New build freezes

[Ron AF Greve]

Hi,

Built a new system with the following specs:

Antec Sonata case III PSU 500W midi tower
3 Seagate baracuda's 7200.11 500GB 7200rpm SATA
Asus ENGTX260 TOP/HDTP 896MB
Kingston 4GB kit (2x2) KVR1333D3N9K2
Intel Core 2 Quad Q9550 2.83 GHz
Gigabyte GA EP45T DSR S775 4xDDR3 (replaced this with asus P5QC however)
Lite-on iHAS120 20 DVDRW SATA

Unfotunately the system freezes completely randomly. First it used to freeze
a few times when I started the system The I could play for instance
oblivion for a while work in visual studio during the day and when playing
again in the evening it might freeze one or two times. But the system seemed
to become more unstable. Especially freezing very fast when playing any 3D
game. But also when playing video files or less frequently (but still a few
tmes per day) when just on the desktop.

I tried setting all kinds of parameters on the GA EP45T DSR board when I
found a similar problem/thread on the
http://forums.tweaktown.com/f69/ga-ep45t-ds3r-weirdest-behavior-28220/index2.html
site. However even with adapting that to my memory it didn't solve anything.
Fair to say that the memory is not listed on the QVL however the board is
sold as able to handle more than 4GB and I see no modules of over 2GB on the
QVL.

I replaced the 260 with an old matrox mystique card but still the freezes
were there. In the end I figured the motherboard might be broke. I ran it
with one stick of memory and the other either one froze the system although
less often ( in hindside I am not sure I loaded default values for the
memory again though so at that point they might have been off). Since it is
a dual boot system I also run it in Ubuntu, unfortunately same thing, random
freezes. This made me believe it was always certain a hardware problem.

I decided the motherboard might be broke, instead of RMA'ing and waiting a
few weeks, I ordered an asus P5QC. Of course when putting it together again
I sort of expected it to work. But two minutes in oblivion and another
freeze.

The PQ5C has however a very nice feature which let you run linux from a USB
stick before loading anything else. So I removed harddisks, cd-rom, floppy
and one memory stick (for some reason the mystique doesn't work on this
board?) No freezes, I attached every again no freezes either. The only thing
left is of course the memory stick.

I decided to google for dual channel and freeze, apparently there are lots
of problems with different MOBO's and memory over 2GB. Sigh, well at least
it seems to work now with one 2GB stick.

Though I still won't hold my breath, (the system freezes very randomly) up
until now (24 hours without a freeze) it seems (more) stable. If it is
stable for a week I will try the other memory stick. For now I am too happy
that I finally have a system that at least works for some time.

Well if anyone got a good idea to make it work with two sticks.

----------------------------------Maybe some useful things I noticed during
building the system-----------------------------
When installing the cooler I think it is better to push the white things
that are supposed to go through the holes on the motherboard through with
your nails. This way it is a lot easier to push down the black pins to lock
the white in place (I also installed the processor and heatsink on the asus
board first before screwing it in the case).

Things I like and dislike about the asus P5QC.
* When you have a large card like the 260 only 4 of the 6 sata ports are
available (and one external sata port) Because the card obscures two of the
sata ports (and this is an asus video card in an asus board :-(
* Also all the sata ports are in the same direction meaning you have to
remove the ones above it first (otherwise you can't press the metal release
clip) (and in the 260 case you have to remove the video card otherwise you
can't even press the top ones).
* The board boots up pretty fast (but sometimes tells me there is no OS,
have to look in the bios maybe some timing, anyway not a big deal).

* I like the express gate (where you can start a linux environment by
preparing for instance a USB flash stick and running linux before booting,
network etc are enabled and you can use a second USB stick as a drive to
write downloaded stuff to. It doesn't seem to write to the USB stick (so it
doesn't wear it) (if you don't download stuff of course). Great to check for
problems with a minimal configuration.


The GA EP45T DSR S775
*Has a very nice layout for the motherboard (sata ports are in two rows of
three and facing each other to the metal release clips are always on the
outside.
* The large 260 card doesn't obstruct anything
*What I dislike about the board it that if you choose optimal values it will
overvolt your DDR3 ram (without giving a clue about that). It will set the
ram voltage to 1.7V instead of DDR3 1.5 volt.


Regards, Ron Greve

 

[Ron AF Greve]

Hi,

Thanks for your reply.

I would expect it to. You've got roughly 500-550W of hardware running off
of a 500W power supply. Well, to be more accurate, you've got so much
draw on the +12V rail that most 500W power supplies would not handle your
hardware. What amazes me is that it POSTS at all. Any 500W power supply
other than an -Earthwatts- 500W probably wouldn't even POST that system.

Maybe you are right there, the sonata contains a build-in Earthwatt's power
supply

Note that I thought if it first too, however I didn't see any odd voltages
with monitoring software (the voltage were still on screen during the
freeze).

A 500W will barely handle it, assuming that most of that power is
available on the +12V lines. Otherwise, even a good name-brand power
supply would be under-sized at 500W.

Linux will help a bit, as it stresses your hardware less than Windows
does. In other words, running linux is probably the only way you can try
to make this system stable with a 500W Earthwatts PS.

Yes, I read that you seem to be stable with only one memory stick. That
COULD indicate you got a bad memory stick. But as it's Kingston we're
talking about, the odds of a bad memory stick are pretty low.

If it was my system, I'd upgrade the power supply to at least 650W, and
that's assuming a VERY good name-brand power supply. THEN I'd run
memtest86 overnight two nights in a ROW, to test each memory stick
individually.

Oddly enough (I forgot to mention) I did run memtest a few times for a few
runs and finally an overnight session together about 13 complete passes
using dual channel mode (both sticks installed) never got a single error out
of it (also did never freeze using memtest). That's why I actually figured
it might be the MOBO. I also run prime under windows. Although it didn't
report errors either I stopped it and at one time, letting it run for a long
time, looking at the time when I came back it had run for two hours, the
system was frozen I didn't notice anything odd about the values also frozen
on screen (ok, a sudden power drop might freeze the system and not be
vissible of course), still no errors from prime.

I'm betting that you will find that the system is totally stable once you
get the proper sized power supply in it. NOTE that a bad or under-sized
power supply can cause symptoms that would seem to point to bad RAM or a
bad motherboard or CPU. And AT BEST, your current power supply is
marginal, as you are running it at ~80% or more of it's maximum rating on
the +12V rail constantly. (yikes) -Dave

You might be right, that one stick might just be too much. However since
removing that stick I haven't yet noticed a freeze (but still since in the
beginning I also could run it for a long time, I am still not completely
sure it is sort of solved (still regret I can't put the second stick in).

Gotta think about a larger sized supply. First going to see what it is doing
for now.

Regards, Ron

 

[Dave]

Hi,

Built a new system with the following specs:

Antec Sonata case III PSU 500W midi tower
3 Seagate baracuda's 7200.11 500GB 7200rpm SATA
Asus ENGTX260 TOP/HDTP 896MB
Kingston 4GB kit (2x2) KVR1333D3N9K2
Intel Core 2 Quad Q9550 2.83 GHz
Gigabyte GA EP45T DSR S775 4xDDR3 (replaced this with asus P5QC however)
Lite-on iHAS120 20 DVDRW SATA

Unfotunately the system freezes completely randomly.

I would expect it to. You've got roughly 500-550W of hardware running off
of a 500W power supply. Well, to be more accurate, you've got so much draw
on the +12V rail that most 500W power supplies would not handle your
hardware. What amazes me is that it POSTS at all. Any 500W power supply
other than an -Earthwatts- 500W probably wouldn't even POST that system.

A 500W will barely handle it, assuming that most of that power is available
on the +12V lines. Otherwise, even a good name-brand power supply would be
under-sized at 500W.

Linux will help a bit, as it stresses your hardware less than Windows does.
In other words, running linux is probably the only way you can try to make
this system stable with a 500W Earthwatts PS.

Yes, I read that you seem to be stable with only one memory stick. That
COULD indicate you got a bad memory stick. But as it's Kingston we're
talking about, the odds of a bad memory stick are pretty low.

If it was my system, I'd upgrade the power supply to at least 650W, and
that's assuming a VERY good name-brand power supply. THEN I'd run memtest86
overnight two nights in a ROW, to test each memory stick individually.

I'm betting that you will find that the system is totally stable once you
get the proper sized power supply in it. NOTE that a bad or under-sized
power supply can cause symptoms that would seem to point to bad RAM or a bad
motherboard or CPU. And AT BEST, your current power supply is marginal, as
you are running it at ~80% or more of it's maximum rating on the +12V rail
constantly. (yikes) -Dave

 

[Ron AF Greve]

Hi,

Thanks, for the reply.

If you've nailed it to memory, no more freezes since, done deal -- buy
better quality or a complaint memory brand possibly others are
reporting as satisfactory. Also might explore the memory for a less
aggressive timing settings if that's an acceptable option. Not memory
and randomness continues -- then possibly another PS for a quick
double-check, and then onto grander schemes, like a MB RMA removal for
replacement. Had a couple "false starts" (error code that were
subsequently PWR on/off cycle resettable) from a FORTRON 300W-ish and
MSI/DURON, though an aging setup, nonetheless unacceptable. Dropped
the FSB back to stock, quit overclocking the little beggar, and then
changed out to a nicely reviewed, maybe Thermaltake 500W PS I'd
shelved from sometime last year's give-away-sell. Firstline packrat's
litany for stuff that comes in handy when troubleshooting. Next, if
necessary, I'll tear it down and totally redo the MB <> device
connects with CRC automotive contact cleaner. Most all my MBs are
builds from returns people can't figure, or minorly buggered,
acceptable MBs.

I thought it was the mobo too since it seemed I excluded all other things.
That's why I replaced the Gigabyte board with the asus one. After rebuilding
it I was sure the freeze was gone (so I thought) until I started oblivion it
frooze after a minute or so. It has run longer than it has before without
that second stick.

Yeah, maybe changing the timing might help. However I spent two weeks during
my vacation adjusting timings etc. on the gigabyte board with little luck
(some where better than others however the freezes were to often, sometimes
a few in a row, to use the system).

However I might just try the timings on the asus board, first want to see if
it is stable for a week or so.

Regards, Ron

 

[Flasherly]

If you've nailed it to memory, no more freezes since, done deal -- buy
better quality or a complaint memory brand possibly others are
reporting as satisfactory. Also might explore the memory for a less
aggressive timing settings if that's an acceptable option. Not memory
and randomness continues -- then possibly another PS for a quick
double-check, and then onto grander schemes, like a MB RMA removal for
replacement. Had a couple "false starts" (error code that were
subsequently PWR on/off cycle resettable) from a FORTRON 300W-ish and
MSI/DURON, though an aging setup, nonetheless unacceptable. Dropped
the FSB back to stock, quit overclocking the little beggar, and then
changed out to a nicely reviewed, maybe Thermaltake 500W PS I'd
shelved from sometime last year's give-away-sell. Firstline packrat's
litany for stuff that comes in handy when troubleshooting. Next, if
necessary, I'll tear it down and totally redo the MB <> device
connects with CRC automotive contact cleaner. Most all my MBs are
builds from returns people can't figure, or minorly buggered,
acceptable MBs.

 

[Dave]

Oddly enough (I forgot to mention) I did run memtest a few times for a few
runs and finally an overnight session together about 13 complete passes
using dual channel mode (both sticks installed) never got a single error
out of it (also did never freeze using memtest).

OK, now I'm nearly 100% certain it is the power supply. I'm assuming you
ran memtest off of a boot disk? That would stress your hardware very
little. And if your memory tests OK on memtest, it's probably good. The
only things you haven't eliminated are the CPU and the power supply. The
odds of a bad CPU? Maybe 1 in a million. Not worth considering. -Dave

 

[Flasherly]

I thought it was the mobo too since it seemed I excluded all other things.
That's why I replaced the Gigabyte board with the asus one. After rebuilding
it I was sure the freeze was gone (so I thought) until I started oblivion it
frooze after a minute or so. It has run longer than it has before without
that second stick.

Yeah, maybe changing the timing might help. However I spent two weeks during
my vacation adjusting timings etc. on the gigabyte board with little luck
(some where better than others however the freezes were to often, sometimes
a few in a row, to use the system).

However I might just try the timings on the asus board, first want to see if
it is stable for a week or so.

Regards, Ron

OK. I've run mostly ASUS or MSI, some ABIT, and still think highly of
ASUS brand (other than a live tech chatroom & some participants
alleging they'd worked for ASUS to cite deteriorating QC). Worldwide,
a few years ago, those three accounted for highest production figures
of sold motherboards. Although GIGABYTE is one I'd like to try, a
Johnny Come Lately for their solid-state series that have replaced
capacitors.

Hard for me to imagine you may be looking at 600-watt and higher class
power supplies, and someplace I'd as soon not go for pricing
comparisons. But you want a game box, or one also capable of giving
new game releases a run for their money, performance wise. ...so,
there goes that old cliche again: "make 'em or break 'em gaming
industry." VISTA, latest DirectX, and lord knows what else.

Take an older XP install, SP1, with no hint of a game, a preferred
barer, than not, configuration for stressing older program
compatibility (fancy as I'd go might be some video encoding), and
that's who I am. I might be able run your system complacently without
errors, it takes a newer generation of software possibly to reveal.

Actually, a video board regardless newer or high-wattage drawer
(though you've also mentioned a no-nonsense business-application board
Matrox), and three HDs (DVDs and whatnot) doesn't sound unreasonable.
There are also certainly worse quality case/ps packages than Antec
puts together.

I suppose I'd also have to go along with researching that PS for its
true rated power, at what efficiency level and percentage of deviation
it's designed, and who-all's credentials its listed for underwriting.
It may be of an "acceptable" package sort, Antec wouldn't expect to
under-perform for average starter-upper usage, even if a PS in a PC
industry sense can often be found a contradiction in terms.

 

[westom1]

I thought it was the mobo too since it seemed I excluded all other things.
That's why I replaced the Gigabyte board with the asus one. After rebuilding
it I was sure the freeze was gone (so I thought) until I started oblivion it
frooze after a minute or so. It has run longer than it has before without
that second stick.

You keep replacing parts on speculation. IOW you don't know what
the problem is, so you replace parts - called shotgunning.
Professionals who shotgun quickly find themselves unemployed.

Others posted classic answers based only on "we did this and for
some strange reason the failure stops". Now stop replacing things.
First discover what is wrong. For example, you had voltages from a
voltage monitor. But since you did not post those numbers, then the
few who actually know this stuff stayed silent.

That motherboard monitor must first be calibrated with a 3.5 digit
multimeter. And voltages are best measured only with a maximum load.
In your case, critically important VDC numbers are on the gray, green,
and purple wire both before and when power switch is pressed. And
then on any one of orange, yellow, and red wires when system is
operating. Those numbers are chock full of information you will not
learn about until posting them.

A power supply is like a house foundation. Everything else that is
good may act badly if the foundation is crumbling. Until you have
established a power supply 'system' (yes more than just a power
supply) as definitively good, then all other testing (or shotgunning)
becomes counter productive.

Also important are other facts - long before fixing anything - such
as the system (event) logs where Windows sees a problem, records it,
then works around it. What did those logs report?

More responsible computer manufacturers provide comprehensive
hardware diagnostics for free just for your problem. You don't have
those free diagnostics. So get diagnostics from each component
manufacturer or from third parties. The actual hardware that can
cause your failure is limited to CPU, some motherboard functions,
memory, sound card, video controller, and obviously the entire power
supply 'system'. Until a supply 'system' is confirmed - without doubt
using a multimeter - then other suspect can only remain suspected.

Other useful tools and concepts may be discussed after supply
'system' numbers are provided.

 

[Ron AF Greve]

OK, now I'm nearly 100% certain it is the power supply. I'm assuming you
ran memtest off of a boot disk?

Yes, sort of. I used to ubuntu install disk which also has a menu entry for
memtest and run it from there (so without any os loaded).

That would stress your hardware very little. And if your memory tests OK
on memtest, it's probably good. The only things you haven't eliminated are
the CPU and the power supply. The odds of a bad CPU? Maybe 1 in a
million. Not worth considering. -Dave

I have been working with the system and not seen any freezes anymore without
that one stick. (So it seems the CPU is ok). I agree that the memory seems
to be ok and it would be the PSU. Stilll somewhere it seems unlikely that
one memory stick is just too much for the psu. I wonder where the 1.5 V for
the memory actually comes from I quess it is made from the 5V rail?

Regards, Ron

 

[Ron AF Greve]

OK. I've run mostly ASUS or MSI, some ABIT, and still think highly of
ASUS brand (other than a live tech chatroom & some participants
alleging they'd worked for ASUS to cite deteriorating QC). Worldwide,
a few years ago, those three accounted for highest production figures
of sold motherboards. Although GIGABYTE is one I'd like to try, a
Johnny Come Lately for their solid-state series that have replaced
capacitors.

Hard for me to imagine you may be looking at 600-watt and higher class
power supplies, and someplace I'd as soon not go for pricing
comparisons. But you want a game box, or one also capable of giving
new game releases a run for their money, performance wise. ...so,
there goes that old cliche again: "make 'em or break 'em gaming
industry." VISTA, latest DirectX, and lord knows what else.

Take an older XP install, SP1, with no hint of a game, a preferred
barer, than not, configuration for stressing older program
compatibility (fancy as I'd go might be some video encoding), and
that's who I am. I might be able run your system complacently without
errors, it takes a newer generation of software possibly to reveal.

Actually, a video board regardless newer or high-wattage drawer
(though you've also mentioned a no-nonsense business-application board
Matrox), and three HDs (DVDs and whatnot) doesn't sound unreasonable.
There are also certainly worse quality case/ps packages than Antec
puts together.

Yes, the old matrox board is a pretty simple 2D board. Actually I was
surprised how fast it was in 2D, no problem displaying videos fullscreen,
although on my 1680x1050 resolution it could only display 16bits colors.
However I expect the board to drain very little current. So I figured it
would drain less power However I am not sure if that is using the same
voltage as the memory.

I suppose I'd also have to go along with researching that PS for its
true rated power, at what efficiency level and percentage of deviation
it's designed, and who-all's credentials its listed for underwriting.
It may be of an "acceptable" package sort, Antec wouldn't expect to
under-perform for average starter-upper usage, even if a PS in a PC
industry sense can often be found a contradiction in terms.

The PSU is an earthwatts it should have an efficiency of over 80%. So I
should get 400 Watts out of it. The test's I saw spoke highly of it.
http://www.hardwaresecrets.com/article/526/10

Note that the Sonata case is indeed a charm. You can slide and click most
stuff in place and its big fan cools the case very well (in my old pc's I
also had at least three fans. This one fan seems to do to same.

Regards, Ron

 

[Ron AF Greve]

You keep replacing parts on speculation. IOW you don't know what
the problem is, so you replace parts - called shotgunning.
Professionals who shotgun quickly find themselves unemployed.

Yes, but I don not work in a service center with all the measuring equipment
(at least not anymore ).
The only thing I got today is a analog multimeter. I did test the 5 and 12 V
with it. But like the onscreen display there was nothing odd about the
voltages.

Others posted classic answers based only on "we did this and for
some strange reason the failure stops". Now stop replacing things.
First discover what is wrong. For example, you had voltages from a
voltage monitor. But since you did not post those numbers, then the
few who actually know this stuff stayed silent.

The voltages are less than 20mV of there optimal values (like 11.98 for the
12V) . Although I don't know the exact spec that is unlikely to be a
problem. The only thing I don't know is the the accuracy of the onboard
monitor.

That motherboard monitor must first be calibrated with a 3.5 digit
multimeter. And voltages are best measured only with a maximum load.
In your case, critically important VDC numbers are on the gray, green,
and purple wire both before and when power switch is pressed. And
then on any one of orange, yellow, and red wires when system is
operating. Those numbers are chock full of information you will not
learn about until posting them.

A power supply is like a house foundation. Everything else that is
good may act badly if the foundation is crumbling. Until you have
established a power supply 'system' (yes more than just a power
supply) as definitively good, then all other testing (or shotgunning)
becomes counter productive.

Also important are other facts - long before fixing anything - such
as the system (event) logs where Windows sees a problem, records it,
then works around it. What did those logs report?

No, I checked the logs, nothing visible there unfortunaly (the only I could
find is that it complains it can't start the printer driver (I haven't got a
printer attached anyway).

More responsible computer manufacturers provide comprehensive
hardware diagnostics for free just for your problem. You don't have
those free diagnostics. So get diagnostics from each component
manufacturer or from third parties. The actual hardware that can
cause your failure is limited to CPU, some motherboard functions,
memory, sound card, video controller, and obviously the entire power
supply 'system'. Until a supply 'system' is confirmed - without doubt
using a multimeter - then other suspect can only remain suspected.

Other useful tools and concepts may be discussed after supply
'system' numbers are provided.

Theoratically you are right. Practically I would have to buy a digital
multimeter with logging capabilities. I don't know how much these cost
today, but I know when I worked for a service center of measuring equipment
the price tag for that came near the system I build. So you are right
theoratically but going that way is just not an option for me.

Regards, Ron

 

[westom1]

The voltages are less than 20mV of there optimal values (like 11.98 for the
12V) . Although I don't know the exact spec that is unlikely to be a
problem. The only thing I don't know is the the accuracy of the onboard
monitor.
Theoratically you are right. Practically I would have to buy a digital
multimeter with logging capabilities.

Buy a 3.5 digit multimeter for less than $20 in Kmart or Wal-mart or
any store that sells hammers. Measure those voltages. Report the
numbers. Get a reply that actually says something useful and
definitive. Your analog meter is a museum piece. Get the 3.5 digit
multimeter to also calibrate the motherboard voltage monitor. If
anything relevant is wrong with the power supply 'system', that meter
reading will say so immediately.

Long before those other suspects can be considered, first the supply
'system' must be known good - without doubt. Another erroneous
assumption was a logging meter. Posted was a multimeter sold even to
K-mart shoppers. A meter typically found where hammers are sold; for
about the same price. Find the problem; then fix it. Or as they say
in CSI: Follow the evidence. Get the multimeter. Valid numbers are
measured with power supply under maximum load.

Other facts - many supplies rated at 500 watt may really only be
350. To those who never first learned the numbers, the myths of 500
and 600 watt supplies are routinely promoted. Most computers rarely
consume more than 200 watts which is why an oversized 350 watt supply
is more than sufficient. IOW you have good reason to believe that
supply is more than sufficient. But even the best supply can be
defective which is why the 3.5 digit multimeter is still necessary.

 

[Guest]

I would expect it to. You've got roughly 500-550W of hardware running off
of a 500W power supply.

No, he doesn't. But it's possible his PS can't really put out nearly
its rated 500W for long.

 

[Guest]

Oddly enough (I forgot to mention) I did run memtest a few times for a few
runs and finally an overnight session together about 13 complete passes
using dual channel mode (both sticks installed) never got a single error out
of it (also did never freeze using memtest).

Try Memtest86 or Memtest+ instead. Memtest is horrible, as bad as
Docmem.

See if real-life reliability improves when only one memory stick is
plugged in at a time. Kingston is not high quality.

 

[Guest]

The PSU is an earthwatts it should have an efficiency of over 80%. So I
should get 400 Watts out of it. The test's I saw spoke highly of it.
http://www.hardwaresecrets.com/article/526/10

You said it was 500W supply, so it should be able to put out at least
500W, regardless of its efficiency, provided the efficiency isn't 0%.
The review you cited shows it capable of supplying 606W @ 47C. My
apologies for earlier doubting the quality of this supply.

The fact that the freezing occurred even with an older Matrox video
card indicates power is not inadquate.

 

[Dave]

The fact that the freezing occurred even with an older Matrox video
card indicates power is not inadquate.

Wrong. Swapping a video card might change the load on the +12V rail. But
that is only one of three voltage railes. If one of the other rails is
inadequate, then the power supply is inadequate. FYI, I believe the RAM
(which he's having a problem with) runs off the +5V rail. -Dave

 

[Dave]

No, he doesn't.

Well yes, in dumbed-down PSU marketing speak, he does. He's got enough
+12V rail draw to max out your typical power supply labelled "500W".

But it's possible his PS can't really put out nearly
its rated 500W for long.

Yup. -Dave

 

[Dave]

See if real-life reliability improves when only one memory stick is
plugged in at a time. Kingston is not high quality.

What do you consider high quality? Kingston is at least as good as crucial
or corsair or mushkin or ocz. Sometimes cheaper in fact, making it a better
value. -Dave

 

[Ron AF Greve]

Hi,

Try Memtest86 or Memtest+ instead. Memtest is horrible, as bad as
Docmem.

I believe the memtest that is on the ubuntu install disk is Memtest86+ V2.1
(from the top of my head). It certainly has the + (since I remember that is
flashing on the screen).

See if real-life reliability improves when only one memory stick is
plugged in at a time. Kingston is not high quality.

I tried reinstalling the second stick. It froze within 30 minutes. I removed
it everything is fine again. It apparently does have to do with that extra
stick (and consequently running in dual channel mode).

I wonder how thorough memtest86+ acutally is (does it also test dma (as
opposed to just poking and peeking different values in and from memory).

Regards, Ron

 

[Flasherly]

Yes, the old matrox board is a pretty simple 2D board. Actually I was
surprised how fast it was in 2D, no problem displaying videos fullscreen,
although on my 1680x1050 resolution it could only display 16bits colors.
However I expect the board to drain very little current. So I figured it
would drain less power However I am not sure if that is using the same
voltage as the memory.

The PSU is an earthwatts it should have an efficiency of over 80%. So I
should get 400 Watts out of it. The test's I saw spoke highly of it.http://www.hardwaresecrets.com/article/526/10

Note that the Sonata case is indeed a charm. You can slide and click most
stuff in place and its big fan cools the case very well (in my old pc's I
also had at least three fans. This one fan seems to do to same.

Regards, Ron

And that efficiency rating is an average over varying loads under some
PS new standard. Yes, the reviews are favorable for earthwatts, and
I've been to Antec on more than a few occasions over the years without
complaints. Though I prefer to research and buy my own PS units, if
somebody says 'do what it takes to build me one,' Antec is a likely
place to first start. Very.

The test link review you provided I determined is not a "real" test,
though --

http://www.sysopt.com/features/cases/article.php/12024_3660031_3

I like his better -- lacking an $50,000 industry PS test unit, he has
two MBs loaded up to the gills, spit off into your PS, and he couldn't
even bring it to its knees. Now that's a "torture" test. Which
brings up the possibility you're looking at an Antec QC slip -- you've
been specially signaled out and Antec sent you a faulty unit. Ha -
fat chance.

Statistics say one quarter of PC problems are PS related (at some
discounted variance from a PS just pulled out of the box new, such as
yours) -- so you've 3 stones to pry up and look beneath.

I recall one that came back, where I found (not easily but at the last
minute before ordering replacement parts, probably a "fix all" PS) I'd
missed a ground connection somewhere. Weird problem, (maybe I swapped
in a know good unit and clued into) pulling the MB and totally redoing
the entire assembly with a close eye to detail -- sporadic reboots,
locks, or whatever else it was doing, solved.

That and I'm a bit prejudiced towards anything beyond a very basic XP
-- inclusive of "make or break 'em" games, VISTA, and such. Took a
prybar and 5lb. mallet, actually, to make me let go of 98 and update
to XP.