ASUS A8N-SLI Deluxe, Now I got worry

ragnar · Post by **ragnar** » Tue Feb 08, 2005 11:54 am

I'm waiting on a new system, and the pwrfreaks are really starting to give me an itch. Here's the thing:
ASUS A8N-SLI Deluxe
AMD64 3500+ (Winchester)
2x512Mb Corsair "Value Select" DDR
Sapphire ATI Radeon X800XL 256Mb DDR3
Creative Soundblaster Audigy2
Seagate Barracuda SATA 160Gb
Some 2 fans running 5-7volts

and here it comes:
Fourtron Source 350w PSU (18A on the 12volt rail).

I know these kinds of questions are really getting on everybodies nerves, but I'm wondering if anyone's had any experience with a similar spec and a poor 12volt rail. I don't want to burn the memory or CPU by clipping the PSU first thing (I've been waiting for these components for quite a while now, seems the production isn't what it should, or I'm having a lousy retailer).

[EDIT]
I really don't want to jump the gun here, but here are my options for replacement PSU:
Antec NeoPower 480W ATX v2.0 - the cable managment thing sounds like a design I could support.
Nexus NX-4090 - anything that boasts 16.2 dBa at idle should be to my liking. The modest climb from my 350 to just 400 should be made a bit better by the 2 12volt rails @ 14 and 15A.
[/EDIT]

FreakNGoat · Post by **FreakNGoat** » Tue Feb 08, 2005 2:32 pm

AFAIK, ATI cards don't support SLI... I question some of the component decisions--SLI motherboard but not SLI video card, Seagate harddrive instead of Samsung, and obviously the power supply. Seems like you'd be better off getting a non-SLI board like the Chaintech and using the money saved to get a Seasonic supply...

burcakb · Post by **burcakb** » Tue Feb 08, 2005 3:28 pm

Yup, no SLI on ATI yet (and ATI will use a different tech when they finally do it)

As for the rest of the system, let's see what's on the 12V rail (the rest is insignificant, any PSU can do the 3.3V & 5V)

AMD 3500: 7.4 A (probably much less, let's be on the very safe side)
R X800XL : 3.3 A (this is for XT, XL should use less power)
Seagate : 2.0 A (spin up draw and I'm being generous)

Total : 12.7A on 12V rail. you've got 50% overhead. You're fine.

Note: actual draws may be a little different, put them down from memory, but with 50% overhead, you're still very safe.

ragnar · Post by **ragnar** » Tue Feb 08, 2005 11:49 pm

Thanks for the replies!
I know ATI doesn't supporrt SLI, but here in Sweden there just isn't any nForce4 Deluxe cards around (just plain nForce4, and most of them don't support cool'n'Quiet), and the difference in price isn't that big. I don't care much for NVidias SLI solution anyway. Seems the wrong way to go about with all the heat issues. It was fine in the Voodoo2 era though.

As soon as I've got the stuff and put it together I'll post my progress, if there's anyone else out there in doubt.

[EDIT]
About the other desc.
The PSU is an old one I've got lying around.
The Seagate is because I'm curious on testing the NCQ. Maybe I've been hiding under a rock for too long, but the last time I bought HDD, Seagate was the thing to get.
[/EDIT]

meglamaniac · Post by **meglamaniac** » Wed Feb 09, 2005 1:32 am

Don't worry, Seagate are still reckoned to be good.
Opinion here simply suggests that Samsung are slightly quieter. Samsung drives also support AAM which allows you to sacrafice some seek performance in order to lower noise when the drive is in use if you want. Seagate don't support it due to an ongoing legal dispute.

ragnar · Post by **ragnar** » Wed Feb 09, 2005 3:52 am

I'm also planning on installing a dvd-writer (I'm currently sitting on a cd-rw writer and a dvd-reader - not a good solution). That should be another 2A on the 12volt rail (me being generous today), not that I'm that worried about it anymore thanks to your kind help.

From what rail does the floppy feed? I might have to plug one of those in during XP install in case the BIOS hasen't got any built in SATA-IDE cloaking feature.

meglamaniac · Post by **meglamaniac** » Wed Feb 09, 2005 6:41 am

A floppy?
Blimey, don't even worry about it.

I would imagine it draws half an amp if that, and that's being extremely generous. It's only got a tiny little stepping motor. I imagine the real draw could be measure in milliamps.

Josd · Post by **Josd** » Wed Feb 09, 2005 8:10 am

I played around a little with most of the comparable components. See my experiences at http://forums.silentpcreview.com/viewto ... 015#155015

Tobias · Post by **Tobias** » Wed Feb 09, 2005 11:34 am

burcakb wrote:Yup, no SLI on ATI yet (and ATI will use a different tech when they finally do it)

It is true that ATI will use another tech for their version of "SLI", but that doesn't mean that a motherboard made for SLI today will be useless for a setup with two ATI cards. With an "SLI-motherboard" you will get two 8x PCI-E ports to use with whatever two cards you like. Either with two Nvidia SLI-cards or two ATI cards with that solution when it becomes available (or with one of each, like Tomshardware did:). The PCI-E ports on the motherboard is not part of the technical solution of SLI, it is only a prerequisite.

Just thought I should try to clearify that a little and I hope that was the effect and not the opposite:)

meglamaniac · Post by **meglamaniac** » Wed Feb 09, 2005 12:46 pm

There has to be more to it than that, or loads of manufacturers would have just whacked two PCI-e x16 ports on.
I beleive the chipset plays a role somehow as well: look at nVidia's chipset lineup. There's the nForce4, the nForce4 Ultra, and the nForce4 SLI. The latter is the only one that will support SLI cards, although there are reports that Asus and a couple of other brands have managed to hack the Ultra to make it work as well.
Anyway, specifics notwithstanding, it would seem the graphics cards are not the only determining factor.

Tobias · Post by **Tobias** » Wed Feb 09, 2005 2:35 pm

Actually all PCI-E mobos will support SLI. I know I have read "articles" (although I am way to tired to find any right now) where they ran SLI on one 16x PCI-E and one 1x. The question of making an "SLI-board" is how to divide the "channels" (don't know the exact term to use here). There are only a set number och channels and those can be divided to different ports to fit different use. If you think about it isn't hard to see. Either an SLI-mobo has one 16x PCI-E for the graficscard or it works with two 8x ports (in SLI-mode). The number of channels is set.

For Nforce4 the number of channels seem to be 19. 1x16 + 3x1 or 2x8 + 3*1. However if one looks at ASUS SLI-board it seems that they have choosen to let one of the large PCI-E ports double as either a 8x or a 1x port, so the choices on the board are 1*16 + 3*1 or 2*8 + 2*1. That is, one loose one PCI-E channel in SLI-mode. The reason is of course space on the mobo.

meglamaniac · Post by **meglamaniac** » Wed Feb 09, 2005 2:52 pm

Hmm, odd that nVidia were so upset with Asus then.
On the subject of channels, won't some of the onboard devices be connected with PCI-e as well?

ragnar · Post by **ragnar** » Fri Feb 11, 2005 1:53 pm

Sorry for hijacking my own post.
I setup the new system, but it wouldn't boot, but instead "System failed memory test", which I'm guessing has something to do with power. I've tried numerous memories but to no avail. I've unplugged all non-essential components, one by one trying to get it to boot, but it simply won't. Sadly I don't have any higher speced PSU for testing

Do you think it's related to my poor PSU? The memory seems ok. It runs fine on the current rig.

meglamaniac · Post by **meglamaniac** » Fri Feb 11, 2005 2:28 pm

Make sure you've set the memory timings properly (or set it to auto).
If you want to make sure, set everything on the memory options to it's slowest setting and that should work (but obviously not very quickly).

ragnar · Post by **ragnar** » Sat Feb 12, 2005 12:47 am

Sorry, I was running a fever last night, so I was a bit unclear. The trouble is it won't POST, I wrote boot. Sorry. I'll try and have a really methododical approach here. Maybe it'll be informative.

I really need your advice. Should I go ahead and get a more powerfull PSU? It seems I'm on the low end from what I've heard from people contacting ASUS support. ASUS says the "EZ-plug", the 12v connection to feed the gfx has to be plugged even when not i SLI. In addition you have to plug in an ATX2.0 12v CPU plug (the P4 plug), there's not even any doubt about that this is compulsary it's clearly stated in the manual. My PSU isn't even spec.ed @ATX2.0 (I bought it when ATX2.0 was a silly P4 invention). I do however have the 12v CPU-plug, and it was plugged in.

I wondering if the components are at full blast (à propos power consumption) during POST, and that's what causing the problem.

[EDIT]
A problem with my testing is that I'm having just one ATX-case. It's a bit of a hustle swithing mb:s a number of times, and I find it a bit hard to breathe when I'm not able to get on the web

. I'm waiting for another case (I'm planning a ambitious project of turning a cheapo $37 case (exl. PSU)dead silent). So my progress will be slow until it arrives.)
Right now I'm just trying to make out if I should get another PSU or not. The 350watt FS might be good for my HTPC-project,

By the way, the PSU is a FSP350-60PN(PF) 350W.
[/EDIT]

merlyn · Post by **merlyn** » Sat Feb 12, 2005 3:07 am

I find it highly unlikely that the PSU is to blame. The GamePC article measured a similar system to yours peaking at 200W although i admit their PSU probably has dual 12V rails.

There is a fairly simple way to verify this however. You don't need to have the motherboard mounted inside a case in order to test it's functionality, just lay it on a piece of foam or similar. I usually use the bubble wrap it came in. Then just use the bare minimum of components to ensure it can pass POST. CPU, memory and graphics. Without a HDD connected startup power draw will be lower. Personally I would go straight into the BIOS and select "Default settings" or similar, most BIOS's have a safe defaults type setting. The main thing I think you need to check is that you have the memory in the correct sockets, read your manual very carefully. If memory serves the A64's require dual channel.

ragnar · Post by **ragnar** » Sat Feb 12, 2005 5:55 am

Yes, maybe you're right, merlyn. I would however like to emphasize that my PSU only has 18A at the 12V rail. That's the big drawback.

Isn't there any risk of shorting the mb if it isn't grounded and bolted to the case? Otherwise it would simplify a lot for me, naturally.

When I tested, I used all possible configurations for memory. I bought 2x512Mb DDR (DS) from the QVL on the ASUS site. I installed to the letter, first the two sticks, then switched them, one of them in one slot. The other one in the same slot etc, etc. until all combinations had been excluded. I then went on to my old Single sided DDR memories and did the same procedure with them. POST failed every time while testing memory.

A stupid question would be if I could mount my 20pin powerconnector in the wrong way in the 24pin mb. I assumed I would go from the top down, leaving the bottom 4 pins (which I understand are for an extra 12V). Should it be the other way around, or is the plug foolproof?

Tzupy · Post by **Tzupy** » Sat Feb 12, 2005 7:03 am

It's odd that you say your Fortron has 18A on the 12V, I have the same PSU and it's only 16A. And that's the specification AFAIK...
Since I'm getting the same mobo, I'm highly interested with your progress. Have you tried to use a 20-to-24 pin adapter?

burcakb · Post by **burcakb** » Sat Feb 12, 2005 7:16 am

Most fortrons that come installed on cases are the ATV model and have 16A on the 12V. 60PNs sold seperately usually have 18A on the 12V.

ragnar · Post by **ragnar** » Sat Feb 12, 2005 8:40 am

Yes, I bought my Fortron separately. I'm actually leaning towards getting a antec neopower 480watt. It seems a nice thing to have anyway:
http://kylning.com/info.php?products_id ... 6602fdcc13.

@Tzupy:
I'll make sure and post my results as soon as I find the time to do some proper testing. I'll check the attic. I think I have a few different PSUs lying around. And I'll try POSTing with a PCI-VGA card (a SIS 4Mb!).

On a more human side, the trouble is my wife has this habit of feeling jealous of my computers when they take up too much of my time. I don't believe spending a saturday night testing will be very popular, concidering I spent most part of last night too in front of her "rivals".

Tzupy · Post by **Tzupy** » Sat Feb 12, 2005 10:35 am

So I got a lesser FSP 350...

It still powers my Barton 1.83 and GF 5900 @450 fine.
After looking on the Anandtech forums for people with A8N-SLI it seems that for your system the NeoPower 480 would be OK. But some people with 2 GF 6800 GTs, a couple of Raptors and overclocked Winnies got into trouble. The solution was Antec True 550 EPS12V (36A on the 12V). I don't know how noisy that would be...
An overclocking suggestion: get the 3200 Winnie and Corsair 4400PT25 (the overall price should be the same) and overclock it to 2.5 GHz. You will definitely want to change the chipset cooler - I suggest to get a Thermalright NB-1 and use Asus' Q-fan to lower the noise. Or, if you are not going to overclock, the Zalman NB47J should be fine with some airflow.

ragnar · Post by **ragnar** » Sat Feb 12, 2005 12:53 pm

@Tzupy
I'm not really planning on oc'ing anything. Anyways, I'm not even planning on running two gfx-cards. I just wanted a nForce4-board, and the SLI chip was the only one available here in Sweden. I've already replaced the chipsetfan/HS with the Zalman NB47J.

I've ordered the NeoPower 480. That should give me enough breathing room for my purposes. I'll get back with results as soon as I can.

[EDIT]
@merlyn.
I was a bit fast in the reply to your post. What makes you not putting the blame on the PSU other than the reviews? I'm really in the dark when it comes to these kinds of things. For me, whenever one of my builds refuse to work proper-like, I don't know in what direction to look. Firstly I try to check if I installed the components with the woollen socks on (being psyched by the manufacturers saying the static electricity may fry your components, I've even bought a wristwrap ground which I plug to the wall-radiator, or something else grounding, when installing things). But from then on, I'm more or less on thin ice. Earlier, I never suspected the PSU, I mean for my AT-builds, I remember having something like a 180watt PSU supporting both the computer and the 14" monitor. And it was always sufficient.
[/EDIT]

merlyn · Post by **merlyn** » Sun Feb 13, 2005 10:28 am

the manual states a minimum requirement of 17A for a system with much higher power draw that yours. in the "Light" column it references 2xPCI-x SLi graphics cards, 2x HDD's and 3 USB devices.

the ground is provided by the ATX connector. usually motherboards are isolated from a case for EMI shielding. if u look carefully around the mount points on the mainboard you will see there is no electrical connection.

your post however seems to answer your own question and also highlight my lack of knowledge on bleeding edge systems! Your motherboard has a 24-pin ATX connector therefore you require a 20 to 24 pin adaptor cable. thanks to Tzupy for pointing that out. i find it extremely worrying that u were able to insert a 20-pin plug into a 24-pin socket, how many more people will do this?

as an aside, i used anti-static products during the first 6 months of my apprenticeship. with a little experience you become aware of static and reasonable precautions. in most cases grounding yourself off against the case is more than adequate.

ragnar · Post by **ragnar** » Sun Feb 13, 2005 10:57 am

Thanks for your answer Merlin!

I've been looking around a bit, and people contacting Asus support has been told that it actually works using a 20pin power connection, but then they recommend a high power PSU, something like a 550watt! The extra 4 pins are a 3V, a 5V and a 12V and a ground. I guess they're afraid making it compulsary with the 24pin, i.e. ATX2.0 PSU, they would scare customers off (my kind of people who like slim lowpowered PSUs). Anyway, I thought it be best and get the information first hand, and have contacted Asus support to hear their opinion.

Tzupy · Post by **Tzupy** » Sun Feb 13, 2005 11:43 am

Ragnar, were you able to POST with the PCI card and the Fortron PSU (just to check if the mobo is working properly)? If it worked then my Fortron will have to go too

I downloaded the manual of the NeoPower 480, after reading some unfavourable comments on it (in conjunction with Asus A8N-SLI), will report back if there's something suspicious.
My concern is if NeoPower 480 will be able to handle 2 6600GTs and an overclocked Winchester safely.

Tzupy · Post by **Tzupy** » Sun Feb 13, 2005 2:19 pm

Well, until I'll know that the 4MB SIS post-ed, my bet goes on a defective CPU or mobo (secretly hoping to keep my Fortron...).
After reading the dox on Neopower 480, it seems that the 12V2 (15A) goes to the CPU and the 12V1 (18A) for the rest. For people with two 6800GTs, two Raptors and other stuff, this 18A (official rating) may not be enough. I am unsure about which 12V should be used for the EZ-plug, I believe the 12V1 (although it looks like there's more spare juice on the 12V2 - with a Winnie anyway). Having the 12V1 close to the graphics cards makes sense since it doesn't cross the mobo anymore.

ragnar · Post by **ragnar** » Mon Feb 14, 2005 12:52 am

From what I've read, the Asus support seems very supportive of the NeoPower 480. Seems to have something to do with the 24-pin power supply and two 12V rails, with the extra 12V feeding from the second line and adding stability. Supposedly, the two 12V rails and the 24-pin power connection was designed for multi CPU servers with rack-HDD clusters.
I haven't gotten to testing with the SIS-gfx yet. I'll have to tear it out of my old TYAN P166 Voodoo2-SLI first. And that one's in the attic.

I'll see if I'll get around to it before my NeoPower arrives, which it should do tomorrow.

@Tzupy
I think I know what You men, but I could interpret it as You're hoping I'll have faulty CPU/MB, both of which I've voided the warranty on by 1. Installing Zalman 7700AlCu, 2. Installing MB47J on the NB. But then again from your perspective, rather me than You, right

Anyways. I hope I can rule out a faulty CPU, since the POST got through the CPU test. And with all the negative feedback I've been hearing about this MB, I wouldn't bet anything against a faulty board. Keep your fingers crossed, some people aren't having any trouble at all. The problem is most people having PSUs at the lower end seems to get inte trouble as soon as they install anything above the bare minimum, e.g. installing a PCI card)

Tzupy · Post by **Tzupy** » Mon Feb 14, 2005 2:55 am

Ragnar, I didn't mean it that way... just hoping to keep my Fortron until I'll need a second 6600GT, and upgrade my PSU and case at that time.
Right now I don't have another option than Neopower 480, a couple of months later I might be able to get a new Seasonic.
I don't think you voided your warranty by installing the Zalman 7700 AlCu and replacing the chipset cooler (at least if you put the old chipset cooler back and the shop people wouldn't notice).

Tzupy · Post by **Tzupy** » Mon Feb 14, 2005 3:42 am

I believe there is an explanation to the memory test failure:
1) The motherboard undervolts the memory (how much, I don't know).
2) When using a 20-pin PSU with the 24-pin mobo the voltages are also degrading slightly.
3) Since the mobo wants to POST at 2.6V and SPD timings, it fails the memory test.
By using a memory that can run at 2.5V ot less, or a DDR500 that at POST runs at DDR400 with maximum relaxed timings, it may be possible to reach BIOS settings and to sort out the undervolting problem.

ragnar · Post by **ragnar** » Mon Feb 14, 2005 3:43 am

I didn't really take it that way. What You wrote just sounded kind of funny to me. Anyway. I just got a tracking number on my NeoPower. I'll post my findings. I'm a bit worried that your in a worse place than I am with your 16A Fortron PSU. The manual says "light load" (which is 2*6600GT, but nevertheless) >=350W, 17A.

Here in Sweden we're debating a vote from 1980 where the people voted to cancel nuclear energyproduction. Maybe I have to rethink my position on that one, now that I'll be needing a reactor of my own to power my new system.