New Build - Need Help!

[AreEss]

I have been setting the affinity in Prime95 everytime I've tried running two instances.

Weird. It may be Prime95 itself, then perhaps. I don't use Prime95 as a burnin or bench here - very counterproductive when you have the exact application and datasets that will be used.
Sadly, don't know anything decently multithreaded that could be used to properly test in your situation. :/

 

[Slartibartfast]

Weird. It may be Prime95 itself, then perhaps. I don't use Prime95 as a burnin or bench here - very counterproductive when you have the exact application and datasets that will be used.
Sadly, don't know anything decently multithreaded that could be used to properly test in your situation. :/

I don't use prime much either, but it was the first thing that came to mind for something that I could run two instances of. Maybe I'll check out something like worldbench. I'm not one for bragging about benchmark numbers, I just want something that's going to stress it.

 

[AreEss]

I don't use prime much either, but it was the first thing that came to mind for something that I could run two instances of. Maybe I'll check out something like worldbench. I'm not one for bragging about benchmark numbers, I just want something that's going to stress it.

http://www.spec.org/gpc/downloadindex.html

SPECviewPerf9 is particularly brutal on systems. If anything's gonna go wrong, SPECvp will show it.

 

[Slartibartfast]

The answer is....

I'm retarded.

Again.

All I had to do was set the cpu affinity via task manager, and leave it to "any" in Prime. Currently priming 100% both cpus, and the system is still usable. I knew I paid all that money for something I'm going to let it run for a while and see if anything happens *crosses fingers*

http://www.spec.org/gpc/downloadindex.html

SPECviewPerf9 is particularly brutal on systems. If anything's gonna go wrong, SPECvp will show it.

I will definitely check it out. I really want to brutalize this thing, and then I'll be confident that it's ok.

Today I shipped out the bad parts and put in an order for a Seagate 7800.10, 320gb 16mb cache. As soon as it arrives I'm going to partition it and see if I can't dual boot both xp's.

edit: well, the new thing is CPU1 is overheating. When I prime on it, it overheats. I reboot and hop into BIOS, and it's 8 - 10 degrees C hotter than CPU2. I scraped off the stock thermal compound and put some AS5 on it, but it's only helped a little bit (ie it takes a few minutes longer to overheat). The bottom of that HSF looks pretty scored, so I'm either going to see about lapping it or getting an aftermarket HSF.

Anybody know offhand if the XP-90 will fit in the H8DCE alright?

 

[AreEss]

The answer is....
I'm retarded.
Again.
All I had to do was set the cpu affinity via task manager, and leave it to "any" in Prime. Currently priming 100% both cpus, and the system is still usable. I knew I paid all that money for something I'm going to let it run for a while and see if anything happens *crosses fingers*

$50 says it's fine now. Definitely a bad DIMM by the codes. You have no idea how helpful that is in debugging a system that's misbehaving. Get MCE codes, run 'em, no guesswork. Right part gets replaced the first time, every time.

I will definitely check it out. I really want to brutalize this thing, and then I'll be confident that it's ok.
Today I shipped out the bad parts and put in an order for a Seagate 7800.10, 320gb 16mb cache. As soon as it arrives I'm going to partition it and see if I can't dual boot both xp's.

I don't believe you can, but I may be wrong. I don't do dual booting really, and I'm not very keen on XP still.

edit: well, the new thing is CPU1 is overheating. When I prime on it, it overheats. I reboot and hop into BIOS, and it's 8 - 10 degrees C hotter than CPU2. I scraped off the stock thermal compound and put some AS5 on it, but it's only helped a little bit (ie it takes a few minutes longer to overheat). The bottom of that HSF looks pretty scored, so I'm either going to see about lapping it or getting an aftermarket HSF.

Anybody know offhand if the XP-90 will fit in the H8DCE alright?

XP90 will fit, though I strongly recommend the SI120. Odd that CPU1's overheating though. This sounds like a case flow problem; not enough pull at the back or push at the front. Stock heatsinks don't overheat unless there's a real issue with the case itself; conflicting flow, bad flow, low pressure, etc. Can you get some photos of the install?

 

[Slartibartfast]

$50 says it's fine now. Definitely a bad DIMM by the codes. You have no idea how helpful that is in debugging a system that's misbehaving. Get MCE codes, run 'em, no guesswork. Right part gets replaced the first time, every time.

Yeah, that was really helpful. It's certainly a new experience working with parts that aren't from the bottom of the bargain bin.

I don't believe you can, but I may be wrong. I don't do dual booting really, and I'm not very keen on XP still.

Well, no harm in trying, right?

XP90 will fit, though I strongly recommend the SI120. Odd that CPU1's overheating though. This sounds like a case flow problem; not enough pull at the back or push at the front. Stock heatsinks don't overheat unless there's a real issue with the case itself; conflicting flow, bad flow, low pressure, etc. Can you get some photos of the install?

I will try to take some pictures, but the only digital camera I have access to re-defines "craptacular." Unless you're taking pictures in broad daylight, they're either too dark, or completely washed-out from the flash.

I was investigating this some more last night and I agree, I think it's an airflow problem. With this case, there are two 92mm intakes in the back right at the top (psu mounts on the bottom). For exhaust, there are two 120mm fans on top, towards the back (pic). So most of the airflow is in the upper-left portion of the case (assuming you're facing the mobo). So it cools CPU2 very well, but CPU1 is more towards the center of the case and away from that airflow. I ordered a hard disk cooler that mounts in a 5.25" bay, which blows air over a HD mounted in it. Just eyeballing it, it looks like I can mount it so that the air will be blowing right onto CPU1.

 

[AreEss]

Yeah, that was really helpful. It's certainly a new experience working with parts that aren't from the bottom of the bargain bin.

If it's a desktop, it's usually Abit though Asus is trying to woo me. (Still not as confident in their reliability under load.)

I will try to take some pictures, but the only digital camera I have access to re-defines "craptacular." Unless you're taking pictures in broad daylight, they're either too dark, or completely washed-out from the flash.

Don't have to be great pictures, just functional really.

I was investigating this some more last night and I agree, I think it's an airflow problem. With this case, there are two 92mm intakes in the back right at the top (psu mounts on the bottom). For exhaust, there are two 120mm fans on top, towards the back (pic). So most of the airflow is in the upper-left portion of the case (assuming you're facing the mobo). So it cools CPU2 very well, but CPU1 is more towards the center of the case and away from that airflow. I ordered a hard disk cooler that mounts in a 5.25" bay, which blows air over a HD mounted in it. Just eyeballing it, it looks like I can mount it so that the air will be blowing right onto CPU1.

Wow. That's the most IDIOTIC airflow I've ever heard. They SHIP it that way? Christ. Amazed that thing isn't burning up boards left and right.
Switch the 92mm fans to exhaust and the 120mm's to intake. (Check the directional indicators. Should be molded into the frame.)
Drive 120mm's should blow across the drives (intake.) The 92/120 setup is just horribly, horribly unbalanced, and whoever thought 2 92's as intake with 2x120 as exhaust was a good idea needs to go back to first grade math.
Your airflow path is maybe a few square inches in the top left corner (facing left side of the case.) That's it. That's IT. The 120mm's outflow the 92mm's horrifically. So two 92's are basically feeding a single 120mm. Don't think you'll be able to get that case into a positive air pressure situation because of all the holes in it. It'd actually work good in a positive pressure setup; aluminum cases don't act as a sink if you're dumping the air too fast. Positive pressure would distribute the hot air far more evenly, which allows the aluminum to absorb a bit more, and more importantly, makes all those holes actually do something.

 

[Slartibartfast]

Don't have to be great pictures, just functional really.

That's thing, sometimes even "functional" is pushing it I'll see what I can do.

Drive 120mm's should blow across the drives (intake.)

There's actually two more 120mm's in the bottom of the case. Unfortunately, these blow from the left side, over the HD cages, and through a grill on the right side, so their airflow doesn't contribute to the rest of the case. I will try reversing the fans, not sure how easily those top fans come out - I was looking at them last night and it was non-obvious.

 

[AreEss]

That's thing, sometimes even "functional" is pushing it I'll see what I can do.

Heh. Time for a new camera too, then.

There's actually two more 120mm's in the bottom of the case. Unfortunately, these blow from the left side, over the HD cages, and through a grill on the right side, so their airflow doesn't contribute to the rest of the case. I will try reversing the fans, not sure how easily those top fans come out - I was looking at them last night and it was non-obvious.

The ones in the bottom should be left as they are. That's suboptimal because it's a crossflow right in front of a front to rear (the PSU), so you get some really fucked up currents there. It's arguably better to push drive heat into the PSU/blankspace with a front-rear flow, but meh, no easy way to do that. I've been totally unimpressed with Silverstone's design efforts of late; it's like they're just slapping crap together with no real testing, and it's very annoying. They have good build quality and the cases look good. But if they're nonfunctional, like the TJ07 (no WAY that thing can handle 15KRPM SCSI-U320's safely,) then I can't use 'em!

 

[Slartibartfast]

Heh. Time for a new camera too, then.

Meh, I'm not much of a photographer anyway Hell if I got a new one it would probably get most of its use taking pics for the forum

The ones in the bottom should be left as they are. That's suboptimal because it's a crossflow right in front of a front to rear (the PSU), so you get some really fucked up currents there. It's arguably better to push drive heat into the PSU/blankspace with a front-rear flow, but meh, no easy way to do that. I've been totally unimpressed with Silverstone's design efforts of late; it's like they're just slapping crap together with no real testing, and it's very annoying. They have good build quality and the cases look good. But if they're nonfunctional, like the TJ07 (no WAY that thing can handle 15KRPM SCSI-U320's safely,) then I can't use 'em!

I don't think the NewEgg pics show it, but the drive cages have a very confined airflow. The drives aren't just mounted on some rails: the sides are actually completely enclosed, except for the screw holes. So all of the air being pushed by those bottom 120mm's is going right over the hd's and out the opposite side.

I was considering the Lian-Li v2100A Plus cases, but they just didn't seem to have enough room for the PSU - most people using that case and a PCP+C (my psu is just as deep) had to hack apart the drive cage or mod the psu just to get it to fit. I also read some complaints about the fact that the cages were riveted directly to the case, and hence amplified the drive vibration.

 

[AreEss]

Meh, I'm not much of a photographer anyway Hell if I got a new one it would probably get most of its use taking pics for the forum

Failing to see a problem!

I don't think the NewEgg pics show it, but the drive cages have a very confined airflow. The drives aren't just mounted on some rails: the sides are actually completely enclosed, except for the screw holes. So all of the air being pushed by those bottom 120mm's is going right over the hd's and out the opposite side.

That's exactly what it should do. Problem is that it's still taking away from the PSU's flow, and the heat that's picked up by the rails (which is significant) bottles up between, in front, and only gets pulled at the rear. Not enough of a connection with the aluminum to get wicking worth a damn.

I was considering the Lian-Li v2100A Plus cases, but they just didn't seem to have enough room for the PSU - most people using that case and a PCP+C (my psu is just as deep) had to hack apart the drive cage or mod the psu just to get it to fit. I also read some complaints about the fact that the cages were riveted directly to the case, and hence amplified the drive vibration.

They don't, and the V2100A's suck hard for airflow. Lian-Li really screwed those up bigtime. Very good negative pressure design up front, and a very bad design at the rear. There's not enough airflow at the top of the case, which is where all the heat ends up going. And you can't fit the 1KW without pulling the rear drive cages or using an extender bracket. I've heard LiLi changed them to screws in the later revisions, but really, screws amplify vibration even more. (Screws vibrate, fit isn't as tight, which amplifies the more audible mid-low vibration noises as opposed to the high.) Those things are, regardless, unbearable with 15K drives -before- you see how badly the rear drives are cooking off.

 

[Slartibartfast]

So just out of curiosity, what cases do you use for builds?

 

[AreEss]

So just out of curiosity, what cases do you use for builds?

LiLi PC73SLB with custom built drive modules.
LiLi PC626 with customization.
LiLi PC767 with custom built drive modules.
Yeong Yang YY-W2xx, Y0430, YY-0221.
Complete custom cases.

I used to use Enermax CS-BB series, but sadly, Enermax no longer builds decent EATX cases. (Or decent cases period, some might argue.)

 

[Slartibartfast]

Ok, I got all the fans switched around. And let me tell you, was it a huge pain in the ass. Those top two 120mm are held in this grill stuff that's attached by 10 tiny screws. I had to pull the mobo tray out, disassemble the motherboard tray rails, and flip the case over to pry the thing out. Then the two screws on the front short end wouldn't go back in, so sometimes it makes this rattling noise but anyways...

So I reversed the direction of the airflow, and it seems to have helped a tiny bit. As in, it takes longer for CPU1 to overheat I'm hoping that when I get that drive bay cooler it will make a difference, becaue it will be blowing air right over that proc. Honestly I think all I managed to do was make CPU2 run hotter

edit: ok I took some pictures, they suck but you'll get the idea. The lower segment of the machine (psu and drives) is cut off, but they don't provide much to look at anyway. They're really small but all I've got for now, I'll get some bigger ones later on (I just finished putting the thing back together and am not quite up for taking it apart again).

 

[AreEss]

Ugh.
You SHOULD have the flow, unless, surprise, the fans lie through their teeth on CFM. It's just not enough flow.
You're gonna have to get better 120's in the top, that'll solve it. A drive bay cooler isn't going to do a whit of difference here. The 92's as intake are just sucking in hot air from the PSU. (HINT TO DESIGNERS: Pick ONE and only ONE direction for airflow.) My recommendation would be to rig up a pair of Panaflo 1212L BX's in the top - they're 120x38's that do 70CFM at 30dBA. (Make sure you get ones with tach output!)
http://www.sidewindercomputers.com/pa12ulqu.html
Replace the 92's with 0912M1BX's. (Again; MAKE SURE YOU GET ONES WITH TACH OUTPUT!)
http://www.sidewindercomputers.com/pa92mesp.html

If those don't get the job done, THEN it's time for new heatsinks. Or a new case that doesn't suck. *sigh* That HUGE open area is the problem; too many places for the heat to go. In fact, better still, do this:
Grab some cardboard.
Cut it to fit in the back of the 5.25 bays.
Now tape it in the case so that it makes a wall between the 5.25" bays and the CPUs.
See how they do with that.

 

[Slartibartfast]

Alright, I'll give the cardboard a shot, see if that helps.

What about, changing the fans back to what they were (it was running cooler that way, right now I'm idling at 52 and 57), and making some sort of duct that will channel the hot air from the psu away? I have the tower right next to the couch, and I could easily channel that hot air over and under the couch, that way the 92mm intakes wouldn't be getting any of that hot air.

update: cardboard didn't help at all.

another update:
ordered some case fans. I got 2 of these:
http://www.frozencpu.com/fan-87.html?id=KTYaGi8U&mv_pc=648

and 2 of these:
http://www.frozencpu.com/fan-193.html?id=KTYaGi8U&mv_pc=1100

Those 120's have a bit more cfm than the fans you recommended, so hopefully they'll do the trick.

 

[AreEss]

Alright, I'll give the cardboard a shot, see if that helps.

What about, changing the fans back to what they were (it was running cooler that way, right now I'm idling at 52 and 57), and making some sort of duct that will channel the hot air from the psu away? I have the tower right next to the couch, and I could easily channel that hot air over and under the couch, that way the 92mm intakes wouldn't be getting any of that hot air.

update: cardboard didn't help at all.

Boggle. That's not right. I'm obviously missing something somewhere, that's causing the hotspot. As it stands, the 120mm's should be cycling enough to avoid the hotspotting, unless they're a lot weaker on air pressure than claimed. (Which wouldn't terribly surprise me.)
Can you try 120mm's as exhaust and 92mm's as exhaust, and see what that does? The only real cause I can think of here is air staying close to the board - the stock heatsinks blow downwards. However, the 120's should be pushing enough to prevent that level of stagnation on the board. Especially since IIRC, the stock heatsink orientation would have the open sides north-south, not east-west. If it's stagnant air near the board, then SI120's would obviously be vastly superior. (High rise, get them out into airflow.)

I think if setting all four fans to exhaust doesn't do the job, only real option's going to be SI120's or similar.

 

[Slartibartfast]

Boggle. That's not right. I'm obviously missing something somewhere, that's causing the hotspot. As it stands, the 120mm's should be cycling enough to avoid the hotspotting, unless they're a lot weaker on air pressure than claimed. (Which wouldn't terribly surprise me.)
Can you try 120mm's as exhaust and 92mm's as exhaust, and see what that does? The only real cause I can think of here is air staying close to the board - the stock heatsinks blow downwards. However, the 120's should be pushing enough to prevent that level of stagnation on the board. Especially since IIRC, the stock heatsink orientation would have the open sides north-south, not east-west. If it's stagnant air near the board, then SI120's would obviously be vastly superior. (High rise, get them out into airflow.)

I think if setting all four fans to exhaust doesn't do the job, only real option's going to be SI120's or similar.

heh, updated my previous post while you were posting this.

Yes you're right, the hsf's are aligned vertically, so I too would expect the 120's to push the air right through them. Of course, on the cpu that's overheating, the ram sticks are blocking some of the airflow into the hsf's.

I'm going to do some research into hsf's (never used aftermarket before) and in the meantime wait for those big honkin' new fans to arrive.

I'm also going to hit up the hardware store for some sandpaper - maybe I can shave off a few degrees if I lap the heatsink in question.

 

[AreEss]

heh, updated my previous post while you were posting this.

REALLY do not recommend the 92 H's or the 120 M's. Those are pretty darn loud.

Yes you're right, the hsf's are aligned vertically, so I too would expect the 120's to push the air right through them. Of course, on the cpu that's overheating, the ram sticks are blocking some of the airflow into the hsf's.

Otay, wasn't absolutely positive to be honest. I didn't get anywhere near enough sleep tonight, so..

I'm going to do some research into hsf's (never used aftermarket before) and in the meantime wait for those big honkin' new fans to arrive.

SI120's are what I use in the vast majority of my builds. They're excellent quality, very quiet, and work exceptionally well overall.

I'm also going to hit up the hardware store for some sandpaper - maybe I can shave off a few degrees if I lap the heatsink in question.

Possible, but I personally doubt it. Maybe 1-2C at the extreme. The stock heatsinks are pretty good. It's gotta be stagnant air. *sigh*

 

[Slartibartfast]

The case as-is is probably .25% as loud as my previous monstrosity, so I don't think it will bother me too much. I'll take a little extra noise for some extra airflow. I can always just turn the music up louder

I agree that it's stagnant air. Under load, cpu1 just keeps getting hotter and hotter, so obviously it needs to have more air pushed away from it.

As far as the SI120, I have a few questions. Does it hang over the edge of the mobo on cpu2 at all? (If I'm gunna get one, I'll probably get two so they match). I'm a tad worried about it fitting. Would I need to pull the standard heatsink plate off the back of the mobo to mount these things? (Never used an aftermarket hsf on a cpu before).

edit:
Here's a question:
The stock voltage on the 248 is 1.4, correct?

 

[AreEss]

The case as-is is probably .25% as loud as my previous monstrosity, so I don't think it will bother me too much. I'll take a little extra noise for some extra airflow. I can always just turn the music up louder

Heh, I'm just really vrather picky about the noise level I'll accept. I'm sensitive to white noise so it doesn't take much to bug me. (Fans above 33dBA or so are enough that I'll hear them over music.)

I agree that it's stagnant air. Under load, cpu1 just keeps getting hotter and hotter, so obviously it needs to have more air pushed away from it.

Yeah, and the TJ07 just has a piss-poor layout for it all in all. The way it's designed, there's no airflow over 75% of the board. It just cuts the corner.

As far as the SI120, I have a few questions. Does it hang over the edge of the mobo on cpu2 at all? (If I'm gunna get one, I'll probably get two so they match). I'm a tad worried about it fitting. Would I need to pull the standard heatsink plate off the back of the mobo to mount these things? (Never used an aftermarket hsf on a cpu before).

No, but I forget which way you have to turn it off the top of my head. It will be painfully obvious, because one way will pin it against the backplate. (That's the wrong way, obviously.) The SI120's are a little tight, but actually fit rather nicely on the H8DC's. Unfortunately, yes, you do have to replace the backplate because of the custom bracket for the SI120. So it may not be the best option for you, after all.
But, I have the answer; the Ultra-90. It's a vertical tower that eats a 90mm fan. They'd be mounted for front-rear airflow, though, which presents more than a bit of concern given the corner-cutting. They do however, use the stock backplate and are very good heatsinks. Certainly the equal of the SI120. The HR01 is not a good match due to weight and case layout.

edit:
Here's a question:
The stock voltage on the 248 is 1.4, correct?

http://www.amdcompare.com/us-en/opteron/details.aspx?opn=OSA248BLBOX
http://www.amdcompare.com/us-en/opteron/details.aspx?opn=OSA248BLWOF

So in reality, 55-60C isn't unacceptable. But it's certainly far higher than I recommend or allow average.

 

[Slartibartfast]

Thanks for the info, I would definitely prefer a hsf that won't require completely pulling out the mobo.

On the voltage: I'm not sure why this popped into my head today, but I *think* my system might be putting 2.0 volts into the cpu's. I seem to remember seeing that number in the bios, but not thinking anything of it because I was using the bios defaults. Fortunately the system is off right now, but I'll definitely look into it when I get home.

 

[AreEss]

Thanks for the info, I would definitely prefer a hsf that won't require completely pulling out the mobo.

On the voltage: I'm not sure why this popped into my head today, but I *think* my system might be putting 2.0 volts into the cpu's. I seem to remember seeing that number in the bios, but not thinking anything of it because I was using the bios defaults. Fortunately the system is off right now, but I'll definitely look into it when I get home.

Hrm, that is not right. Should be set on Auto. Bear in mind, the H8DCE's hardware monitoring is not perfect. However, 2.0V into the CPUs would definitely be wrong. Definitely worth looking into; that would decidedly cause overheating like that.

 

[Slartibartfast]

Hrm, that is not right. Should be set on Auto. Bear in mind, the H8DCE's hardware monitoring is not perfect. However, 2.0V into the CPUs would definitely be wrong. Definitely worth looking into; that would decidedly cause overheating like that.

Yeah, if I remember correctly and it is running at 2.0v then that's def. the cause. If so, let's hope the cpus didn't sustain any damage from having all that current in there. Just one of those things that I never thought to look at.

edit: looked up the Ultra-90, def. seems like a good choice. Sidewinder has it surprisingly cheap, too.

another edit: Voltage was fine, have no idea what I thought it might have been otherwise. I broke down and bought a pair of them there ultra 90's. It's only money, right? I'm going to get this freakin' thing cooled down if it kills me. Or my credit rating.

 

[Slartibartfast]

Ok so I got the new stuffs in.

The 120mm fan's don't fit apparently silverstone used abnormally thin fans for the top, which I never even noticed before (this is seriously the first case I've owned that didn't use all 80mm). So I'm not sure if I'm going to hang on to them or sell them (anybody interested? )

I installed the 92mm panaflo's on to the ultra 90 heatsinks, and that helped a lot - it's brought cpu temps down about 10 degrees - under full load I was only hitting the low 60's.

Installing those heatsinks was a blast, and by a "blast" I clearly mean a huge pain in the ass. I pulled out the motherboard try, scrubbed the cpu's down, got the heatsinks installed, and discovered that they wouldn't fit So I had to take them off, slide the tray in, mount the heatsinks, and then try to plug everything in. But hey, it's all good now.

So now all I have to do is wait for that last stick of RAM to show up (should be mon or tues) and this beast will finally be complete.

And then I start waiting/hoping for some nice 285 prices drops

Thanks to everybody for their assistance!

 

[unhappy_mage]

How are the temps with those? Stability a little better?

 

[AreEss]

Ok so I got the new stuffs in.
The 120mm fan's don't fit apparently silverstone used abnormally thin fans for the top, which I never even noticed before (this is seriously the first case I've owned that didn't use all 80mm). So I'm not sure if I'm going to hang on to them or sell them (anybody interested? )

I've got no use for 'em, sad to say. I just restocked my exotics on Friday. Don't think I have any 120x20's, but I'll check - might have some PetitAce's that would fit.

I installed the 92mm panaflo's on to the ultra 90 heatsinks, and that helped a lot - it's brought cpu temps down about 10 degrees - under full load I was only hitting the low 60's.

*sigh* Definitely stagnant air. Definitely. Remind me to beat the crap out of Silverstone. The TJ05 I think it is didn't have this problem. (Might be thinking TJ03, really not sure off the top fo my head.)

Installing those heatsinks was a blast, and by a "blast" I clearly mean a huge pain in the ass. I pulled out the motherboard try, scrubbed the cpu's down, got the heatsinks installed, and discovered that they wouldn't fit So I had to take them off, slide the tray in, mount the heatsinks, and then try to plug everything in. But hey, it's all good now.

*bites tongue clean off*

So now all I have to do is wait for that last stick of RAM to show up (should be mon or tues) and this beast will finally be complete.

And then I start waiting/hoping for some nice 285 prices drops

Screw that; wait for 258's.

Thanks to everybody for their assistance!

Never a problem. Be a real jerk if I didn't back up parts I recommended.

 

[Slartibartfast]

How are the temps with those? Stability a little better?

It's actually been pretty stable since I was able to identify and remove the old ram. There were some quirks getting x64 to cooperate with the soundcard drives, but I've got it successfully dual-booting both versions of win xp. I have no idea if that will ever be useful, but I like having the option. The cpu temps have definitely come down with the Ultra 90's. I may take off some of the 5.25" drive bay covers directly in front of CPU1 and see if having some more fresh air would help at all. These heatsinks work really well, because I have them both blowing towards the back, where the fans are blowing out. Additionally, the fins line up so that the 120mm's in the top are blowing right into the heatsinks.

I've been playing around more with the dual prime 95, and in x64 it still acts really, really weird (basically the way I described before). I am using supposed x64 version of P95, so maybe I'll try the regular one and see how it works.

*sigh* Definitely stagnant air. Definitely. Remind me to beat the crap out of Silverstone. The TJ05 I think it is didn't have this problem. (Might be thinking TJ03, really not sure off the top fo my head.)

Yeah, I've definitely been dissapointed with this case. Nothing fits together even remotely right and the airflow doesn't make sense.

I have a case here with a slight mod: I cut a piece of plexiglass to the size of two 5.25" bays, then cut a hole in it and mounted an 80mm blue led fan in it. It looks pretty sharp and really helps cool the case, as it blows in right over the cpus (dual celly, BP6). I'm thinking of doing something like that to the silverstone. The problem with that one though is that it's in the bottom two bays of the tower, so I could epoxy it in on three sides. On the silverstone I would have to mount it higher up, so there would be no support on the bottom. I don't know how well it would work.

*bites tongue clean off*

Don't forget to freeze it so they can sew it back on

Screw that; wait for 258's.

hehe, well actually before any of that I have to wait for my bank account to fill back up Is AMD planning on releasing any new socket 940 cpu's? I figured with AM2 just released and Socket F next quarter (last I heard anyway) that they would abandon 939/940.

 

[Asian Dub Foundation]

hehe, well actually before any of that I have to wait for my bank account to fill back up Is AMD planning on releasing any new socket 940 cpu's? I figured with AM2 just released and Socket F next quarter (last I heard anyway) that they would abandon 939/940.

prolly should only expect 1 more speed bump (ie 200mhz) for socket 940s. everything opteron will move to socket F (don't know about 1xx). can't wait to see the new boards. im drooling at prototype asus dual socket sli socket F boards

 

[Slartibartfast]

Ok, so I'm seeing these errors in the system log again, so far only in x64. Unlike last time, they're not popping up in the sys tray, they're just in there. Here is the mcat output:

Event Source 0 - WMIxWDM
Processor Number : 0
Bank Number : 4
Time Stamp (0x): 01C69196 0A60C9B8
Error Status (0x): A47E2100 00000A13
Error Address (0x): 00000000 7FFFFFF8
Single bit errors:
Error found by scrubber
Uncorrectable ECC error
Error address valid in MCi_ADDR
Error not corrected
Error valid
Bus Error Code:
Participation processor: Local node responded to the request (RES)
Time-out: Request did not time out
Memory transaction type: Generic read (RD)
I/O: DRAM memory access (MEM)
Cache level: Generic (LG)
North Bridge Error MC4:
Extended Error Code: 0x0 - ECC Error
Error Code: 0x0A13
DRAM memory access (MEM) Generic read (RD), on Generic (LG) cache
Uncorrectable Multi-bit ECC Error
Error address at 2047 MB

---

Event Source 1 - WMIxWDM
Processor Number : 0
Bank Number : 4
Time Stamp (0x): 01C691CB B20DE422
Error Status (0x): A47E2100 00000A13
Error Address (0x): 00000000 7FFFFFF8
Single bit errors:
Error found by scrubber
Uncorrectable ECC error
Error address valid in MCi_ADDR
Error not corrected
Error valid
Bus Error Code:
Participation processor: Local node responded to the request (RES)
Time-out: Request did not time out
Memory transaction type: Generic read (RD)
I/O: DRAM memory access (MEM)
Cache level: Generic (LG)
North Bridge Error MC4:
Extended Error Code: 0x0 - ECC Error
Error Code: 0x0A13
DRAM memory access (MEM) Generic read (RD), on Generic (LG) cache
Uncorrectable Multi-bit ECC Error
Error address at 2047 MB

---

Event Source 2 - WMIxWDM
Processor Number : 0
Bank Number : 4
Time Stamp (0x): 01C69230 82E979E4
Error Status (0x): A4092100 00000A13
Error Address (0x): 00000000 7FFFFD38
Single bit errors:
Error found by scrubber
Uncorrectable ECC error
Error address valid in MCi_ADDR
Error not corrected
Error valid
Bus Error Code:
Participation processor: Local node responded to the request (RES)
Time-out: Request did not time out
Memory transaction type: Generic read (RD)
I/O: DRAM memory access (MEM)
Cache level: Generic (LG)
North Bridge Error MC4:
Extended Error Code: 0x0 - ECC Error
Error Code: 0x0A13
DRAM memory access (MEM) Generic read (RD), on Generic (LG) cache
Uncorrectable Multi-bit ECC Error
Error address at 2047 MB

---

Event Source 3 - WMIxWDM
Processor Number : 0
Bank Number : 4
Time Stamp (0x): 01C69252 B59FA530
Error Status (0x): A479A100 00000A13
Error Address (0x): 00000000 7FFFFF38
Single bit errors:
Error found by scrubber
Uncorrectable ECC error
Error address valid in MCi_ADDR
Error not corrected
Error valid
Bus Error Code:
Participation processor: Local node responded to the request (RES)
Time-out: Request did not time out
Memory transaction type: Generic read (RD)
I/O: DRAM memory access (MEM)
Cache level: Generic (LG)
North Bridge Error MC4:
Extended Error Code: 0x0 - ECC Error
Error Code: 0x0A13
DRAM memory access (MEM) Generic read (RD), on Generic (LG) cache
Uncorrectable Multi-bit ECC Error
Error address at 2047 MB

---

Event Source 4 - WMIxWDM
Processor Number : 0
Bank Number : 4
Time Stamp (0x): 01C69285 30C2132E
Error Status (0x): A479A100 00000A13
Error Address (0x): 00000000 7FFFFF38
Single bit errors:
Error found by scrubber
Uncorrectable ECC error
Error address valid in MCi_ADDR
Error not corrected
Error valid
Bus Error Code:
Participation processor: Local node responded to the request (RES)
Time-out: Request did not time out
Memory transaction type: Generic read (RD)
I/O: DRAM memory access (MEM)
Cache level: Generic (LG)
North Bridge Error MC4:
Extended Error Code: 0x0 - ECC Error
Error Code: 0x0A13
DRAM memory access (MEM) Generic read (RD), on Generic (LG) cache
Uncorrectable Multi-bit ECC Error
Error address at 2047 MB

---

Event Source 5 - WMIxWDM
Processor Number : 0
Bank Number : 4
Time Stamp (0x): 01C69298 202E2B3E
Error Status (0x): A40EA100 00000A13
Error Address (0x): 00000000 7FFFFEB8
Single bit errors:
Error found by scrubber
Uncorrectable ECC error
Error address valid in MCi_ADDR
Error not corrected
Error valid
Bus Error Code:
Participation processor: Local node responded to the request (RES)
Time-out: Request did not time out
Memory transaction type: Generic read (RD)
I/O: DRAM memory access (MEM)
Cache level: Generic (LG)
North Bridge Error MC4:
Extended Error Code: 0x0 - ECC Error
Error Code: 0x0A13
DRAM memory access (MEM) Generic read (RD), on Generic (LG) cache
Uncorrectable Multi-bit ECC Error
Error address at 2047 MB

---

Event Source 6 - WMIxWDM
Processor Number : 0
Bank Number : 4
Time Stamp (0x): 01C692F0 2F5B019C
Error Status (0x): A416A100 00000A13
Error Address (0x): 00000000 7FFFFDE8
Single bit errors:
Error found by scrubber
Uncorrectable ECC error
Error address valid in MCi_ADDR
Error not corrected
Error valid
Bus Error Code:
Participation processor: Local node responded to the request (RES)
Time-out: Request did not time out
Memory transaction type: Generic read (RD)
I/O: DRAM memory access (MEM)
Cache level: Generic (LG)
North Bridge Error MC4:
Extended Error Code: 0x0 - ECC Error
Error Code: 0x0A13
DRAM memory access (MEM) Generic read (RD), on Generic (LG) cache
Uncorrectable Multi-bit ECC Error
Error address at 2047 MB

 

[AreEss]

mage, can you decode? I'm not even at home right now so no MCAT.

Best guess off the top of my head, driver being stupid and trying to read past end of memory.

 

[Slartibartfast]

mage, can you decode? I'm not even at home right now so no MCAT.

Best guess off the top of my head, driver being stupid and trying to read past end of memory.

that was the mcat output from the .evt file. Are you looking for something else?

edit: here's the actual event from x64

Event Type: Error
Event Source: WMIxWDM
Event Category: None
Event ID: 122
Date: 6/18/2006
Time: 12:00:31 PM
User: N/A
Computer: ID
Description:
Machine Check Event reported is a fatal Bus or Interconnect error.
Memory Hierarchy Level: 3
Participation: 1
Request Type: 1
Memory/IO: 0
Address: 2147483112

For more information, see Help and Support Center at http://go.microsoft.com/fwlink/events.asp.
Data:
0000: 00000001 00000001 2f5b019c 01c692f0
0010: 00000000 00000000 00000004 00000000
0020: 00000a13 a416a100 7ffffde8 00000000
0030: 00000000 00000000 00000000 00000000
0040: 00000000 00000000 00000000 00000000
0050: 00000000 00000000 00000000 00000000
0060: 00000000 00000000 00000000 00000000
0070: 00000000 00000000 00000000 00000000
0080: 00000000 00000000 00000000 00000000
0090: 00000000 00000000 00000000 00000000
00a0: 00000000 00000000 00000000 00000000
00b0: 00000000 00000000 00000000 00000000
00c0: 00000000 00000000 00000000 00000000
00d0: 00000000 00000000 00000000 00000000
00e0: 00000000 00000000 00000000 00000000
00f0: 00000000 00000000 00000000 00000000

and again with the Data set to Bytes:

Event Type: Error
Event Source: WMIxWDM
Event Category: None
Event ID: 122
Date: 6/18/2006
Time: 12:00:31 PM
User: N/A
Computer: ID
Description:
Machine Check Event reported is a fatal Bus or Interconnect error.
Memory Hierarchy Level: 3
Participation: 1
Request Type: 1
Memory/IO: 0
Address: 2147483112

For more information, see Help and Support Center at http://go.microsoft.com/fwlink/events.asp.
Data:
0000: 01 00 00 00 01 00 00 00 ........
0008: 9c 01 5b 2f f0 92 c6 01 œ.[/ð’Æ.
0010: 00 00 00 00 00 00 00 00 ........
0018: 04 00 00 00 00 00 00 00 ........
0020: 13 0a 00 00 00 a1 16 a4 .....¡.¤
0028: e8 fd ff 7f 00 00 00 00 èýÿ....
0030: 00 00 00 00 00 00 00 00 ........
0038: 00 00 00 00 00 00 00 00 ........
0040: 00 00 00 00 00 00 00 00 ........
0048: 00 00 00 00 00 00 00 00 ........
0050: 00 00 00 00 00 00 00 00 ........
0058: 00 00 00 00 00 00 00 00 ........
0060: 00 00 00 00 00 00 00 00 ........
0068: 00 00 00 00 00 00 00 00 ........
0070: 00 00 00 00 00 00 00 00 ........
0078: 00 00 00 00 00 00 00 00 ........
0080: 00 00 00 00 00 00 00 00 ........
0088: 00 00 00 00 00 00 00 00 ........
0090: 00 00 00 00 00 00 00 00 ........
0098: 00 00 00 00 00 00 00 00 ........
00a0: 00 00 00 00 00 00 00 00 ........
00a8: 00 00 00 00 00 00 00 00 ........
00b0: 00 00 00 00 00 00 00 00 ........
00b8: 00 00 00 00 00 00 00 00 ........
00c0: 00 00 00 00 00 00 00 00 ........
00c8: 00 00 00 00 00 00 00 00 ........
00d0: 00 00 00 00 00 00 00 00 ........
00d8: 00 00 00 00 00 00 00 00 ........
00e0: 00 00 00 00 00 00 00 00 ........
00e8: 00 00 00 00 00 00 00 00 ........
00f0: 00 00 00 00 00 00 00 00 ........
00f8: 00 00 00 00 00 00 00 00 ........

 

[unhappy_mage]

that was the mcat output from the .evt file. Are you looking for something else?

That's what mcat shows:

C:\Program Files\AMD\MCat>mcat 0x01C691960A60C9B8 0xA47E210000000A13 0x000000007FFFFFF8
Processor Number  : 0
Bank Number       : 0
Time Stamp    (0x): 00000000 00000000
Error Status  (0x): A47E2100 00000A13
Error Address (0x): 00000000 00000000
Single bit errors:
   Error found by scrubber
   Uncorrectable ECC error
   Error address valid in MCi_ADDR
   Error not corrected
   Error valid
Bus Error Code:
   Participation processor: Local node responded to the request (RES)
   Time-out: Request did not time out
   Memory transaction type: Generic read (RD)
   I/O: DRAM memory access (MEM)
   Cache level: Generic (LG)
Data Cache Error MC0:
   System line fill error into data cache
   Syndrome: 0xFC
   Error was detected on a scrub

 

[AreEss]

Sometimes Windows XP Pro 64 misreports, so I always verify with MCAT. You can see the differences in what they say here. Windows is saying reading past end of memory; MCAT's saying it was caught by the scrubber and prevented from causing a serious issue.

In this particular case, I'd say it's most likely a driver with something akin to an off-by-one which is trying to read past end of memory.

 

[Slartibartfast]

Sometimes Windows XP Pro 64 misreports, so I always verify with MCAT. You can see the differences in what they say here. Windows is saying reading past end of memory; MCAT's saying it was caught by the scrubber and prevented from causing a serious issue.

In this particular case, I'd say it's most likely a driver with something akin to an off-by-one which is trying to read past end of memory.

So is this something worth worrying about? It hasn't done it in a while and it didn't seem to affect the system at all.

 

[AreEss]

So is this something worth worrying about? It hasn't done it in a while and it didn't seem to affect the system at all.

Only as a long-term stability issue. Can't say for sure what driver's doing it - though it's probably nForce again. As long as the system isn't blowing up about it, I wouldn't worry too much. If you can find a reproduction case though, that'd be helpful.

 

[Slartibartfast]

Only as a long-term stability issue. Can't say for sure what driver's doing it - though it's probably nForce again. As long as the system isn't blowing up about it, I wouldn't worry too much. If you can find a reproduction case though, that'd be helpful.

Alright, I'll keep an eye on it. The thing is I got no indication, I just happened to be looking in the event viewer and noticed it.

Damn windows

 

[Slartibartfast]

Well, after abusing the system in both versions of xp, that error only ever comes up in x64 and not very frequently at that, so I'm not going to worry about it.

On a positive note, my rma stick of ram came in the other day and works fine. Dual channel gives the system a little boost.

Of course, XP 32, in all it's glory, actually sees less ram now

 

[Slartibartfast]

Ok, so after playing around a lot with different airflow configurations, as well as re-seating the Ultra 90's a few times, I just can't get the cpu's to idle under 50c. Now it could have something to do with the fact that it's usually pretty hot in the apartment, but I'm wondering if the mobo is just not reporting temps accurately, the reason being that no part of the hsf feels even remotely warm. Even the base where it contacts the cpu feels colder than the air surrounding it. I don't have any sort of thermometer immediately available to see what it's reading but I will keep my eye out for one.

I'm more concerned about the south bridge - that heatsink always feels very hot, even with fresh air blowing through it.

 

[AreEss]

Ok, so after playing around a lot with different airflow configurations, as well as re-seating the Ultra 90's a few times, I just can't get the cpu's to idle under 50c. Now it could have something to do with the fact that it's usually pretty hot in the apartment, but I'm wondering if the mobo is just not reporting temps accurately, the reason being that no part of the hsf feels even remotely warm. Even the base where it contacts the cpu feels colder than the air surrounding it. I don't have any sort of thermometer immediately available to see what it's reading but I will keep my eye out for one.

Hrm. That's concerning and a half. The H8DCE uses the on-die, so the CPUs may not be reporting correctly themselves. However, I did some poking around. You have RevF's there, I'd wager. They're the latest and greatest, and would explain the memory errors previously. The F's got a new register for uncorrected ECC error counting, and predictably XP64 mishandles the register.

I'm more concerned about the south bridge - that heatsink always feels very hot, even with fresh air blowing through it.

Never, ever, ever gets any better. I'm still working on a waterblock to handle the H8DCE's chipsets (way easier to get both at once just because of the layout.) It'll stay in operating range, but yeah, it runs HOT.