[HowTo] Building an overclocked Core2Duo box for OSx86.

[bofors]

Any particular reason why you have chosen a Xeon 3060 over a core 2 duo E6600 (for instance)?

 

Absolutely, the "Xeon" branded version are of higher quality than their "Core2Duo" counterparts. In short, they can be expected to overclock better but are not much more expensive.

[joe75]

I am using PowerFractal (set at the highest "Maximum Count" of ~16.7 million) to test for 100% stability. This loads both cores for about half an hour

 

What kind of temps do you get at full load

[bofors]

Two of my Xeon 3060's give temperatures about 75C (but the third, which does not appear to be quite as good only reads about 65C).

 

It is important to note however, that the meanings of these "tempeatures" is not the same as with other chips. Intel is using a new digital sensor systems here and these chips are much cooler than other putatively at the same the temperature. For example, people at XtremeSystems have commented that before when a chip said it was at some 70C, that the heatsink would also be hot. Here, the heatsinks are not really much above ambient and I know they have good contact to the chip.

[Alessandro17]

Absolutely, the "Xeon" branded version are of higher quality than their "Core2Duo" counterparts. In short, they can be expected to overclock better but are not much more expensive.

 

Thanks

 

You are a great source of hardware info. I have taken what you say very literally and I have been very satisfied with the results.

[bofors]

I want to post some more detailed results and pictures from my new Xeon 3060 / BadAxe2 builds, but I am really busy now and will probably not have time for a while. So results are going to piecemeal.

 

The BadAxe2 LAN controller is different and is not exactly working "out-of-the-box". I have not tried to edit any kexts to fix this but have noticed that the LAN controller becomes functional if one goes to System Preferences->Network (and perhaps select in Ethernet in the "Show:" pull down menu). It thereafter works (even after rebooting) for most things like Safari and iTunes but apparently now everything (I had problems doing running the packet analysis tool "Snort").

 

EDIT: The "Snort" problem I had now appears to be unrelated to the BadAxe2 LAN controller (it would not run right on BadAxe1 either and appears to not be complied for x86).

 

I have not even tried the BadAxe2's audio codec (my main speakers are USB SoundSticks).

 

One of my three Xeon 3060 is clearly not as good as the two others and putting it on a BadAxe2 did seem to improve its overclocking at all. It is not much to be complaining about, but OS X is not 100% stable (by Power Fractal as detailed above) at 3.6 GHz on this chip. This chip seems to be 100% stable at 3.55 GHz though, but the vCore is all the way to up 1.6 V. The odd thing about this chip is that it reports a much lower temperature than the other two (one might expect a higher temperature).

[joe75]

I would call or email Intel about that chip and see what they will do for you

[bofors]

I would call or email Intel about that chip and see what they will do for you

 

I was thinking about it, but the only legitimate complaint I could come up with is perhaps that the temperature is being report wrong (much lower than the others).

 

Nevertheless, I guess it is worth a shot.

[joe75]

I'd tell em you have three and you know there is a problem with that one and mention the faulty temps and make up some {censored} abut errors or reboots. Im not saying to mention you cant clock it like the others but you could probably get an exchange out of it.

 

Once you mention you have three, they will take care of you

[bofors]

BTW, I just noticed that TigerDirect has BadAxe2's for only $200 (... damm, I did not see that coming):

 

http://www.tigerdirect.com/applications/Se...mp;Sku=I69-2145

[Alessandro17]

The BadAxe2 LAN controller is different and is not exactly working "out-of-the-box". I have not tried to edit any kexts to fix this but have noticed that the LAN controller becomes functional if one goes to System Preferences->Network (and perhaps select in Ethernet in the "Show:" pull down menu). It thereafter works (even after rebooting) for most things like Safari and iTunes but apparently now everything (I had problems doing running the packet analysis tool "snort").

 

I have not even tried the BadAxe2's audio codec (my main speakers are USB SoundSticks).

 

So now the question is: which board is fully compatible with OS X and Core 2/Xeons?

 

Maybe the ASUS P5W-DH DELUXE?

 

Edit: I found an entire thread dedicated to the ASUS P5W-DH and it seems to have problems as well.

[bofors]

The BadAxe1 certainly meets my standard for OS X compatible, all it needs is the modified AppleHDA.kext for stereo output (which I have posted in this thread). The BadAxe2 LAN certainly does work for almost everything once one uses the System Preferences -> Network to get it recognized (and my problem with "snort" could have been due to something else).

 

I expect that the BadAxe2 audio codec will not be hard to get running either. It might work with a version of AppleHDA.kext that already exists for all I know.

[bofors]

XBits labs has a recent air cooler review that makes the ThermalTake Big Typhoon again look like the way to go. I am pretty happy with mine, but like the VX model better. It is usually is only a few dollars more, but comes with a much nicer fan. I have actually had problems with regular Big Typhoon fan not starting up regularly when connected to the motherboards rear fan (which is where I used these) header. Too bad the VX's fan controller is integrated though, I wanted to use them with the Sunon's on the CPU. But I bought some Zalman FanMate 2's instead, which work better than the RadioSchack rheostat, to quiet my 40+ dBA Sunons down. Slowing down the Sunons actually does not seem to hinder cooling much at all, probablly due to less turbulence and the 38mm thickness.

 

 

http://xbitlabs.com/articles/coolers/displ...man-scythe.html

[casse]

XBits labs has a recent air cooler review that makes the ThermalTake Big Typhoon again look like the way to go. I am pretty happy with mine, but like the VX model better. It is usually is only a few dollars more, but comes with a much nicer fan. I have actually had problems with regular Big Typhoon fan not starting up regularly when connected to the motherboards rear fan (which is where I used these) header. Too bad the VX's fan controller is integrated though, I wanted to use them with the Sunon's on the CPU. But I bought some Zalman FanMate 2's instead, which work better than the RadioSchack rheostat, to quiet my 40+ dBA Sunons down. Slowing down the Sunons actually does not seem to hinder cooling much at all, probablly due to less turbulence and the 38mm thickness.

I agree the Big typhoon seems like the way to go for d975x motherboards, with its effectiveness and cooling the area surrounding the cpu. Regarding the VX model, however, I've read of people doubting the reliability of the plastic plugs that fasten the Typhoon to the motherboard. This review, which is very in depth and seems reliable, states that:

After mounting the Big Typhoon VX several times on both my AMD and Intel test systems, I’m less than impressed with the new tool-less mounts. The AMD clip doesn’t hold the big cooler in position very well and to be honest, I don’t really trust the plastic push-pins on the Intel mount. Even after the push-pins were fully seated and locked, it was too easy to pop them back out without applying much pressure. Push-pins may be OK for some of the smaller coolers but for a very large cooler like the Big Typhoon VX, I don’t trust them…

and in the conclusion:

While both of the new mounts are potentially easier to install than the original Big Typhoon, the new mounting schemes are not as robust as Thermaltake’s classic H-style steel plate system. In the end, users will have to decide whether or not they value the added performance and ease of installation over a more secure mounting system. For my money, I will stick with Thermaltake’s sturdy mounting system used on the original Big Typhoon.

Further, the only other difference between the VX model and the old one seems to be the improved fan. But it's still just a 25 mm fan, right? I read a really comprehensive comparison of heatsinks, including Big typhoon, with different fans. I thought it was in the Silentpcreview forums but I can't find it now. Anyway, the conclusion was that you really need a fan that moves more air through the densely packed fins on the typhoon, to make real use of it. This is your analysis as well, Bofors, is it not? My point is:

 

1) If you swap the stock Everflow 25 mm fan for a 38 mm fan, the assembly will become even heavier and the potential plastic plug issue will be even more pronounced.

 

2) If you're going to swap the stock fan for another, there's not much point getting the VX model anyway.

 

Both these points favor the original Big typhoon model. On the other hand, if the problem with the plastic plugs is exaggerated (some reviewers seem to think it works fine) I'm thinking the stock, adjustable speed, fan might serve well as an exhaust fan for the case, where there's less need of changing the speed and high CFM is not necessery. Then the VX model might be a good buy anyway, all things considered.

 

Just another bit of armchair analysis (to keep the thread alive). Thanks for reading.

[bofors]

Regarding the VX model, however, I've read of people doubting the reliability of the plastic plugs that fasten the Typhoon to the motherboard.

 

Regarding, the use of "push pin" mounting on the VX version of the Big Typhoon, I can say that I was not thrilled to see them when I open up my first box either. However, after working with the original and two VX's I really do not have problem with them. The original Big Typhoon's mounting system is certainly more rugged, but it much more difficult (and time consuming) to install. Using the "push-pin" systems (which is almost identical to what Intel uses on their stock coolers), one has to make sure that the pins themselves actually cross through the motherboard and that they are properly locked (the real problem with "push-pins" is that people will certainly fail to do this). If properly installed, it is very secure. Furthermore, provided the spring tensions are correct, the "push-pins" should provide the right amount of pressure on the CPU (whereas with the other system one could possibly over-tighten it).

 

EDIT: I moved my machines and noticed that at least two of these "push-pins" came loose in the process. This is a big negative for me and refastening them with the board in the case was a hassle too.

 

On a related noted, my experience installing three ThermalRight HR-05 on these Northbridges has left me with some doubts about that mounting system. First of all, they are not able to be mount completely parallel to the PCI slots, the mounting clip forces the fan to be mounted higher on the heatsink than it should otherwise be, only two of the four hooks on the motherboard are used, and the result is that the heatsinks feel a little loose. Nonetheless, I will probably buy more of them.

 

Further, the only other difference between the VX model and the old one seems to be the improved fan. But it's still just a 25 mm fan, right? I read a really comprehensive comparison of heatsinks, including Big typhoon, with different fans. I thought it was in the Silentpcreview forums but I can't find it now. Anyway, the conclusion was that you really need a fan that moves more air through the densely packed fins on the typhoon, to make real use of it. This is your analysis as well, Bofors, is it not?

Yes, I have read an analysis (whichI believe I posted the link to in this thread before) that shows that the Big Typhoon (which aside from the fan and mounting mechanism is the same as the Big Typhoon VX) performs very well with 38 mm thick fans with high amounts of air flow (above 100 CFM). However, my own limited experimentation with the 38 mm Sunon's that I use seems to indicate that the maximum cooling effect can be reached with far lower flow rates (I have slowed these fans significantly to make them quiet and there seems to be little or not effect on the CPU temperature being reported). I am attributing my results to lower turbulence at lower speeds (which must be an important factor when pushing air through the grill).

 

My point is:

 

1) If you swap the stock Everflow 25 mm fan for a 38 mm fan, the assembly will become even heavier and the potential plastic plug issue will be even more pronounced.

 

I do not think the weight difference between these fans in significant here, and again, if you are careful to make sure the "push-pins" cross the motherboard and are properly locked there is no problem with them.

 

2) If you're going to swap the stock fan for another, there's not much point getting the VX model anyway.

On the other hand, the stock VX fan may not even need to be replaced either. Although I suspect that the thicker 38 mm fans generate higher pressure (which I think may be more important here than actual air flow), my Sunons' only have five blades while the VX fans have nine (which I think should also increase pressure).

 

Ideally, I think I would want a 38 mm fan with nine or eleven blades here (but I have yet to properly the research the aerodynamics of fans pushing air into grills). Delta makes some such fans, but they are much more expensive than the $8 or so I have been paying for these Sunons.

 

I'm thinking the stock, adjustable speed, fan might serve well as an exhaust fan for the case, where there's less need of changing the speed and high CFM is not necessery. Then the VX model might be a good buy anyway, all things considered.

 

That's exactly the way I am looking at it. I have put all of my stock Big Typhoon fans (including VX's) on the exhaust (the stock CoolerMaster Centurion fans are pretty useless and I do not use them at all). The one original Big Typhoon fan (i.e not the VX) I have, regularly has problem starting up off the rear motheboard fan header and often needs a push (which would make it almost useless if my computers were not usually on 24/7). The nine-blade variable-speed VX are great though, they look like something that Silverstone would sell for about $12 (which makes them a bargain since the VX model can be had for only a few dollars more than the basic Big Typhoon).

[mdg]

Bofors,

 

I'm having the opposite obeservation re: OSX vs. Win XP.

 

@ 3.29 Ghz (all I've really done thus far), OSX is 100% stable on my E6600 BadAxe2 Intel board, but I can't boot Windows.

 

Also, I'm trying to reach 400 Mhz (to get 3.6 Ghz, but I seem to be maxing out at 366'ish). What settings are you using to get to 3.6Ghz?

 

Thanks,

 

/mdg

 

PS. How did you get your sound working (Sigmatel 9227 or the 9274D)?

[bofors]

Also, I'm trying to reach 400 Mhz (to get 3.6 Ghz, but I seem to be maxing out at 366'ish). What settings are you using to get to 3.6Ghz?

 

Well, if you have decent cooling and not scared to take a little risk, I would start setting your memory to 1066 strap (BadAxe2 "reference" 266) with timings 5-5-5-15 at 533 and a vDimm of 2.2 V. Then turn up vCore, vMCH and vFSB all the way and see what your chip can do. Then I turn the vCore, vMCH and vFSB (in that order to the point where instability is reached). I also enable HPET for OS X 10.4.8 and disable EIST (aka "Speedstep" which is an energy saving throttle).

 

You might want to mention what you are using for cooing here and type of memory here too.

 

PS. How did you get your sound working (Sigmatel 9227 or the 9274D)?

 

For the 9227 (BadAxe1), I used a hacked version of AppleHDA.kext which I posted in this thread previously. However, it does not seem to work for the 9274D (BadAxe2), so as of now that is an open problem.

[mdg]

Well, if you have decent cooling and not scared to take a little risk, I would start setting your memory to 1066 strap (BadAxe2 "reference" 266) with timings 5-5-5-15 at 533 and a vDimm of 2.2 V. Then turn up vCore, vMCH and vFSB all the way and see what your chip can do. Then I turn the vCore, vMCH and vFSB (in that order to the point where instability is reached). I also enable HPET for OS X 10.4.8 and disable EIST (aka "Speedstep" which is an energy saving throttle).

 

You might want to mention what you are using for cooing here and type of memory here too.

For the 9227 (BadAxe1), I used a hacked version of AppleHDA.kext which I posted in this thread previously. However, it does not seem to work for the 9274D (BadAxe2), so as of now that is an open problem.

 

Bofors, I've been running your fractal program for 24 hour straight now, and have been able to do a cold boot after 60 minutes off. I'm steady at 3.4Ghz. I've gotten to 3.6 Ghz before, but settings got wiped out (long story).

 

 

I will eventually get to 3.6 Ghz and above...I am confident. I just need some time.

 

I am using Super Talent PC6400 DDR2 800 Mhz, which I am running at 2.28 volts at 4-4-3-8 timings at 3.4 Ghz. I ran this timing at 3.6 Ghz as well. These sticks supposedly hold up to 500 Mhz FSB (1000 Mhz) at 4-4-3-8 before needing to drop to 5-5-5-15. Time will tell

 

Regarding sound, I thought this board had both of the above?

 

I have Dev ID 7621 Ven ID 8384 / Sub 8086 0419

 

I thought I would just be able to slightly mod the existing kexts and I was good. But I have zilch

 

Any ideas on sound? Should I wait for someone to come up with a fex or buy an alternative? What are you using?

 

/mdg

[sg]

i have a few questions about OC'ing and what exactly my results mean...

 

basically i've determined that my FSB on my p5w-dh can go up to 410-415 and be stable with my e6400. my ram is some partiot ddr2 800 4/4/4/12 and it will run that those timings with a voltage of2.1 (default). however, if i raise my FSB too high (over 360), it wont even run at 4/4/4/12 even if my memory clock is 800 or lower.

 

if i leave my SPD timings to auto it will boot easily at FSB 420mhz, but if i try to run it at spec (4/4/4/12) or even 5/5/5/15 while still running at 800mhz it wont even POST. i also tried to increase my voltage to 2.2 or even 2.3, no difference.

 

with my fsb at 320mhz my ram will be stable with 4/4/4/12 even at 915mhz.

 

so i'm a bit confused what all this means.

[sg]

However, OS X is not very stable above 3.6 GHz while Windows clearly is. This can be attributed to the microkernel-like archetecure of OS X versus the monolithic structure of Windows. While OS X may be easier to port, it is a little less stable.

 

xnu isnt a micro kernel.

[asap18]

well the mach microkernel is the core of the xnu hybrid.

[mdg]

You need to take you ram timings down to 533 instead of 667 or 800 prior to OC'ing. This effectively makes your CPU to Ram timings 1:1. Now as you increase your FSB to 400 Mhz, you will reach a Ram timing of 800 Mhz (266 for CPU x 1.5 = 400 Mhz / 533 for Ram x 1.5 = 800 Mhz). Based on the kind of Ram you have you might need to set your timings to 5-5-5-15 and then go nackwards to tighten.

 

BTW - my Windows issue had nothing to do with OC. It was because my BIOS was set to AHCI and I didn't have the drivers installed.

[sg]

well the mach microkernel is the core of the xnu hybrid.

 

apple doesnt use the mach microkernel (3.0) which failed as a kernel. read this article:

 

http://www.roughlydrafted.com/0506.mk1.html

[sg]

You need to take you ram timings down to 533 instead of 667 or 800 prior to OC'ing. This effectively makes your CPU to Ram timings 1:1. Now as you increase your FSB to 400 Mhz, you will reach a Ram timing of 800 Mhz (266 for CPU x 1.5 = 400 Mhz / 533 for Ram x 1.5 = 800 Mhz). Based on the kind of Ram you have you might need to set your timings to 5-5-5-15 and then go nackwards to tighten.

 

BTW - my Windows issue had nothing to do with OC. It was because my BIOS was set to AHCI and I didn't have the drivers installed.

 

yes, when i bring up my fsb and i bring my ratio to 1:1 i make sure i clock my ram down to 800fsb, yet it still fails unless i use some really high timings or use automatic SPD values.

[mdg]

yes, when i bring up my fsb and i bring my ratio to 1:1 i make sure i clock my ram down to 800fsb, yet it still fails unless i use some really high timings or use automatic SPD values.

 

Start with Ram timings of 533 5-5-5-15 @ 1.9v, and then move the FSB up (I do 7 at a time). Then up the voltage or MCH/ICH as appropriate.

 

/mdg.

[bofors]

Bofors, I've been running your fractal program for 24 hour straight now, and have been able to do a cold boot after 60 minutes off. I'm steady at 3.4Ghz. I've gotten to 3.6 Ghz before, but settings got wiped out (long story).

 

I will eventually get to 3.6 Ghz and above...I am confident. I just need some time.

 

For the record, I am have a Xeon 3060 on a BadAxe 2 running at 3.6 GHz with the follow settings:

 

vCore (CPU): 1.5375 V

vMCH (NorthBridge and SouthBridge): 1.550 V

vFSB : 1.450 V.

 

Those appear to be the minimum settings to complete PowerFractal at the maximum count (about half an hour). However, my software seems to need a little more vCore for long runs (8 to 12 hours).

 

Regarding sound, I thought this board had both of the above?

No, it comes with either the 9274D or the 9227 (and I do not think this version, the LAD975XBX2LKR, is generally available).

 

http://www.intel.com/products/motherboard/D975XBX2/index.htm

http://www.intel.com/design/motherbd/bx2/bx2_available.htm

 

Any ideas on sound? Should I wait for someone to come up with a fex or buy an alternative? What are you using?

 

Well, this really is not a problem for me because I am using these machines as number crunching workstations and I also have some USB speaker sets which work fine.

 

However, I want to promote the BadAxe2 as a "fully" OS X compatible motherboard and sooner or later I will be needing it to support audio, so I am definitely interested in working on a solution to this problem. I have yet to deal with these types of audio support issues on OS X before, but my impression is that the community knows how to fix them.

 

So, I think we need to try contacting some people and ask for advice. Kiko might be able to help (as he got the BadAxe1's audio working), as well as James2mart in the Sigmatel 9200 forum ( http://forum.insanelymac.com/index.php?showforum=81 ) and I think Munky has some experience here too.