Printable View
You have the same set up as me - the only difference is the washer mod. Use plastic washer as pictured by 3oh6 a few pages back or cut up an unused credit or shop store card - place these under the springs on the back of the mother board for additional increased clamping force of the H/S assembly to the chips.Quote:
Originally Posted by ACE76
My N/B is around 42 idle but hits 68 under load hence my quest for H2O !!!.
I'm a fan of Swiftech products, but the MCW-NBMAX falls short by failing to address the core problem with Asus' stock HPA: the surface of the waterblock is too far removed from the surface of the chipset it's intended to cool. The NBMAX copper base rests against a copper plate that, in turn, makes contact with two copper pipes that only make indirect contact with the chip surface through a copper plate beneath the pipes. The only advantage the NBMAX might provide would be any increased volume of water it may (or may not) hold (it appears larger than the stock NB wblock) and a more finely machined base plate, providing better contact. IMHO, those differences don't offset the horribly inefficient HPA design. You can replace TIM and add as many washers as you like: it's still 'lipstick on a pig.' ;)Quote:
Originally Posted by rob2k
To stay within budget, how about removing the HPA and cooling FETS, SB and VR with conventional heatsinks by Swifty or EnzoTech, then using an EK-NB S-Max for the NB as one alternative?
Typically safe, functional vcore range for these processors is (approximately) 1.28 - 1.60v, depending on how aggressive the overclock. Some will perform short runs/benchmarks with voltage as high as 1.70. I can't see the system booting, much less completing a benchmark, at a vcore of 1.10. Is this a typo?Quote:
Originally Posted by svrz
As you know, quads simply run hotter than dual cores. The seem to run especially hot on this board. The consensus seems to be that this is attributable in large part to the 16-phase power supply design, a design far better suited to dual core CPU's. Temps in the 60's when running Prime95 aren't unusual, nor are temps in the high 30's - mid-40C at idle.
I'm going to put all my EK waterblocks (NB, SB and Mosfets) on this board once I get a few more parts in stock.
have two questions :
1) whats the best 2x2gb Ram for this board ?
2) whats the max safe 24/7 voltages for :
-CPU PLL
-FSB Termination
-North Bridge
running a Q9650.
Thanks Mezzro - I might just go for it. Compiling a list of parts right now :hehe:. Looks like my case is limited to the dual radiator so a direct contact on the N/B may compensate for that.
Just to echo your comment on quad temps, on air with one of the best heatsinks available I get 63 on prime and 75 OCCT :eek:. In fact I only let OCCT run for 5 minutes in fear of the CPU melting through my case. NB is also on 68 at 450FSB hence my need for water ;)
What is your cooling setup and planned O/C ? - as you would have read temps are an issue.Quote:
Originally Posted by quattro_
for me temp are not an issue NB Idls @38c & 43c load SB is alil higher around 45c with RE HS :) ,and im getting the MCW-NBMAX next week so my temps will be even better .Quote:
Originally Posted by rob2k
ill be happy if i can reach 450x9 stable ! still searching for good ram ,the one i have is not good with RE .
Quote:
Originally Posted by Mezzro
Unfortunately i have to say you're right. :(
Tonight i replaced my stock fusion block with the Swiftech one and also place some nylon washers onto the screws that hold the HPA.
As is it seems i gained 1ºC CPU temp and only 2-3ºC on the NB temp.
The bright side is that the delta between idle and load (load= HCI memtest) seem to have lowered some 5 ºC.
Even with the washers in place i couldn't get decent contact between the HPA and the NB chip.
After i removed the HPA, there was still a triangle of untouched copper visible on the copper plate that touches the NB.
A little warning to people who will install this block: i'm using 8mm-11mm compression fittings with the o-ring that comes with those fittings but it leaked like mad.
I had to use the o-rings that come with the swiftech barbs as well to stop the leaks.
Now it's all nice and tight.
Also, the Swiftech logo on mine was upside down. :shrug:
echoing exactly what i have been saying in the NBMAX thread. the HPA is just a dog and despite all the effort, still comes out sucking compared to a direct contact cooler.Quote:
Originally Posted by Mezzro
having just slapped an MCW30 directly on the X48 IHS and having tested NB temps thoroughly out of the box and post nylon washer mod...the HPA is the biggest draw back by far and aside from the benefit of not having adapters out the ace for the Fusion block, i wouldn't expect a degree of difference between it and the NBMAX in a capable loop. the 1/2" barbs though are worth the price of admission for the NBMAX if running a 1/2" loop.
my suggestion to anyone wanting to really lower NB temps and running a water loop, go direct contact with an MCW30 or the EK blocks mentioned, slap some MC14s or Enzotech heat sinks on the CPU/NB PWM and enjoy sub 40C temps regardless of voltage :up:
Anyone know where i can get one of the Crossbow backplates from in the UK?
TIA
And i take it you have seen EKs nb block (like swiftech's)...................
http://ekwaterblocks.com/shop/produc...013f54428b8c2c
from :Quote:
http://images.hardwarecanucks.com/im.../re/heat-9.png
The term "stock" in the chart refers to before we removed the heat sink/heat pipe assembly and re-mounted with the nylon washers. Like we said, it is an absolute no brainer as to what lowered temps almost a whole 10C on average under load with our overall overclocked settings of 477FSB and an increased vMCH voltage to 1.50v under load, just to add more heat to the testing. Idle temps lowered quite a bit and the drop off of the temperature from load was quite a bit steeper and far less gradual. So in the end, the stock heat pipe/heat sink assembly really isn't bad at all, heck, we would even go as far as saying it is pretty good. The only issue is that it appears that the assembly needs a little bit of help and added mounting pressure. Keep in mind, too much pressure and you will get into a warping issue with the PCB which we did encounter but we are willing to forgo that for much better MCH temperatures.
http://www.hardwarecanucks.com/forum...review-16.html
hehe, that is my review, and what i was referring to when i was talking about going from out of the box to nylon washer mod to direct contact in my previous post.Quote:
Originally Posted by quattro_
if i was to add the direct contact with MCW30 to that chart in that test, it wouldn't go over 33C (likely not over 30C) and would be more dependent on ambient temps than anything else. i just wish i wasn't so blind when doing that testing to see that the MCW30 can mount, you just can't use the plastic stand offs as the screws have to be loose to reach in to the MCW30s mounting tabs.
and had i, the tone of the stock assembly would be toned down a bit because it would have been obvious just how much it held back the cooling. at that point, it really wasn't known that the X48 MCH could run that cool with a direct contact waterblock on this board.
oops never knew that sorry .
can someone help me with :
Quote:
Originally Posted by quattro_
I'm having strange lockups in boot process . The computer hangs at "Checking NVRAM" , just before it begins to boot from HDD . The problem is only happening when i have raid mode enabled in ICH9R and have a plextor 760sa plugged in one sata port (5 or 6) .
Removing the sata burner , everything works properly .
Anyone else with this problem ???? Raid mode and optical (ROM or burner) in sata ports ???
3oh6, what about direct contact with an air cooler like this http://www.sidewindercomputers.com/encnfoconoco.html , plus say the Enzotech mosfet sinks plus sinks on the NB mosfets like you did? I don't have watercooling and wonder if it would be worth the trouble. I already have the NB and SB sinks plus a bunch of BGA and such. I would only have to get the 2 Mosfet sinks to do this.
oh no worries, no need to apologize at all. i tend to try not to promote the reviews i do as not to come off as trying to promote traffic but this is one of those situations where it really is warranted and was just thinking of posting that chart when you did, that's why i thought it amusing.Quote:
Originally Posted by quattro_
this was the next question i was asking myself after finding out how well water does on the NB with direct contact. i have an NC-U6 as well as an HR-05-SLI and will maybe try to do a quick little test sometime tomorrow similar to what i did with the review but with those two coolers, and the MCW30 direct contact. i am thinking 500FSB tRD 7 (perhaps 8, not sure what is stable) 1.50v? anyone have any better suggestion? maybe i will do a couple different voltages, 1.50v and 1.40v or something like that.Quote:
Originally Posted by Pntgrd
i am not sure about the Enzotech heat sink you linked, will likely need a fan on it is my first assumption.
I lost two motherboards due to these flimsy barbs: one Rampage Extreme and one Maximus Extreme. After the lose of the ME, I thought I knew what to check for in avoiding a repeat. The problem is that the leaks can be insidiously small -- an almost imperceptible trickle of coolant. Cases with inverted mobo mounting are particularly problematic, as the NB sits inline with the CPU socket beneath it. This last leak took out a QX9650 CPU along with the mainboard. I'm super careful setting up loops, leak testing before running any power to the board, etc. When the worm clamps aren't tight enough -- leaks. Then they are too tight, barbs are easily crushed -- even worse leaks.Quote:
Originally Posted by Zeus
While I think many of us here are in agreement that Swiftech's solution misses the mark, it's still a far better solution than the native NB block for this very reason. The Swiftech's 1/2" barbs actually hold up to a little torque from wormclamps: you can get a decent seal, if not decent temps. :shakes:Quote:
Originally Posted by 3oh6
Yes, I went through this nightmare too. Keep your RAID setup as simple as possible when installing the OS. Make sure the Silicon Image controller is disabled if running Intel RAID (at least during setup). Preferably, chuck software RAID and get yourself an Adaptec or Areca RAID card. :D Problem solved!Quote:
Originally Posted by Stelios
Note: Wtf would they sell a $400 motherboard with only two PCIe 16x slots? Someone, please? LOL. For those of us that are hardware RAID junkies but need both PCIe 16x slots for Crossfire, this is a real problem. On the Maximus Extreme, I could run Crossfire and still have a spare 16x slot to run my 8x Areca card from. Makes life so much simpler. :) While not a complete solution, a partial fix is to use Areca's 1200 RAID card -- a RAID card that runs from PCIe 1x slot. Not ideal, but a nice alternative from ICH9R for two OS boot disks in RAID 0.
It's a great review. I wasn't aware of it, but am glad someone linked it. :DQuote:
Originally Posted by 3oh6
RE EnzoTech: Agreed that active cooling is a must here. I linked to one of their heatsinks in my earlier post that has a low profile fan attached.
Thank you Mezzro for the confirmation . We need another bios as it seems . There are some bugs also with the Tweakit function .
yes that is the best review out there i had it saved in my bookmarks for some time now :)Quote:
Originally Posted by Mezzro
i would like to WC the NB but the only thing holding me back is the mosft HS .
Hi I wonder if anyone knows weather the corsair memory fan will fit over the G.Skill Pi ram?
like a glove...Quote:
Originally Posted by rddave
http://img.photobucket.com/albums/v2...m/_MG_1219.jpg
Thanks very much for that 3oh6, you never can tell from tje product photos on the web :)Quote:
Originally Posted by 3oh6
looks like im going to be getting the PI :D
Thank you for the reply, Mezzro. Believe it or not, no this is not a typo. :) I entered the lowest operating voltage value, as written in the manual, in the BIOS. Miracle or not, it booted just fine. :)Quote:
Originally Posted by Mezzro
Noted. Thank you for clarifying this matter.Quote:
As you know, quads simply run hotter than dual cores. The seem to run especially hot on this board. The consensus seems to be that this is attributable in large part to the 16-phase power supply design, a design far better suited to dual core CPU's. Temps in the 60's when running Prime95 aren't unusual, nor are temps in the high 30's - mid-40C at idle.