My first AMD

[pascal]

This is the first time I build an AMD system. My goal is to have a stable, quiet and cool system for home office and casual gaming

The configuration is below:
- AMD Athlon XP2100+ OEM AXDA2100DUT3C (TBred B) *
- GlacialTech Diamond 2000 (2500 ~ 4500 rpm) *
- Asus A7N8X rev. 2.0 BIOS 1005 *
- 2x256MB DDR266 Crucial *
- Asus V8200T2 Deluxe (GF3Ti200)
- SB Live and pair of generic speakers
- Maxtor 40GB 7200rpm
- TDK velocd 16/10/40x CDRW
- generic 400W PSU *
- generic midtower case *
- generic 2500rpm 80mm fan in the back of the case (negative pressure) *
- WinXP Pro

* new itens

It was a cheap upgrade and some itens I already had in my previous system (P3-S, tusl2-c). I am not overclocking and I am using conservative BIOS settings. It is working great. I burned some CDs, tested USB pen drive 1.1, lan is ok, UT2003 is great, folding is fast

I got used to the silent, cool P3-S also I live in a hot country and the system temperature is a litle high for me, specially the disk. The CPU temperature is 50c with the cpu fan at 3400rpm (Asus Probe measures with QFan enabled) after some hours of full load folding. The disk is hot to touch. I will probably move the disk for some 5"1/4 position and maybe try the HD cooler.

Any advice/comments are wellcome. Thanks

 

[bystander]

Have you seen this site?. They sell a lot of components specifically designed to quiet down your PC. Some of the products are pricy (there PSUs are particularly expensive), but its a good site to see what options/products are available.

Edit: Couldn't quite get the formatting correct.

 

[Gollum]

Looks like a nice system, although the video card and HD look slightly underspecced compared to the rest of the setup. OTOH it probably was pretty cheap to assemble these components and if you want it silent, a passively cooled Ti200 isn't exactly the worst choice. As for the HD getting too hot, a HD cooler can help at the expense of some additional noise. Alternatively you can try moving it to a different slot that offers the drive a better airflow.
Have fun with your new system!

 

[pascal]

Thanks for the site bystander
I have thinking about the Zalman coolers.

Gollum, The HD is enough for what I do and the GF3Ti200 is fast enough for most games (UT2K3, U2, etc..). Usually the bottleneck is the CPU. Probably this will change with Doom3, but this card was expensive one year ago. Thanks for the advices

I tried something unusual like underclocking without success, the idea was reduce the speed to 1.6Ghz (XP1900) and the vcore to 1.5v and then reduce the heat generation in 20% (in the level of P4 2Ghz) but I could not reduce the vcore!

Also I cannot find any Athlon XP1900 Thoroughbred anywhere. Why ? :?

 

[BenSkywalker]

With a negative airflow inside the case you are creating turbulence which is bad for cooling, if you want imbalanced airflow it is better to go positive then negative. That said, I would consider picking up a matching fan for the one in the back and adding it to the front of the case, should drop you temps by at least a few degrees.

You say you have a generic case, which type exactly? Is it one of the Chenming based ones(Antec, Alienware style)? If not, is it a decent case in terms of offering places for fans etc? I would certainly reccomend a hard driver cooler for you, it sounds like your HD is running way too hot which can lead to reliability issues. Do you set your drive to spin down after x amount of time? You may want to try it if you aren't already, it can help reduce ambient temps.

What are your ambient temps at, both room and case?

For your HS/F, I'm trying to recall and isn't that a 70mm solution? To really reduce your temps you can pick up a ThermalRight SLK800A and pair it with a low speed 80mm fan. If you do that with a good thermal paste your core temps should drop considerably and your noise level can be reduced by quite a bit also. Combine a SLK800A with a decent 80mm fan, a HD bay cooler and a front case fan to compliment the rear and all of your problems should be cleared up.

 

[epicstruggle]

cut-n-paste from the thread this question came up.

Some simple things you could do, without know how you placed the components, I would suggest giving more space for the hard drive, removing the covers from the pci covers, or even just running the pc with the case off. There are more expensive solutions like adding a HD cooler, using a second fan to increase air flow. Try also looking at whether or not your cables are blocking airflow.

let me know if anything works,

also a picture of the inside of your case might help us see any major problems with componet placement.

later,

 

[Snyder]

CPU: 50°C after hours of folding? Thats quite a good value - you shouldn't be worried about that.

HDD: Well, "hot to the touch" is quite subjective. Is it so hot you burn your fingers when touching it, or just "very warm"? (Eeehm... 8) ) Putting it into a 5,25" slot is always a good idea, though.

 

[pascal]

I sold some old cpu/cooler/mobo/mem for $200 and bought/import new ones from newegg for only $335 (including shipping expenses)
It was supposed to be a cheap/smart upgrade waiting for something better next year.

Ben, my new case is just a cheap and simple generic midtower case with PSU I bought locally and it doesnt have a front place to put the fan Epic, I will try to have an inside/outside pictures tonight.

The Glacialtech cooler is made of aluminiun and was really cheap ($10). The 70mm fan is controlled by the mobo (Asus Qfan) with variable speed from 2500 (quiet) to 4500 rpm (some noise). The cooler pdf: http://www.glacialtech.com.tw/pdf/Diamond 2000.pdf

Ben, the room temperature in my town varies during the year from 22c to 33c (never too cold, never too hot). This morning it was 26c for room temperature, 31c for mobo and 42c for CPU and CPU fan at 2400rpm without folding. It looks good, but once I start folding the CPU go to 50c and the fan go to CPU 3400rpm. I use Asus Probe to measure speed/temp.

When I first build the system it had no back case fan and the CPU was at 50c with CPU fan at 4400rpm while folding. Then the back case fan helped reduce the CPU fan speed by 1000rpm.

I cannot keep the case open because other people will use it in my home. Also I dont like air conditioner.

Snyder, maybe I am overreacting about the CPU, but the HD is really hot and is my priority to cool it. Being more objective I can not keep my finger on it more than 4 seconds without burning it. Definitelly it is much hotter than the CPU cooler and I agree with Ben that it can lead to some reability issues. I cannot keep folding 24/7 this way.

I did not configured the HD to spin down. I will try it tonight. Also I am waiting some parts one friend in Brazil is sending me like HD cooler and some 5,25" slot adpaters then I will try to move the HD to other places.

After that I will try other things like change case, roundcables, new cooler.

Anything I buy from US will take 4 weeks to be in my hands. I think I will have to import some itens.

Thanks all

 

[Basic]

ASUS Probe doesn't actually measure the CPU core temp. It measures the temp at the motherboard under the CPU. The dfference can be >10º C.

MBM5 can show temperatures for motherboard (ambient), motherboard (under CPU), CPU (actual core temp), and HD temp if the HD supports it (but it seems like that reading is of dubious quality).

If you want to cool the HD better, the most important part is that it sits in some airflow. It doesn't have to be some hefty flow from a fan dedicated to it, but make sure that the normal airflow through the case passes the HD. At least that's enough for my Maxtor DiamondMaxPlus9 7200 rpm. (MBM5 says that it's 25ºC, but 35ºC is probably more like it.) It's warm, but definitely not hot. And within the max 55ºC with a margin.

I don't know if it helps (or if it shows through that I just want to show my own odd experiment ), but here's how I've done the cooling in my box.

Temps when folding:
Room temp: 26ºC
MB ambient: 30ºC
MB under CPU: 44ºC (This is the same as ASUS Probe shows.)
CPU core: 56ºC
HD: 25ºC according to MBM5, but 35ºC is more like it.

PS
Round cables are for amateurs. A true origami master can make standard flat cables less obstructive for airflow than rounded cables. And it'll look nicer too.

 

[pascal]

WOW Basic, what nice work you have done
Now I understand how you can keep it folding 24/7.

Home computers are a thermal and acoustic engineering artwork

Then my conclusion is that my CPU temp is probably around 60c when folding. I will download MBM5 and try it. Probably it is time for a better cooler.

My HD is a Maxtor DiamondPlus 8 and I will move it before the end of the week (still waiting the 5.25" adapter).

You probably divided the cables four wires wide, right?

thanks,

 

[pascal]

Basic,
The MBM5 measure the following:

A7N8X rev 2, no load
Asus 2 = case sensor = 31c
Asus 1 = socket sensor = 47c
W83L785TS-S Diode = CPU diode = 43c

Is the CPU colder than the socket?

 

[BenSkywalker]

Basic

I think you have a mistake on your cooling page-

2x Papst 4412FGL, 12 mm,

Either you have a typo there or you have some seriously miniaturized parts in your rig

Pascal-

With an under die reading of 50C you are likely at 60C or higher for your core temps. The readings you just posted are wrong, there is no way your core is cooler then your under die(program is reading something wrong, perhaps on purpose). I have a mobo that supports core temp readings(MSI K7N2 not using the latest BIOS that screws the temps up) and also have a thermal probe underneath my processor and right now with my fan @3800 RPMs my under die temp is 47.6C while my core temp is 59C(mine is OCd decently TBred B 2100XP@2700XP).

Can you post your load temps so we can see where they are at?

I'm at the opposite end of the spectrum compared to Basic, I have a ThermalTake XaserIII case with 12fans total(including CPU, GPU, PSU) and noise isn't too much of a concern for me(with my setup, I'd go nuts if it was ). My big problem is my rig is used for gaming an average of 8-9 hours a day which kills my folding in relation to Basic....

 

[pascal]

Temperature after hours folding:

Asus 2 = case sensor = 31c
Asus 1 = socket sensor = 50c
W83L785TS-S Diode = CPU diode = 47c

Probably something is wrong. It is better/safer assume the CPU is 10c above the socket.

 

[BenSkywalker]

No increase in ambient and 3C on the core? I wouldn't sweat it in the least. Your likely pushing about 60C, the max core temp for the TBred B 2100 is 90C IIRC(although you really don't want to test that out ). What kind of thermal paste did you use? I bring it up mainly as the difference in your temps comparing load/idle isn't much higher then comparing a OEM style heatpad with ArcticSilver, it's rather insignificant(and an indication that you can likely lower your fan a bit more without worrying).

I'd say take care of your HD temps and then not worry about it(the heat that is). If you plan on OCing your core to get folding a bit faster, you may want to spring for a bit better HS/F(pure cu, with an 80mm fan would be best), but for right now I'd turn down your fan a bit more and see what kind of temps you get if noise is an issue.

 

[Basic]

pascal:
Thanks, I did some time thinking before I settled for that idea. At one time I was thinking about doing a full custom case, using big airducts with noise traps. But settled for this rather easy way.

I haven't split any cables, just folded them so they are flat against the back panel, out of the way from any airflow. And where they go out from the back panel to the drives, they are folded so they show their small edge against the airflow (meaning that it has a smaller cross section than a rounded cable.) A few rubber bands can help.

CPU diode cooler than the socket sensor? Yes, that's indeed strange. I've heard that CPU diodes might have some variance, but that's far too much. I know there might be a problem if you run two programs that checks the sensors. First time I noticed it was when I logged in as two users concurrently in WinXP, and both autostarted an ASUS Probe. All voltages and temperatures went haywire, and it scared the shit out of me.
Did you run ASUS Probe at the same time? In that case, does it make a difference if you turn it off?


Ben:
No no, it is 12mm somewhere. I didn't say what part was 12mm. (Edited)

About not being bothered by the noise. Just be carefull. Even if it doesn't bother you, it can still give you tinnitus, and then it sure will. If you're constantly exposed to the high pitched noise from fans and not so good HDs, it can be as damaging as occational exposure to much louder noise.

And yes, you get more gaming out of your rig. My day average is probably below 4 hours/day. But it tends to go in bursts. Lots of days with no gaming at all, and then totally freak out and sit longer than what's healthy.

 

[pascal]

Ben,
I am starting to get used to this new computer and its heat generation. Probably I dont have to worry about the CPU for now. I am using the thermal paste that comes with the HSF, but the fan has variable speed and with folding the rpm increase almost 1000.

I agree about the HD. No I dont plan to overclock.

Basic,
The Asus Probe is installed but not running. Tonight I will uninstall it to see what happens.

Anyway, this new AMD is really fun
Thanks all folks

 

[BenSkywalker]

Pascal-

I am starting to get used to this new computer and its heat generation. Probably I dont have to worry about the CPU for now. I am using the thermal paste that comes with the HSF, but the fan has variable speed and with folding the rpm increase almost 1000.

Can you wire around the auto speed adjusting? I have a ThermalTake SmartFanII and have it hooked to a manual controller on the front of my rig, makes adjusting for optimal heat/noise a lot easier

Basic-

Width of the fan blades....?

About not being bothered by the noise. Just be carefull. Even if it doesn't bother you, it can still give you tinnitus, and then it sure will. If you're constantly exposed to the high pitched noise from fans and not so good HDs, it can be as damaging as occational exposure to much louder noise.

It's not really all that loud, although you certainly notice when you shut it off I'm currently keeping my CPU fan @3400 RPMs which isn't close to as loud or as high pitched as it is when I crank it up, my case fans loaf along somewhere around 2500RPMs, I'd turn them down a bit lower but then they don't light up as much(my rig is quite 'riced out' ).

And yes, you get more gaming out of your rig. My day average is probably below 4 hours/day. But it tends to go in bursts. Lots of days with no gaming at all, and then totally freak out and sit longer than what's healthy.

Are you still running it @2400XP speeds? I don't think my wife has been gaming with the rig too much lately so my figure for average gaming time should hold, but it seems like you are running at nearly double my speed lately. I did just pick up WarCraftIII last week and that has been sucking up a lot of my time, but I wouldn't think it would be enough to double up over my rig.

Hmmm, what are you system settings at? Multiplier/FSB RAM timings etc? Been trying to figure out how to get my rig to fold faster, was thinking about getting around to picking up another 512MB DIMM so I could run dual channel mem mode but didn't think it would get me much so I haven't gotten around to it yet. Are you running single or dual? I would think it wouldn't make too much of a difference as the P4 seems to get whipped by the Athlons for the most part despite their sizeable bandwith edge.

 

[Basic]

Pascal:
I've only seen the problems when more than one sensor app is running. I don't think it will make any difference between not running and uninstalled.

So it's still a mystery why you get such a low reading.


Ben:
Eeeww, a rice-box.
The only "rice" on my box is the barbed wire fan grill. And that was actually an accidental look when I cut it up for better airflow.

WarCraftIII rocks! I restarted it recently. I never finished it the first time because it started to jerk horribly when I came to some crowded Orc levels. But it works just fine with the new box.

Dual channel (2x512MB) @ DDR266 speed CAS2. (Don't remember the other timings.) FSB @ 266MHz. (It's actually DDR333 CAS2 memory, but I haven't overclocked the CPU FSB yet. Hmmm, maybe it's time to play around. )

About the folding:
A given CPU is supposed to get a constant score/time unit when working. At least if they have a performance profile similar to a Celeron 500. But I get the (subjective) feeling that it's not that way for my (more powerful) CPU. I get the feeling that the hard WUs gives better score per time unit. It could be something like: hard units puts lots of pressure on the Celeron cache, but with larger cache there's no problem. So we don't just get the clock advantage, but also less cache thrashing. Ergo hard units give more points per time on powerful CPUs.
And I've got the feeling that after I've been gaming (delayed return of the WU), I get degraded to easier WUs.
Could it be a factor here? That your daily gaming keeps you off the best WUs, but I get more of them?

Either that or the dual channel. My extra memory space will probably make nil difference. I usually have >700MB left.

[Update]
More specific memory timings were 6-3-3-2 @ DDR266.
But from now on (until it crashes ), I'm running DDR333 7-3-3-2, 333MHz FSB, and 12x multiplier. That's still XP2400+ CPU speed, but with a faster FSB.

 

[pascal]

Looks like MBM5 dont have an exactly simple way to tell which sensor is the CPU diode sensor :?

MBM5 site points to this nforcershq forum thread: http://www.nforcershq.com/forum/viewtopic.php?t=16364

Then I will have to do some experiment and analyze the results?
Maybe I try it some weekend.

 

[BenSkywalker]

Basic-

First off, congrats on the big 1K mark First uni processor folder to hit it(and by a sizeable margin at that ).

Eeeww, a rice-box.

Hehe, when you add up my case, PSU and other 'rice worthy' parts I'm out the cost of a R9800Pro/FX5900 I've always had strictly functional plain beige box cases and I got tired of trying to explain to people that its what's on the inside that counts. Now I could have a 486 33MHZ in my rig and people would be blown away.....

At least if they have a performance profile similar to a Celeron 500. But I get the (subjective) feeling that it's not that way for my (more powerful) CPU. I get the feeling that the hard WUs gives better score per time unit. It could be something like: hard units puts lots of pressure on the Celeron cache, but with larger cache there's no problem. So we don't just get the clock advantage, but also less cache thrashing. Ergo hard units give more points per time on powerful CPUs.

I think you are on to something here. When I get the 50-70 point WUs I start closing on on you a bit(a very little bit, but still ), but when I get to gaming too much and get the ~20 point WUs I start fading back again. Makes sense. Also, the GROMACS core WUs I fly through, about twice as fast point for point as the Tinker core WUs.

Either that or the dual channel. My extra memory space will probably make nil difference. I usually have >700MB left.

I know it's not RAM amount, I tend to have plenty of physical left(unless I'm running Mafia or U2) but thought that dual channel could make a difference, although now that you mention the WUs you get assigned it makes more sense.

But from now on (until it crashes ), I'm running DDR333 7-3-3-2, 333MHz FSB, and 12x multiplier. That's still XP2400+ CPU speed, but with a faster FSB.

That would be an Athlon 2500XP if they had them The 2600 is 12.5 @166, the 2700 is 13 @166. I know it's the same core speed, but with you within ~100MHZ of me and running the same FSB it would take me about a year to catch you if I stopped gaming cold turkey(very cold day in hell scenario ).

WarCraftIII rocks! I restarted it recently. I never finished it the first time because it started to jerk horribly when I came to some crowded Orc levels. But it works just fine with the new box.

I finished it off a few days ago, certainly enjoyable although I thought it could use a bit more direction(although on the flip side of that, with more solid direction it would have had less variety). Bouncing back and forth between the RPGish style missions and the more traditional RTS ones kind of had me having a bit harder time getting in to it then I would have liked to. Minor gripe. Also, I found the very end cinematic to be, I guess you would say anti climatic. Really hoping that Frozen Throne finished off some of the elements. These are minor issues, just the game critic in me coming out

I really enjoyed the elements that they seemed to borrow from Sacrifice(the idea of hero units, gaining xp etc), although I think the troop formations could have been done a bit better(although I haven't really glanced, this could be an issue of RTFM ). Overall I'd say it's a 4.5/5 title. I enjoyed it quite a bit more then CivIII which took itself too seriously in some aspects(diplomacy, civilization progress) while others were disgustingly lacking(had a squad of tanks wiped out by a single guy with an axe once ). Solid and enjoyable game.

Performance wise, I played through @1280x960 4x AA 8x AF and don't recall a hint of slowdown anywhere. 1600x1200 also appeared to play smoothly but my refresh rate at that res is a bit less then optimal for ten hour stretches of gaming(76Hz max).

Pascal-

It is a pain on most of the AMD mobos to get an actual core reading. A lot of people freak out when they are used to the socket temp and start seeing the actual temp so a lot of companies make it a pain in the ass to find the accurate temp. MSI's latest BIOS screws it up on my mobo, which is one of the reasons I'm not running it. I want to know my core temp thank you