: Any suggs on CD/DVD combo drives.  (multi format CD burning = required)

Power supply calculation :

x86fix, please specify, what Asus boards were "crap"? Asus makes boards with all flavors of chipsets (Intel, VIA, nVidia, SiS, ATI), also, some motherboards are designed as "low price" (..-X models).

ChrisEddy, you should use 2 memory modules (so you can benefit from dual channel mode).

What power supply you use now? (brand, watts).
You *can* use your old 2x10 pins PSU with new i9x5 motherboards as long you don't buy high end video card (you won't need that, since this computer is not for 3d games). IMHO 350W high quality PSU (by Antec, InWin and so on) will be sufficient.
Also, you *don't* need new PSU to get support of SATA devices. Power adapters costs 1-2USD and works fine.

IDE controller will support your 250 GB disk. Just make sure, that you enable LBA48 support in your OS.

Plextor DVD+-RW PX-712SA (i use non-SATA model - very good, though expensive)

SATA combo:
MSI XA52P (be sure to read conclusion)
ChrisEddyAuthor Commented:

Thank you again for your responses!

On what power supply is used now, I don't have the box physically with me right now, so I don't know.  A constraint is going to be whether the powersupply will physically fit.  Note again that this is a full size tower case, so I suspect that any PS will fit fine.   The harder question is what the size of the power supply should be.

Thank you for the link to the power supply calculator!  Because the D915GAG is a new board, almost none of the drop down choices applied directly, so I chose the item that showed it consumed the most power.  According to this calculator, the minimum power supply is 374watts.  

Remembering back to something I read in the Intel technical documentation, the power distribution rails in the mobo can carry up to 111 watts.  The most power hungry processor listed was 103 watts.  It's looking like the 400 watt supply is the reasonable minimum.  The down side is that both of these power supplies use a 20 pin connector, and the new D915GAG board has a 24 pin connector - but will accept a 20 pin connector, if you also use "the alternate power connector".  Looking at the pictures in the Intel technical document, it looks like the "alternate power connector" is a square 4 pin ATX12V connector.  

From the Toms Hardware article (dated 040709) on the MSI XA52P: "All in all, a product that we can recommend - as long as you use a motherboard with an Intel 865/875/848 chipset - the XA52P doesn't yet work with other SATA controllers, and so we have to sit tight for SATA II with ATAPI support.".  From the MSI website: Supports only Intel 875P / 865PE / 865G / 865GV / 865P / 848P Chipset motherboard".

However, research from another site (dated 040826) states otherwise.  Nutshell: the D915GAG board uses a the ICH6 chipset, which this site has found to "function correctly" with both the MSI XA52P and the Plextor PX-712SA.  The term "functions correctly" is defined as: "In order for these drives to be classified as a "Supported" on a particular controller, we had to be able to be able to see them correctly in the motherboard BIOS, had to be able to boot a disc outside of the operating system, and had to be fully accessible once inside Windows XP Professional.".  

A plus for using the MSI drive, albeit a small one, is that it uses the "older" style of 4 pin Molex power connector rather than the "new" Sata style of power connector..

Original source for the above referenced article:  


Combo Drives (CD burner, DVD reader, SATA interface):
    PX-712SA         MSI XA52P
     $139                                           NewEgg
     $147                                           Next International
     $150                                           CompuHQ
     $153                                           MonarchComputer
                                $67                  Axion Technologies

Power Supplies:
   Antec makes a 380 and a 430 watt model, which run about $70 and $75 respectively.  

Open question: Any suggs on alternates for a good power supply?
ChrisEddyAuthor Commented:

I've placed the order for equipment this afternoon, and I wanted to share this with you.

Motherboard    Intel D915GAG  $112 at NewEgg     Current technology that is about as good as it gets for the uATX form factor.
Memory            PQI 2x256 pair     $99  at NewEgg    These rams have cache latency rated at 2, rather than 2.5 or 3.
Processor         Intel P4 3.0Ghz  $189  at NewEgg    FSB=800Mhz, L2=1Mb
PowerSupply    Antec True430     $75  at NewEgg    Steady, quiet, reliable life insurance.
CD+DVD         MSI XA52P         $55  at ZipZoomFly  If you have to upgrade, simplify with 1 drive rather than 2.


The D915GAG mobo is as good as it gets right now for the uATX form factor, and building a machine using this will cost about .25 what an equivalent Dell would cost to order.  Btw: I sent a question to MSI tech support re: the combo drive, and (replying within 1H!  Go MSI!!!) they said: "As long as the SATA Controller which the drive is connected to it is Intel SATA Controller. The drive will not work with other SATA Controller chips like Via or Promise."  Thank you nobus for getting me started in the right direction!

Since the 800 Mhz FSB can only be achieved with pairs of memory, this is that.  Thank you kpaske for tuning me into this!  (btw: the big reason why I didn't go with your specific suggestion was a) anything over .5G of memory is overkill at this time, and b) what kind of name is "mushkin" anyway?  Is it like Smuckers jellies and jams, where they *have* to be good.  )  DDR2 was not chosen because of it's price, although the Intel technical documentation says that the D915GAG mobo can support DDR and DDR2 (just not at the same time).

The processor speed is safely in the range of state of the art but not bleeding edge, allowing the processor price to be in the range of "reasonable", and for the design of the machine to have a 4 year usable performance life.  I was interested in seeing faster processors with L2=@Mb costing a premium 8-10x over the equivalent sprrd processor but with L2=1Mb.  Thank you again to kpaske for steering me in the right processor direction!

The power supply was upgraded gently.  I'd hate for this box to become unstable for a stupidly shortsighted decision as saving money on teh power supply, at the risk of instability and the tell tale flag of failure that is the smoke test.  The one downside of this PS is that it has a 20 pin power connector, rather than the 24 pin jack on the mobo, but research indicates that there is a proper work around.  Thank you nobus for sending the link to the power supply calculator!  Although going for the True550 powersupply would have been only +$25, and it *does* have a 24 pin power connector, I was already over the estimated max price and didn't want to go any further.   Here's a link to an interesting power supply ranking done by Toms Hardware: http://www.tomshardware.com/howto/20040122/index.html   The True430 is powerful enough for this application even after derating for results from different review measurements.

The MSI combo drive may not be as highly reviewed as the Plextor drive, but they really aren't far apart.  Both of these SATA drives are fast and capable, but the MSI drive costs about .5 as much as the Plextor.  There were some initial availability problems when shopping for the MSI drive, but I did manage to find it by checking some specific websites that I know about.  Thank you tapkep for recommending both the MSI and Plextor drives!

I'm going to leave this question open for about a week.  It's going to take a few days for the parts to arrive, plus things *could* go wrong while building.  And even though I've researched just about everything I could think of, past experience shows that things do have an interesting possibility of failing or otherwise behaving strangely.  Maybe everything will work swimmingly the first time, you never know.  But I'd like to post the results of the build and get your feedback and guidance.

ChrisEddy -

No prob on the suggestions;  I see you've taken them all to heart and done your research, which is always better than buying based on one or two people's opinions.  The Intel board you've chosen sounds really good, and you got the uATX form factor you were looking for (although I'm still inclined to believe your tower probably supports standard ATX).  LGA775 is still fairly new technology, and I'm surprised to see what a good deal you got on that mobo.  As for Mushkin, yeah, I know.  Silly name.  Partly why I bought OCZ memory myself (along with the cool gold heat sinks).  But we really shouldn't discount it just because of the name.  It's become a top rated memory brand, especially among gamers, who are some of the most performance conscious of consumers.  If your budget is tight, 512MB is probably plenty for what you are trying to do.  On the other hand, I would have sprung the extra $125 for twice the memory and a beefier power supply myself, but that's just me.  ;)

ChrisEddyAuthor Commented:
Thank you!

On uATX vs ATX, yeah, the full size tower probably would have supported the ATX board, but I don't have the box with me right now so I don't know that for sure.  I don't want to have an "oh shucks" moment at build time.  Also, a big driver was to keep the price down as a stealthy way of getting excellent technology.  Although LGA775 is new, it's not really that new, but it is newer than the other stuff on the market.  Since this is a rebuild, now is the proper time to consider the 915 chipset stuff, and with a little luck and a lot of homework, the number of "aw shucks" will be small.  

I agree with your assessment on being aware of shallowness when making choices, like judging merely based on a name.  I was playing a bit, because Mushkin is like Smuckers in that they have a psychological name barrier that needs to be overcome with superior product.  However, my mom was given a bottle of red wine by a friend of hers with a sense of humor.  The wine was called "Cheap Red Wine", printed in red letters on a brown paper label.  Let's just say that it was wet and had alcohol in it.  

On the little bit additional, I completely agree.  While still working on the machine, I told my customer what I thought he needed, and in response he asked me what I thought it would cost, and to give a number like my life depended on it.  Well, since I estimated that he'd need a mobo, processor, memory, technically I came in at 20% under budget.  However, in taking this path, two more components needed replacing that were not anticipated at the estimate, which nudged the total price a little over the estimate.   So I'll get the job done with no corners cut, and there will be a reasonable "margin of safety" with the power supply, but not a large margin of growth.  And the fact that the uATX mobo has a limited number of expansion slots will help to practically enforce this in the future.  Think of it as being like the waste gate on a turbo ...

... but the proof is in the eating of the pudding.  Time will of course tell.  My next status will be in about 2 days, but I'll continue to monitor the comments.

Let us know when you will get requested parts and upgrade computer. :)
New power supply should fit in old one's place (thay are standard, unless you have some branded proprietary case).
ChrisEddyAuthor Commented:

The motherboard, processor, and powersupply arrived today!  Go NewEgg!!  Unfortunately, the memory was not shipped in this package.  Hopefully, they and the CD drive will arrive tomorrow.

I've done some further research into each of the components.

The Antec 4300 watt power supply comes with 2 SATA power connectors, in addition to the bevy of 4 pin Molex style power connector.  There are connectors for fans, and mention within the Users Manual that the powersupply can control the fans.  I haven't been able to determine whether it means that the PS will throttle fan speed based on measured temperature by the PS, or just to feed voltage to the little spinners.  Btw: the internal fans are made of a semi-clear plastic, which I suspect is reactive to UV light.

The D915GAG mobo has 2 fan connectors, and two temperature sensors - one on the die of the processor, and one outside but close to the heat sink area to measure ambient temperature.  Further reading of the modo doc indicates that it can and will throttle the fan speed based on temperature needs.  http://www.intel.com/design/motherbd/precisioncooling.htm

I like the optional spdif and com2, dual color on/stndby status LED,  front panel audio+usb+1394.  Connected right, this is good status.

Intel provides utilities for this mobo.  

The Active Monitor is a realtime temperature and fan speed monitor, which kinda looks like a car dashboard.  http://support.intel.com/design/motherbd/active.htm

The Audio Studio allows adjustment of the onboard DSP.  It's made by a company called Sonic Focus.
Here's the INTC description:
Here's the company website:
An article by the UK division of PC Magazine has generally positive things to say:
btw: the driver download is called: "AUD_XP2K_5.10.0.5027_PV.exe", which to me, based on past experience searching for Creative Audigy drivers, suggests that this is an Audigy device.

The Integrator Toolkit allows for non-usual adjustment of Bios attributes.  I like the idea of defining my custom splash screen *in* the Bios!  Perhaps a variation of the (somewhat old, dated) "Intel Inside" marketing theme.

There are also some of the ones you'd expect bundled in with a computed from Dell, eg:
       MusicMatch* Jukebox
      InterVideo* Home Theater Silver
      Norton* Internet Security 2003 with Norton Antivirus*
      AVG Anti-Virus for Linux Workstations
      Kaspersky* Anti-Virus Business Optimal
      NTI CD Maker*
      SoundMAX* 4 XL
      RealONE Player*
      Farstone* RestoreIT* Lite
      Executive Software Diskeeper* Lite
      InterVideo WinDVR* Lite
      InterVideo WinDVD* Creator Silver
      InterVideo WinDVD* Recorder Silver
Unfortunately, there is a provision which tempers enthusiasm: "Included on the CD with Boxed Intel Desktop Boards only.  Some software may not be included with all products.  Check individual CDs or product packaging for software availability."  Yeah, like I'm *not* going to check the CD's when I'm install-the-software mode.

Gotta go and make dinner for the fam.  But I'm reasonably happy with the choice so far ...

ChrisEddyAuthor Commented:

On the Intel Audio Studio, there is some more information which I missed before.  The reason I missed the disclaimer: it was right next to the Download link.  

The disclaimer is: "20-use no-obligation trial version.  Purchase for introductory price of $14.95 until 11/15/04".  So, does this mean that I get 20 hits to tune the whiz bang audio hardware hat comes with the mobo anyway, using a "take the puppy home" enticement for of crippleware? Kind of takes some of the shine off for me.

ChrisEddyAuthor Commented:

The remaining two packages, memory and CD+DVD drive, have arrived this afternoon.

I've talked with my customer and we've agreeed that work will start tomorrow (Friday) afternoon.   I estimate the work will take 2-3 calendar days, partly to do the work of assembly and the installation of software, but also to let the new parts burn in for a couple of days to confirm that the machine is thermally and actually stable.  In the mean time, the current machine is locking up and serving as an impediment to his getting work done.  The sooner I get the machine, the sooner these problems can go away.  

ChrisEddyAuthor Commented:

I recieved the machine yesterday, and this morning began the process of assembling the motherboard.

One problem occurred pretty soon: the processor package is the wrong type for the motherboard.  The D915GAG takes an LGA775 package, what was recieved was a PGA475 package.  It will not fit.  I will need to return the computer to my customer until the correct part arrives.

I have reviewed the order data, and it looks like I overlooked the package type.  I used the above link for the recommended processor, but I did not double check the processor package type.  In small defence, NewEgg doesn't exactly put that data up at the top of the list.  Here's a cut and paste of their product listing:

Intel Pentium 4/ 3.0E GHz 800MHz FSB, 1MB L2 Cache, Hyper Threading Technology - Retail

Model# BX80546PG3000E
Item # N82E16819116171
Model: Intel Pentium 4 w/ Hyper Threading
Core: Prescott
Operating Frequency: 3.0GHz
FSB: 800MHz
Intel Pentium 4/ 3.0E GHz 800MHz FSB, 1MB L2 Cache, Hyper Threading Technology - Retail
Model: Intel Pentium 4 w/ Hyper Threading
Core: Prescott
Operating Frequency: 3.0GHz
FSB: 800MHz
Cache: L1/12K+16K; L2/1MB
Voltage: 1.25V - 1.525V
Process: 0.09Micron
Socket: Socket 478
Multimedia Instruction: MMX, SSE, SSE2,
Warranty: 3-year MFG

An important improvement that could have been added to the automated online ordering process is a red-flag warning when a motherboard of one socket type and a processor of a different socket type are in the same order.  If I were placing the order with a person, I suspect that this difference of packaging would have come up.  There are computer manufacturer websites that already red-flag incompatible configurations when ordering a computer online, so it's not like this is new and unbroken ground or intellectual heavy lifting.  Perhaps this is a way for the otherwise good automation of their ordering process to be improved.  In the mean time, my lesson learned here is to explicitely review the detail of the entire invoice prior accepting, and not presume that each item was chosen right just because I did it (I am, after all, never wrong ...).

So progress stops, for now.  The computer will be reassembled with the old parts and returned.  I've gotten an RMA for shipping the wrong processor back, and will do that today.  And I'll need to order the correct processor today, which will probably arrive on Wednesday.  I'll report more status at that time.

Yeah, that's an easy mistake to make.  As you probably figured out, the link I posted above was for the Socket 478 processor to go with the Asus motherboard I recommended.  At least with NewEgg's fast shipping you won't have to wait too long to get your exchange.
ChrisEddyAuthor Commented:

The replacement processor is inbound.  Fedex tracking shows that it was picked up on Sunday morning, and left Sunny CA yesterday morning, but no status since then.  Perhaps this 1+ day lag on status is by design, and "delivered" status will be reflected more promptly.  But our dogs will announce the presence of the delivery much sooner.

ChrisEddyAuthor Commented:

The computer has arrived, the new parts installed, the new install of XP Pro is complete, and everything is functioning.  I've also gotten 2 80mm case fans, mounted front and rear, to allow the mobo to throttle fan speed should it find the need to defend itself.

I've downloaded the Intel Hardware Monitor, to watch temperatures and fan speeds, while seti@home is run.  Since hyperthreading is enabled, the task manager shows two processors, and seti@home is only running on one of them.  Running the StressTest application in continuous mode does drive the processor to 100% utilization with a few small notches.

Room temperature = 69F.  The conputer case is closed but not sealed, because there is a missing panel from the front, and 3 card openings on the back.  The reported temperatures are system zone 1, system zone 2, and processor zone 1.  With just seti@home running continuously: System zone 1 temperature is cruising at 118F, fan speed of 2200.  System zone 2 temp is 118, fan speed is 2100.  The processor zone is at around 147F, fan speed is 3100.  About once every 5-15 seconds, there is a warning popup that "system zone 1 temp has exceeded it's recommended maximum temperature."   The summary screen, updates 1/sec, shows a monir transient temp spike of 1-3 seconds, where 1sec is the norm.

With both seti@home + StressTest running at the same time, I can hear an occasional increase in fan speed.  After running for a couple minutes, system zone 1 temp = 118, fan speed = 2100; system zone 2 temp = 117, fan speed = 2050; processor zone temp = 149, fan speed = 3450.

According to the Intel documentation on the processor temperature, the green zone is 100F or less, yellow ranges from 100F to 152F, and red  is above 152F.  

Although these temps aren't exactly room temperature, I'm not too worried about this because over the past 1H or so, the temps have been reasonably constant, with the occasional and short lasting temperature blip above the max.  

Q: Am I right in not being too concerned about these temps, or just complacent?

Q: How can I confirm whether the heatsink is mounted properly.  Note: this is an LGA775 package, and these clips are definitely different from the PGS475 clips.  Although all 4 white nylon barbs are set below the mobo, the black hat plunger on one of the standoffs is not all the way down like the others are.  

Thanks for noodling on this with me!

if the black plunger is not all the way down, it is not locking the white part open, so it can come loose. This MAY indicate that the cooler is not completely flat mounted. I always check this mounting with the motherboard still out of the box, by looking at it from all sides.
ChrisEddyAuthor Commented:

I did mount the heatsink while the board was still out of the box.  The mobo warped noticeable, almost 1/8 of an inch.  Soon after pressing in the 4th mounting pin, one of the other mounting pins popped up.   Most disconcerting.  Looking more closely, the black pins actually weren't down all the way in a locked position.  

Considering the non-flat flex of the mobo, I felt more comfortable squeesing the pin down onto the board between thumb and forefinger, rather than pressing down on the pin into the board.  Twisting the pin 90 degrees made a sound like fingernails on a blackboard.  

All four of the white nylon split barb retainers are below the mobo and holding.  Three of the black pins have a narrow black plastic finger showing with the bottom of the nylon retainer, keeping it spread apart and is the final locked position.  The fourth black pin was about 1/8 of an inch from the bottom of the nylon retainer.  I squeezed that one down, and watched the lock click into place.  Mounted in the case, all four pins feel fairly stiff when wiggled gently with an index finger.

I left the machine running overnight with seti@home plus StressTest.  I could hear the fan speeds vary up and down slightly, which I take is a good thing.  This morning, I stopped the stresstest, and checked out the progress of seti.  Over 8 hours, it completed about .48 work unit.  Not impressive, but it was probably displaced by the more agressive stresstest program.  

Note that seti@home is running continuously in the background as a method of consuming cpu and producing heat, to see how well a a worst case scenario can be cooled.  Although I will not be returning the computer with this running on it, I do think that he should have the opportunity and the option to run something cpu heavy and not get temperature complaints like I'm getting now.

I've decided to disable hyperthreading.  Although the task manager gives a cool 2 processor display, and I can see different consumptions on both "processors", I like the idea of a process getting 100% of the cpu rather than up to 50%.  This should dramatically help the seti@home scores (in about 1H, an additional .25 work unit has completed.  that's more like it).  But more important to this application, will improve the performance of the speech-to-text application to near realtime.

With the case closed, I'm getting frequent (about once every 15 seconds or so) and annoying popup complaints (it's modal, so it is the topmost window) from the hardware monitor, reporting that the system zone 1 temperature has exteeded a max temperature.  

  System zone 1 temperature cruises at 117F but briefly peaks at 121F, system fan 1 speed is about 2200rpm.  

  System zone 2 temperature also cruises at around 117F, briefly peaking at 118F, system fan 2 speed is about 2000rpm.  

  Processor temperature is cruising at 145F, briefly peaking at 147F, processor fan speed is about 2650.  

With the case open:

  System zone 1 temperature cruises at 109F, briefly peaking to 113F, fan speed is about 1720rpm.

  System zone 2 temperature cruises at 109F, briefly peaking at 111F, fan speed is about 1600rpm.

  Processor temperature is cruising between 138F and 140F, fan speed is about 2260rpm.

Note that the current temperature problem is that an area of the mobo is complaining that it is too warm.  The processor temp is fine, and the other part of the mobo is fine.  

The following is a description of the current method of cooling.  Hopefully there can be a recommendation for improvement.  

The front case fan (80mm) is at the bottom front and pushing air in.  Having my hand in front of the exhaust, I can barely feel air moving.   But it is quiet.  The rear case fan (80mm) is in the middle, just below the power supply fan and behind the processor cooler, and is pushing air out.  It's force is also something less than blowy, and it is quiet.  The powersupply fan is also sucking air up from above the processor, and pushing air out of the back of the case.  It's air force is a little stronger than the other two, but the temperature of the air coming out is a nice comfortable warm feel, not hot.  In the back of the case, where there used to be boards plugged into slots, there are now 3 openings.  There are grey ribbon cables which connect the legacy floppy and hard drive.  The case cables, for reset/ hdd led/ power, have been tucked up and away do as not to droop into the airstream of the lower front fan.  

There need to be at least 2 fans, which are both controlled by the mobo.  These are measured things.

The Antec TruePower 430 watt power supply has a 2 wire plug on the wiring harness labelled: "psu fan monitor connector".  I presume that this is to monitor the spinning of the PS fan, and not to give it signal to go faster.  It is unused.

Perhaps the ribbon cables could/ should be changed to something with less of an air deflecting profile.  

Perhaps replace the front fan with something bigger.  There us currently a nice slidable plastic cradle which is designed to hold an 80mm fan.  However, there is about another 3/4 of an inch total around the fan that would allow a larger fan to be installed, just mounted different.  

Btw: seti@home has completed the first work unit and has started on the second.  It has worked for 27 minutes, and is 11.95% done.  That calculates to about 3.75H per work unit.  Rock on!

For a test, turn off fan controls at BIOS (so fans will spin at maximum speed). If there is temperature change - you should use bigger fans, or install more fans.
I would recommend, if possible, cranking up the speed of the rear case fan.  What you need is more air to be pulled OUT of the case.  You should plug the holes behind the PCI ports, as this will cause more of the warm air from the mobo to be pulled out by the fan, and prevent pockets of heat within the case.

80 mm fans are plenty big enough.  They are often programmed to spin slow, but if you crank them up they will push a lot of air.
ChrisEddyAuthor Commented:

The only way I can get the fans to spin at their maximim speed is to connect them direct to the power supply.  The Bios for the Intel D915GAG board has a limited fan configuration, which allows the slowest fan speed to be either off or low.  Low is currently selected.  

I've varied several things since my last report.  I know, I know, you're supposed to vary one thing at a time and see how things are affected, but it empirically seemed like the right thing to do at the time.  

The power cables have been rerouted a little.  Before, they were bunched together just after they came out of the power supply, which is above and to the right of the processor cooler.  Now, the bundle has been shifted down to the next natural bend, and that bundle was parked in the unused phantom space above the CD+DVD drive.  The intent was to create a larger unencumbered space near the processor for air to flow more freely.  

I've replaced the 80mm fan in the bottom front with a larger fan that is about 5 inches across.  It's rated at moving 2x the air as the 80mm fan.  Now, I can actually feel some air coming out, where I coudn't with the smaller fan.  However, there are a fair number of holes in the front of the case, so it is reasonable that the case is not pressurizing because of this.  

Btw: on the topic of negative pressure ventilation (pulling air out) versus positive pressure ventilation (pushing air in), positive pressure tends to be much more effective in clearing smoke from a room after a fire, especially if your fan is on one side of the room, and a window on the opposite side of the room is opened.  Note that the catch is that the room is sealed, only one window is open, and only the one door is open.  Negative pressure ventilation tends to remove the smoke nearby the fan, and does not pull enough air away from the opposite walls, where positive pressure will.  At least, this is what I was taught in Fire Fighter Academy.

The two pair of flat grey ribbon cables have been replaced with round ones.  The idea is to remove obstructions to the airflow, and the cost of the round floppy and round pata cable was a special situation price I could not refuse.  So I bought alot of them ...

The 3 open slots in the back have been closed, eliminating air leakage there.  

The side panel is in place, and all unused bay slots in the front of the computer are closed.

Here are the measured "after" results:

  System zone 1 temperature cruising at 115F and briefly peaking to 122F, fan speed is about 1775rpm.

  System zone 2 temperature ranges from 115F to 117F, fan speed is about 2100rpm.  

  Processor temperature is cruising at 145F, fan speed about 2500rpm.

  I'm still getting complaints from system zone 1 being too warm, about once every 15 secs or so.

Net net, I don't think I've varied the temperature symptom enough to say I've made a measurable difference.

However, one thing I did find out was that the hardware monitor can have the alert temperature varied.  And according to the INTC documentation, the "green" range goes up to 100F, the "yellow" range goes up to 153F, and the "red" range starts at 154F.  Having an alert level of 120F may be a pessimistically low practical temperature, and therefore invalid.  And considering how often the "temperature exceeded" warning is presented, this is noise.  How's all that for rationalization.  ;-)  But I am going to change the alert temperatures for system zone 1 and 2 to 130F (122F) and 129F (121F) respectively.

I've done about as much as I want to do on cooling for now.  I'll watch it for the next day or so, and if things are still running too warm, I'll add another larger exhaust fan to the back of the case.  

Next is to activate windows.  I'm a little concerned, because I've already activated this license key after the fresh install with the old mobo with the keyboard controller failure causing a monsterous volume of interrupts, and things are going to be radically different than before.  I guess I'll see whether MSFT is cool with the idea of real people making major hardware changes or not...

ChrisEddyAuthor Commented:

Last night, MSFT interactively reactivated windows fine.  They asked how many times I've activated windows, for an explanation of what happened, and whether this involved many hardware changes.  I was kinda worried about this, but now it's positively behind me

I've posted a separate 250 point question on memory.  In a nutshell, the memory comes rated as being 2-3-3-6, but the Bios is showing it as 2.5-4-4-8.  Is this a problem?

A hardware problem occurred.  While still in the Bios, the sound of the fan speed changed.  Booted into Windows, and the hardware monitor showed that system area 1 and 2 fans had stopped (back interior of case), and that the interior case temperature was going up well into the red zone.  The processor temperature showed negative numbers, which I took as being very bad because it's strange, and I don't remember it's fan speed.  During this time, there was the sound of a fan going faster.  I shut the machine off fast, and didn't note down the specifics.  Left machine off for about 2 minutes, turned the machine back on, booted into Windows.  Hardware monitor shows temperatures and fan speeds within normal parameters.  The alert history shows that both system zone fans stopped before, and system zone 2 temperature went high.   I don't remember seeing whether this D915 mobo has built in overheat protection, like Asus does.  

Overheat protection is built in in Intel P4 CPUs, it works independently from motherboard. Some motherboards, however, monitors CPU fan speed and can shut down system, when that fan stops.
Since D915 motherboard is fairly new, i suggest to visit Intel website regulary to see, if there is BIOS update.

About hyperthreading: it is efficient only when several *different* tasks are running. Generally, when several tasks are running, they enter queue to get CPU cycles. If hyperthreading is enabled, there are 2 queues, instead of 1 - one for calculations with floating point, one - with integer numbers. So if you try to run seti@home, which uses floating point operations a lot, you will only use "floating point calculations queue", while "integer queue" will idle. I suggenst you to enable hyperthreading - that won't harm and sometimes will background programs will be able to run faster.

Right now, inside case, you get 25oC above ambient temperature. IMHO it is acceptable, but what temperature will be in summer? On the other hand, these temperatures are when CPU is fully loaded. Maybe you will get lower values, when using Dragon Naturally Speaking.
Positive vs. negative pressure:  The point you made about the smoke is a good one, which I've never heard before.  However, I think the situation you are describing is a little different.  When you are creating the positive or negative pressure, you say that the fan is located near a door, and there is a window on the opposite side of the room that is open.  This room is nowhere near "sealed".  If you take the fan and move it into the window and seal the edges, the results would be very different.  I believe the reason is because of the way air moves behind the fan:  it tends to suck in a lot of air right behind it.  Therefore, if the edges around the fan aren't sealed, it tends to suck a lot of the air in front and next to it right back in also.  On the other hand, if the edges are sealed, like in a window fan, it can only draw air from behind it.  Even more effective would be to shut the door, as it would create more pressure inside the room.

The way i discovered this was by experimenting with window fans in my father's house one summer.  He never used air conditioning, and summers in the Washington DC area got really hot, so I bought two window fans, both with intake and exhaust modes.  I had originally intended to use one on intake and the other on exhaust to get air moving through the room.  But that only dropped the temperature minimally - maybe 5 degrees.  When I tried both fans on exhaust though, it easily dropped the temperature 20 degrees.  It truly felt like the room was air conditioned, compared to the rest of the house.

However, remember that the source of the heat in my situation was OUTSIDE the room - opposite from a computer case.  There are some natural physics properties working for us inside the case.  Heat naturally dissipates towards cooler regions, so really, the heat wants to escape the case, we just need to help it along.  Exhaust fans doing most of the work should clear most of the heat out, but having an intake fan blowing across tends to help break up the heat pockets around the other heat producing chips.  In actuality though, in most cases, it would be better to seal off the holes, including those in the front of the case, because all they really do is lower the pressure inside, making the exhaust fan(s) less effective.

In my server pc's, I use a combination of fans, most of them placed at the top of the case.  I've got a single intake fan at the bottom front to pull some cool air in and move it across the mobo.  The processor fan (a really nice Zantec with a huge copper heat sink) pushes air away from the proc, and the intake fan gets that heat moving around inside.  Then all my exhaust fans are at the top:  one for the psu (which has its own intake fan on top), two built into the TOP of the case, and two each on my hard drives (they each have a heatsink unit attached).  If you look at the overall design, I have a fan near every heat producing component, blowing air across or away from it, and all the exhaust fans are positioned near the top because heat rises.
ChrisEddyAuthor Commented:
Good news about the built-in overheat protection!  There is a reflash available for the bios, thanks for the reminder!  Here at 8500 feet in the mountains of CO, max summertime temp is 85.  However, I'd hate to design a system that thermally excluded a place like Long Island.

I like your point on case ventilation pressure, and ack on heat rises.  Yours may be the best with computer cases in general!  

I'm remembering back when I grew up on Long Island, it's known to top 105F + 95% humidity during the dog days of August.  A lot of people used to have 36 inch diameter whole-house fans mounted in the ceiling at the center of the house and at the top of the stairs.  Although it sounded like a turboprop airplane, and you wouldn't run it much more than 15 minutes at a time because of the noise, it was extremely effective in rapidly turning over the air in the entire house.  An added bonus was a reduction in cleaning, because all of the little dust bunnies in the house tended to collect just under the fan, and the ones that were downstairs would actually roll up the stairs.  

So did the BIOS update help solve the problem?

Yeah, I had some problems with overheating in the past, so when I designed my current setup, I decided not to fool around.  I don't tend to spend a lot of money on my cases, so they generally need to be modified for optimal cooling.  There is usually just enough room in the top of a mid-tower to install two 80mm fans side-by-side, between the psu and the drive bays.  Also, if you can find a decent hard drive cooler - the kind with a heat sink and two 40mm fans - and install it in the top drive bay, it goes a long way to not only reduce heat inside the case, but also prolong the life of your hard drive.

A bit off topic, but still relevant, I've been looking into super-efficient house designs and have come across the cement "geodesic dome".  Apparently due to their reduced surface area, sealed design, and excellent insulation, most of those homes are super cheap to heat and cool.  A small exhaust fan or two at the top is enough to keep them cool, even in a hot climate.  Now I'm not suggesting that we build little dome shaped computers, but maybe case sealing and insulation would be something to look into.  It would also cut down on the noise that so many people seem to be worried about these days.  Of course, there is always water cooling, but that's a bit extreme for my taste, not to mention expensive.  I've gotten by just fine using smart design techniques, heat sinks, and a bunch of fans (although I admit, for a Prescott processor, it's worth it to buy a $50 Zantec fan).
ChrisEddyAuthor Commented:
On the Bios reflash, didn't get that far yet, but I will.  Needed to catch up on some email to a friend.  Next is serving dinner to the rest of the crew.

I like your idea about case sealing, and is thinking that I'm folding into the next computer I build, but not the cement geodesic dome part.  Ehem.  Sealing the case would substantialy reduce the issue of noise, which will soon be be near and dear to my heard on a non-carpeted floor, and would help manage the general direction of the airflow - provided the interior shape was done right.  I've been looking at Lian-Li cases, perhaps oogling would be a more accurate description, and so far I like several things about them.  Good interior shape, good intake and exhaust airflow paths, fairly thick aluminum case material, good construction.  I need to see at least one before I make a buying decision.  The case will be painted the color of our Volvo wagon, which they call Saffron, but I think it looks more like Metallic Pumpkin Pie.
This is the car, but not ours: http://img.yezzz.com/fi2101737.jpeg

ChrisEddyAuthor Commented:

After reviewing the list of devices in the Device Manager list, there is a big yellow question mark next to Other Devices, for a thing of type "PCI Device".  The properties say it's location is bus 0, device 27, function 0.  Device status is: The drivers for this device ar enot installed (Code 28).  

The device was uninstalled, then a scan for new hardware was launched.  The Found New Hardware Wizard completes with a "Cannot install this hardware" status.  

How can I learn more about what this device is?

ChrisEddyAuthor Commented:
Everest has been DL, installed, and run.

Under Devices->Windows Devices->Unknown,
There is a big yellow question mark for a "PCI Device", it's reported properties are:
  Memory=FFA3C000 - FFAA3FFFF.

Under Devices->Device Resources:
  IRQ 11 is shown as "shared" by a "PCI Device".

Under Benchmark: (for personal curiosity)
  Memory Read=4999MB/s    (Ath64 3800+/MSI K8TNeo2-FIR = 5980MB/s, P4-3.2E/Abit IC7-MAX3=5780MB/s)
  Memory Write = 1913MB/s (Ath64 3800+/MSI K8TNeo2-FIR = 2500MB/s, P4-3.2E/Abit IC7-MAX3 = 1740MB/s)
  Memory Latency=96ns         (Ath64 3500+/MSI K8NNeo2Platinum/2-2-2-5=48.3ns, P4-3.2E/AbitAA8-DuraMAX/3-3-3-8=84.5ns)

This is a new tool to me, so I may be using the tool wrong or missing something, because I don't have more information now than I had before.  For all I know, it could be The Un Known Comic device, except the representative icon is not an upside down brown paper bag.  But the white noise from the fans do kinda sound a little like laughter...  Any suggs?

Try this then : The unknown devices identifier


You see, we 've got always something else up on our sleeve.....
ChrisEddyAuthor Commented:

Btw: I just completed reflashing the Bios.

The only generic item on the list is:
PCI Device
  OEM Device from Intel Corporation
  Chip: Intel Corporation
          PCI Device
      Chip Vendor
          Intel Corporation
      OEM Vendor
          Intel Corporation

... So it looks like an Intel thing.  
Any suggs on addl detail?

kpaske, were you able to find *quiet* HDD cooler with 40x40 fans? From my experience - they (one made by Titan, another - by Hama) are quite noisy. I ended placing HDDs in 5.25" devices place and mounting 80x80 fan in front of them. Had to drill several holes in front of the case, too :)

ChrisEddy, have you installed chipset drivers as well as integrated devices drivers? (i know, you should have already done that, just checking :) ). Sometimes helps selecting that "unknown device" and ordering windows to "reinstall driver". Usually USB controller's drivers are already in the system, but somewhy they don't get properly configured from the first time.

ChrisEddyAuthor Commented:

Yes, the latest and greatest chipset software ( has been installed.  (good question, though.  this *is* the most likely suspect)
Just for grins, and to rule out pilot error, I'll rerun it.
Btw: All MSFT updates have been applied, including SP2.
Done.  Restarting.

A dialog box is presented after booting up and logging in:
  Title: Tray Program for Intel (R) Desktop Utilities
   Body: Exception source: TfmIntelTray
              More than one queue object exists.
   Button: OK

Checked startup folder.  Empty.
Checked HKLM->sw->MSFT->Win->run, nothing relevant or interesting.
Checked Device Manager, Other Devices -> PCI Device
    Properties: type: unknown, Manuf: unknown, Location: PCI bus 0, device 27, function 0 (same as before)
    Device status: The drivers for this device are not installed [code 28], to reinstall click it ...

Yes, I have gone through the process of uninstalling and reinstalling this unknown device.
But just for grins, and to eliminate pilot error, here goes.
    "cannot install this hw.  the hw was not installed because the wizard cannot find the necessary sw:..
    Button: Finish

Changed device usage from "use it" to "don't".

Checking the integrated device drivers.  
  Integrated 10/100 lan driver (got it. ) (reinstall apps via repair) (reinstall netw drivers) (uninstal+reinstall unk pci device.  failed)
  Integrated audio (got it.)
  Integrated graphics  (got it.)
  INTC Chipset sw install utility (got it)

New news: I've run out of disk space on C:!!!  It's a 130G partition on a 250G drive.  I've scavenged some space back by deleting temp files.  SP2 is applied, so LBA definitely is supported now.  Any suggs on stretching the C: drive to use the full physical disk, *without* reformatting and reinstalling?

ChrisEddyAuthor Commented:
False alarm ...
The low disk space complaint was reported when installing the integrated audio drivers.
Looking at the consumption of C:, I was looking at the usage pie-chart all wrong.  
Free disk space is 131G, Used is 6G.  
... but I am still interested in stretching the C: drive to the max physical size, using a $0 solution if possible.

ChrisEddyAuthor Commented:
More information.
Changing the target directory to recieve the unzipped installation, from reusing the same directory name to something named "foo", the unpacking and installation completed swimmingly.  
Looks like another case where the application is misreporting an error.  I love wild geese ...

ChrisEddyAuthor Commented:
Reinstalled integrated graphics sw, and the chipset sw.  Restart.
Restart.  Check device manager.  The mystery PCI device is still present.

Any suggestions, other than it's time for lunch.

I happen to be cruising through the built in speech-to-text feature of XP, and I was presented with an interesting error message:
"Could not initialize volume meter.  Please select a different audio input device."
Hmmm, could this mystery PCI device be related to this?

Bringing up the "Sounds and Audio Devices Properties" box from the control panel,  on the Volume tab it says "No Audio Device".

On the Audio tab,
Sound playback uses the default device, which is: "Modem #0 Line Playback",
Sound record uses the default device, which is: "Modem #0 Line Record."
I'm expecting something that says something like: Azalia, Dolby 5.1, or something spiffy like that.

Clicking on Volume Control, an error message is presented:
"There are no active mixer devices available.  To install mizer devices, go to Control Panel, Printers and other hardware, and then click Add Hardware."
Tried to install the audio sw for this, failed.  

So it looks like the audio subsystem is nonfunctional, in addition to the presence of the mystery PCI device.  
The absence of audio is going to be a killer issue, because the prime usage for this newly rebuilt computer is speech-to-text.  

I'll check on the INTC website re: anything like this.
But in the mean time, if there are any suggs, I'm all ears...

What lunch?! It's 22.00 already... :)

No more ideas right now, unless some obvious ones: uninstall Intel add-on and demo software (Audio Studio, integrator toolkit), then reinstall audio drivers. My experience with i915 chipset motherboards is (still) from Asus motherboards, i haven't tried Intel value-add software. For troubleshooting, everything should function with drivers only (without audio studios and so on).
ChrisEddyAuthor Commented:
Found it!

The mystery device was the onboard high end Realtek Azalia audio subsystem.

The "full disk" complaint was merely a distraction to redirect focus away from the real problem.

I had originally DL'ed and installed driver version *.5032.  It is the only choice presented by that INTC DL site.  

The fix was to go to a different part of the INTC website, and DL and install driver version *.5027.   Technically this is a downgrade to the driver, but this wouldn't be the first time that "new and improved" isn't quite as good as "old and crappy".

Now, sound plays, and the RealTek demo plays nicely!  There is a cool effect which makes the sound sound like it is going below you  and then above you, in sync with a graphic of a globe circling a graphical listener.

Gotta go and take the kids to the Dr.  The daughter has an absolutely raging case of tonsillitis, and the son is not feeling so "new and improved" either.  

Later gators ...

ChrisEddyAuthor Commented:

The computer was delivered tonight and was received well!  

The overall responsiveness of the machine, especially the speech to text conversion, was snappy, safely staying in near realtime.  Or in the words of my customer: "the conversion stays contemporaneous with my speech."  The conversion was never more than 2 or 3 words behind, but was cruising at about 1 word behind.  The CPU utilization during this speech to text conversion cruised at around 30%.  

The only thing that wasn't dead on was the location where the new CD+DVD combo drive was mounted, right above the floppy drive, which positions the dark tray to be very easy to accidentally kick when it's left out - which is surprisingly common for me.  I'm thinking about spraying it a lighter color, maybe traffic cone orange ...  Perhaps there's a timer available to autoretract the tray after so much time has elapsed.

Tonight was a session of installing and configuring software, monitoring, final testing and troubleshooting, installing a printer, and buying software to placate the obnoxious nag screens.  There are several little things that still need to be done, but the volume is asymptotically approaching zero.  

This job took longer than it really should have from start to finish, but he's reasonably happy with the result.  I'll be returning in a couple of days to refine some things.  

I really want to thank you for your assistance, guidance, encouragement, and friendly nudging back towards center!  And thank you for sticking with me, even though the question evolved to cover the project.  I found this to be a constructive method of working and troubleshooting, partly because having it in writing helps me to stay on task with running status, but also because I don't otherwise get a chance to metaphorically "rub elbows" with people who know many more things than I do.

I'm glad it worked out for you.  It's nice to see someone who builds a system, and pays attention to the details, and also helps research their own problems.  It's a learning experience for all of us.
I had the same with the audio device ; look here : 3 last comments

ChrisEddyAuthor Commented:
At least I'm in good company.  Now I don't feel so bad.  ;-)

I failed to mention before that one of the solutions was to repoint the audio inputs away from the modem.  
While I was trying to identify the mystery PCI device, I decided to check the volume controls, and found that the modem was the only choice.  That strange absence of choice raised my level of suspicion about the onboard audio possibly being the mystery PCI device, probably on the reptilian or conspiracy theory level.  I knew I had to solve the audio problem, and hoped the mystery device would also be solved, which it ultimately was.  

Btw: the proper driver includes a slick little sound tweaker from the manufacturer of the sound chip, RealTek.  It has a demo mode that does a pretty good simulation of a repetative sound slowly rotating vertically or slowly horizontally, in addition to some other predefined sound types.  

ChrisEddyAuthor Commented:

I'm having an interesting challenge of an intermittent  problem, eg: once every week or so, but with the speech to text translation files getting corrupt after a graceless powerdown, which is motivated by the machine becoming hung.  

Here's a link to the new problem if you're interested:


Did yopu bother to consider my orignal post- the 915 chipset is buggy.
Good luck maybe a new bios upgrade is available.  It is your only hope.
x86fix, please provide links to your "scary news about i915 chipset". I guess they are about recalling some motherboards with defective chips.
Also, you have not answered about "crappy Asus boards".

You need to stop guessing and start reading.  
It is not hard to find reviews from reputable sources.
Reputable sources does not include the marketing department.
Funny that you criticize the comment about Asus, I have only typed it in a couple times and I have been told by others that they suspected the board but didn't wan't to replace it - after reading my comments they were motiveted to change boards and found the problems went away.

I don't recall where I read it and it is not incumbent upon me to defend that comment-  It is simply relaying a fact to help others decide.  If you read tomsharware or anandtech omong others you might learn something- you may even find the article I referenced.  It is up to yuo to take other peoples comments and learn from them.

This is a forum for free advice - those who  il to recognize the value of the advice they get might learn from their mistakes.  This question is a perfect example of that and also of the fact that these 915 chipsets dont have a proven stable record.  The early bios versions are especially buggy and the newer ones may still have problems.

I don't see bug problems with the 865 or 875 chipset.  The 915 will have it's day but my point is you have to decide if you want to be a beta tester or not.  I think it is ok to choose yes and buy this for yourself.  I also think any tech worth his salt thinks twice before putting this in a computer without first warning the user that it is not the most stable product.  Clearly you are either a hobbiest or a weak player tapkep.
Question has a verified solution.

Are you are experiencing a similar issue? Get a personalized answer when you ask a related question.

Have a better answer? Share it in a comment.

Join & Write a Comment

Featured Post

Cloud Class® Course: CompTIA Healthcare IT Tech

This course will help prep you to earn the CompTIA Healthcare IT Technician certification showing that you have the knowledge and skills needed to succeed in installing, managing, and troubleshooting IT systems in medical and clinical settings.

  • 28
  • 9
  • 7
  • +2
Tackle projects and never again get stuck behind a technical roadblock.
Join Now