Jump to content
Sign in to follow this  
EjectButton

speed and temp results from switching to passive northbridge and pci-e

Recommended Posts

I'm sure this issue has been beaten to death by many people but I just wanted to relate my own experience since I still occasionally see people asking about it in threads, wondering if there is a performance hit or high-temperature instability. For the many who I'm sure consider this a solved-problem, you can ignore this, I just wanted to add a little clarification for those who still occasionally ask about it. And post this to contribute to the forum which has helped me out a couple times.

 

The issue I am talking about is the one that has been brought up many times, the northbridge fan on the nf4 ultra-d is very noisy but also unfortunately placed right under the end of the pci-express slot1, which is the only one that can run at 16x.

 

When I got this motherboard the first thing I wanted to do was replace the noisy fan with a passive cooler (larger heatsink and no fan), but I didn't want to have to build a custom one or bend up an existing one to get it to fit. So I figured I would just move the video card to the other full-size pci-express slot, then I learned that you have a choice between running (slot1 16x - slot4 2x) or switching the jumpers to sli-mode and getting (slot1 8x - slot4 8x).

 

My concern was that at 8x I would take a performance hit by limiting the bandwidth available to the video card, as others have said, this is not an issue.

 

listed below are benchmark scores and northbridge temperature readings comparing-

(video card in slot1 at 16x - stock northbridge cooler) vs (video card in slot4 at 8x - passive northbridge cooler)

 

the northbridge cooler I used is the zalman zm-nb47j heatsink, I would suspect other heatsinks of similar size would also be suitable, as for air circulation, the fan in my power supply is almost always off, even at load, as is the cpu fan (CNPS7000B-ALCU). So the only real air movement comes from a cheap 120mm fan in the front blowing in and an identical fan on the back blowing out, and the vga cooler which just blows hot air into the center of the case, I would suspect the efficiency of a passive northbridge cooler in this scenario is largely dependent on case air flow, I never put a whole lot of work into mine and it seems more than sufficient.

 

 

------------------------------

northbridge temp

 

(stock fan)

idle 48

load 60

 

(passive)

idle 48

load 54

 

------------------------------

quake4 - necro666

(slot1) 59.1 fps

(slot4) 59.2 fps

 

 

quake3 - four

(slot1) 276.5 fps

(slot4) 279.4 fps

 

 

FarCry - Pier, demo: 1.tmd

(slot1)

Score = 60.41 FPS (Run 1)

Score = 60.07 FPS (Run 2)

Average score = 60.24 FPS

 

(slot4)

Score = 62.30 FPS (Run 1)

Score = 62.07 FPS (Run 2)

Average score = 62.18 FPS

 

------------------------------

 

for some reason my scores actually went up a tiny bit on the (8x slot4) as opposed to the (16x slot1), why this happens I do not know but I read a couple other people having similar experiences on other forums. As for the temperatures, the at-load temp dropped with the passive cooler.

 

so for those being annoyed by the noise of the northbridge (and not running two video cards) you can move to slot4 and swap in a northbridge heatsink and as long as your case air flow isn't terrible you should be fine.

Share this post


Link to post
Share on other sites

What thermal compound did you use on the Zalman - AS5? I've thought about doing this and have one of those Zalman's not being used right now. Only down side is I have to completely take my brand new system apart lol. This is the one thind I wish I'd thought about before I put it all together!

Share this post


Link to post
Share on other sites

I've done the same mod on an Ultra-D

 

Temps didnt change mutch, but at the same time i did the mod, 1 other fan was also removed from the system.

 

Good case ventilation is always a must, and this is no exception.

 

Used AS5, temps dropped by 1-2 degrees.

 

I still need to add another quiet 80mm intake fan, Antec TX1050B case.

I expect another drop of 1-3 degrees when the new fan is added.

 

::edit::

I kinda wedged a 120mm fan in the bottom of the case.

Very poor fit, didnt expect mutch of it.

However, check this out...

How the hell do i get images to display in the post?

Share this post


Link to post
Share on other sites
What thermal compound did you use on the Zalman - AS5? I've thought about doing this and have one of those Zalman's not being used right now. Only down side is I have to completely take my brand new system apart lol. This is the one thind I wish I'd thought about before I put it all together!

 

I used the stock thermal grease that came with the heatsink itself (Zalman CSL 850). The contact area on the northbridge that needs to be coated is very small (maby 1/4 or 1/6 the surface area of an A64) so you don't need much regardless of which one you use. I mis-placed my other thermal paste so I just used the included stuff and the results have been acceptable.

Share this post


Link to post
Share on other sites

Im going to test this myself tonight because I have to see it for myself to believe it! I just would be surprised if there is no difference let alone and improvement.

Is it the changing of the jumpers alone even on a non SLi board that determines the lane bandwidth allocation? i.e. I have an Ultra-d non sli...do I have to do the SLi mod in order to change to x8 slots? Im thinking no, because x8 slots still wont allow SLi to run until you do the hard mod.

This right?

Share this post


Link to post
Share on other sites

Thats right.

Just doing the jumpers alone enables 8x each.

You need an SLI chipset, or need to do the SLI mod to get SLI.

I like what DFI call it... "Dual eXpress Graphics" lol

What they mean is "we know, and the users know the ultra can be modded to SLI, so we'll fit 2 x16 slots to this board".

 

Good case circulation is a must if fitting a pasive cooler.

With 1 x 80mm intake fan in the hard drive cage, i ended up with ~47-50 degrees

I added a 2nd 80mm intake fan the other day, now chipset is 35 degrees.

I was getting 50-55 on the stock cooler.

 

BTW, i'm running my 6600 in the 1st x16 slot again, i bent a few fins on the zalman to make it fit.

Looks like SLI might be an option after all ;)

Share this post


Link to post
Share on other sites
Thats right.

Just doing the jumpers alone enables 8x each.

You need an SLI chipset, or need to do the SLI mod to get SLI.

I like what DFI call it... "Dual eXpress Graphics" lol

What they mean is "we know, and the users know the ultra can be modded to SLI, so we'll fit 2 x16 slots to this board".

 

Good case circulation is a must if fitting a pasive cooler.

With 1 x 80mm intake fan in the hard drive cage, i ended up with ~47-50 degrees

I added a 2nd 80mm intake fan the other day, now chipset is 35 degrees.

I was getting 50-55 on the stock cooler.

 

BTW, i'm running my 6600 in the 1st x16 slot again, i bent a few fins on the zalman to make it fit.

Looks like SLI might be an option after all ;)

 

Yeah Im going for ultimate silent setup. Trying to get upmost silent PC whilst still mainting good overclock. Im looking to use as little fans as possible really, so I will see if I am able to run the zalman with no fans blowing over it. I doubt it. Will see. Will post back my results of slot 4 hopefully tonight. Or infact early evening for you guys in US.

Share this post


Link to post
Share on other sites

Ok well here is my experience. At the moment im running spec in my sig but on a conservative overclock of 10 x 285 HTT (2.85ghz) with ram @ 1:1 (cpu can do 11x multi and can probably go higher).

 

EDIT:

 

See my post below for results.

Share this post


Link to post
Share on other sites

OK update. Started from scratch and this time everytime I changed jumper settings I blitzed all drivers and reinstalled them using the latest Omega 6.1 Catalyst drivers. Driver cleaner pro was used to get rid of any old driver traces and everything was defnitely kept constant with all fans @ 100%. Also got rid of TRIXX and ATI Tool and had just the Omega 6.1 Catalyst package on there.

 

All benchmarks were ran three times and an average was taken (mean value).

 

PCI-e Normal x16slot

 

3dmark 2001: 27684

3dmark 2003: 11838

3dmark 2005: 5423

 

PCI-e Slot 4 (lower slot) x8

 

3dmark 2001: 27656 = (- 0.101%)

3dmark 2003: 11791 = (- 0.397%)

3dmark 2005: 5417 = (- 0.111%)

 

PCI-e Slot 4 (lower slot) x2

 

3dmark 2001: 25350 = (- 8.431)

3dmark 2003: 10739 = (- 9.283)

3dmark 2005: 5066 = (- 6.583)

 

 

So it would appear that my original benches were going wrong somewhere. I think maybe the slot was still running at 2x. Anyway...these benches clearly show that people are right when they say that the difference is minimal. I mean I wouldnt have said 5-10% would have been noticeable in any game, let alone 0.3 % !!!

 

Bare in mind, this was with the GFX card clocked at stock settings. Had the gfx card been overclocked, I dont know if the difference would have been greater slightly. But theoretically, it should be pretty much the same difference as both slots would still be running the same clocked card. I guess I could test this to be sure, but frankly, Im happy already with using my lower slot, until cards really need x16. Saved me chopping up my passive zalman anyway.

 

 

If anyone else could test to be sure that would be great.

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Sign in to follow this  

×
×
  • Create New...