the Abit IN9 32X-MAX thread

ALSO... I forgot to ask.. Are you running a divider? That is also important. If your RAM was not designed for heavy overclocking, you may want to run it as close to 800mhz if possible. If that means setting a divider, no biggie.. My ram runs MUCH faster with a 400FSB on a 3:2 divider @ 1066mhz than my default 266FSB on a 1:1 ratio @ 1066mhz. Run Everest Ultimate.. TONS of goodies for benchmarking your CPU and RAM.
 
Thanks for the replies officermartinez :). To be honest, the divider business has me somewhat confused. Let me explain.

btw I have tried originally running the RAM at the default voltage and this seemed OK when the settings were linked, even at the reported 1125 DDR2 rate.

When I ran the settings as above, all seemed OK, as long as I stayed at 375FSB. I tried setting a 5:4 divider, which gave me 1000 DDR2 rate memory (iirc, I will double-check that), but that was unstable.

If I unlink the memory, and set the FSB to 375 (1500 in bios) and set the memory to 800 manually, it is fine. If I then increase the FSB (say to 380), the DDR2 rate will drop from 800 to 775. Is this normal for an Intel board?

TBH I'm not convinced the RAM is the problem. What's puzzling me is the uGuru in Windows. Setting that to 380 or above instantly gives the overclocking failed message, before it's even tested it properly.

I will try as you suggested however and see what happens. Update soon :). Thanks again.
 
Thanks for the replies officermartinez :). To be honest, the divider business has me somewhat confused. Let me explain.

btw I have tried originally running the RAM at the default voltage and this seemed OK when the settings were linked, even at the reported 1125 DDR2 rate.

When I ran the settings as above, all seemed OK, as long as I stayed at 375FSB. I tried setting a 5:4 divider, which gave me 1000 DDR2 rate memory (iirc, I will double-check that), but that was unstable.

If I unlink the memory, and set the FSB to 375 (1500 in bios) and set the memory to 800 manually, it is fine. If I then increase the FSB (say to 380), the DDR2 rate will drop from 800 to 775. Is this normal for an Intel board?

TBH I'm not convinced the RAM is the problem. What's puzzling me is the uGuru in Windows. Setting that to 380 or above instantly gives the overclocking failed message, before it's even tested it properly.

I will try as you suggested however and see what happens. Update soon :). Thanks again.

I know the LINKED vs UNLINKED can be a little confusing. Personally, I run LINKED. Having owned the IN9 32x-max board, there are SEVERAL ratio settings.. MORE than my current XFX 680i LT as a matter of fact. I would try to get my ram as close to 800 as possible.. Soooo.. 775 might be one avenue.. If you decide to try setting a different divider, and your ram is stable at 825, that might be another avenue to consider..

The thing to remember is if you are searching for a stable overclock with your specific ram, try keeping it near the 800 mark. Dividers are the key. When I had my IN9 32x-max board, I tried both linked and unlinked. I always had my best results with it being set to LINKED.
 
Now running 375FSB, linked with a divider of 3:2 (not 5:4 as I mentioned above), giving me 1000 DDR2 rate (1.9v). Manually set timings to 5-5-5-12-2T and verified with CPUZ. Running Orthos now on blend.

Same error with uGuru however, and I suspect I'd get the same problem if I tried it from BIOS.
 
I know the LINKED vs UNLINKED can be a little confusing. Personally, I run LINKED. Having owned the IN9 32x-max board, there are SEVERAL ratio settings.. MORE than my current XFX 680i LT as a matter of fact. I would try to get my ram as close to 800 as possible.. Soooo.. 775 might be one avenue.. If you decide to try setting a different divider, and your ram is stable at 825, that might be another avenue to consider..

The thing to remember is if you are searching for a stable overclock with your specific ram, try keeping it near the 800 mark. Dividers are the key. When I had my IN9 32x-max board, I tried both linked and unlinked. I always had my best results with it being set to LINKED.

Yes I've found the best settings are linked so far, but, I only see a choice of 3 dividers, well 4 if you count sync (what's that all about?), I have 1:1, 3:2 and 5:4. 3:2 gives me the closest @ 375 (1000), but obviously that will go up as I raise the FSB. Can you specify a divider manually when it's linked?

*edit* CPUZ shows a divider of 3:4 FSB/DRAM
 
OK new results

Have tried memory @ 1.9v @1000 DDR2 rate, 3Ghz CPU (375FSB - 1500). Linked 3:2 ratio. Orthos was quite happy blending.

Tried same settings, but raised FSB to 1525. PC locked 1.5 mins after starting Orthos. CMOS clear required.

Tried same settings, FSB 1525, dropped multiplier to 7x and set RAM to Sync (auto sets to 775). PC locked 1.5 mins after starting Orthos. CMOS clear required.

Seems to be a pattern here. FSB limit? Is there another voltage I should increase?
 
What is your CPU speed? ..I can tell that probably the weakest link in your setup is going to be your ram.. However, that's not such a bad thing. You know you are stable at a 375 FSB..
 
What is your CPU speed? ..I can tell that probably the weakest link in your setup is going to be your ram.. However, that's not such a bad thing. You know you are stable at a 375 FSB..

CPU is 3Ghz (from 2.13). Can you explain why you think the RAM is the problem from the results I posted above? the last one shows memory @ 775 yet it still locks. Doesn't this point to more a FSB issue?
 
A few things about your setup.. It looks pretty solid. I would even say the Antec 500w PSU is fine (although, there are a few on here that hate Antec PSU's and would disagree with me).

For dual cores, you have a good 680i board. Options galore.

At 3.0ghz, that might be the spot where you need more VCORE. When I had my E6600 on my IN9 32X-MAX, at 3.1ghz, it need more VCORE. If your system boots fine and only reboots or locks after running prime or orthos, then at this point, it might not hurt to add a little VCORE.. My general rule of thumb is, if you fail prime SEVERAL minutes or an hour or so into it, bump it up only 2 notches from where you are at.. If it fails within just a minute or so, bump it up 4 notches.

Remember, the C2D's can take quite a few notches of voltage.. However, the HEAT is the by-product of VCORE increases and that can cause instability. When I refer to your RAM being the "weak link", I am refering to the specifics of the ram.. Often, the phrase is, "you get what you pay for". I find this to be a general guideline for ram.. When ram is produced, all ram is NOT considered equal. Would you expect cheap ram that was NOT intended to be used as high end, overclockable ram to work the exact same as the really high end stuff? Probably not.

I am not being negative about Corsair. To the contraire.. I own 4gigs of Corsair Ram myself. However, if you want to achieve higher clock speeds than where you are at currently, maybe a better investment of ram might be in order. If you are happy with 3ghz, then the ram you have is perfect for you. Also, consider many other factors.. Do you happen to be one of the unlucky ones with a botched processor? Is it at its limit? How effecient is your PSU? Is there an introduction of heat or high temps within your case that might cause instability? Do you have a bad stick of memory? Etc..
 
A few things about your setup.. It looks pretty solid. I would even say the Antec 500w PSU is fine (although, there are a few on here that hate Antec PSU's and would disagree with me).

For dual cores, you have a good 680i board. Options galore.

At 3.0ghz, that might be the spot where you need more VCORE. When I had my E6600 on my IN9 32X-MAX, at 3.1ghz, it need more VCORE. If your system boots fine and only reboots or locks after running prime or orthos, then at this point, it might not hurt to add a little VCORE.. My general rule of thumb is, if you fail prime SEVERAL minutes or an hour or so into it, bump it up only 2 notches from where you are at.. If it fails within just a minute or so, bump it up 4 notches.

Remember, the C2D's can take quite a few notches of voltage.. However, the HEAT is the by-product of VCORE increases and that can cause instability. When I refer to your RAM being the "weak link", I am refering to the specifics of the ram.. Often, the phrase is, "you get what you pay for". I find this to be a general guideline for ram.. When ram is produced, all ram is NOT considered equal. Would you expect cheap ram that was NOT intended to be used as high end, overclockable ram to work the exact same as the really high end stuff? Probably not.

I am not being negative about Corsair. To the contraire.. I own 4gigs of Corsair Ram myself. However, if you want to achieve higher clock speeds than where you are at currently, maybe a better investment of ram might be in order. If you are happy with 3ghz, then the ram you have is perfect for you. Also, consider many other factors.. Do you happen to be one of the unlucky ones with a botched processor? Is it at its limit? How effecient is your PSU? Is there an introduction of heat or high temps within your case that might cause instability? Do you have a bad stick of memory? Etc..

Again, thank you for your responses sir :). Let me reply

I have always had good PSUs from Antec. This NEO500 served me during my AM2 phase and so far, has been rock-solid (going by BIOS and uGuru readings) during my C2 phase. The PSU is rated at 85% efficiency @50C. 3 12v rails @17 each.

I've been overclocking a long time, well 11 years or so, and I know it's not a given ;), but I'm sure there's something else wrong here. The issue that's puzzling me here is the FSB. If I push it over 1500, even to just 1525, it *may* get into Windows, however quite often it will simply go to *safe* settings, requiring a reboot, and quite often, a CMOS reset. This is regardless of the voltage settings. From the research I did on this mobo, it should easily go higher than that. Hence why I tried dropping the memory and multiplier to test.

I'm not bothered about applying a higher vcore, my temps are just fine, even when pushing 1.4v, but the problem is, it craps out as soon as I go over 1500FSB, either in DOS or Windows (uGuru).

The RAM (at least in the AM2 setup) was rock @ 250 (2.1v) and appears fine here too. As I said before, I'm pretty rusty in the Intel scene, if FSB was the underlying problem, and especially with this board, what is the best setting to be changing? does the HT need more volts?
 
Violator and officermartinez,
Read the latest review of the IN9-32X Max at X-bit labs for some excellent insight on this main board and OC - especially memory settings and FSB Strap. It will help clear alot of issues and help with determining OC's. Linky:

http://www.xbitlabs.com/articles/mainboards/display/abit-in932x-max.html

hey officermartinez, sounds like your still diggin the XFX 680i LT. Been lookin at the DFI version for my son. Also, the Gigabyte 680i seems pretty sweet....but also $$$.

Also linflas (from Abit forums) PM'd me and let me know he went with a Stryker Extreme - and is in Quad Core OC'ing heaven;)
 
I could not get my memory stable at 800 with a 6600 running 400X8 linked/synched no matter what voltages I messed with and the memory was rated for it. Well I found the problem, most of the stuff I've read online says to lock down the pci-e frequencies and disable linkboost. Once I turned that stuff on auto and enabled linkboost and EX it ran Blend for 10 hours last night without a hitch. The board wanted the HT at 1.4v so I left that alone as well. I've been pulling my hair out for days trying to stabilize the memory, and lo and behold, let the board do more of the work for you. Newegg had a big sale on my memory, so I'm going to try 4gigs Wednesday and see what happens.

CPU CORE VOLTAGE.................... - 1.36
DDR2 SDRAM VOLTAGE.................. - 2.05
DDR2 REF VOLTAGE.................... - 0%
NB DDR2 REF VOLTAGE................. - 0%
CPU VTT VOLTAGE..................... - 1.30
NB VOLTAGE.......................... - 1.45
SB VOLTAGE.......................... - 1.50
HYPERTRANSPORT VOLTAGE............... - 1.40
 
Thanks for the reply and linky Missile Maker.

An update after reading that review. I tested 1600, 1650 and 1700 FSB and discovered a) my CPU/Mobo doesn't appear to like a 7x multi. When I used 1700FSB and a 6x multi, dropped the memory down to 750, windows booted OK. Ran Orthos and it gave me an error after 1:53 (would normally just hard-lock). Now, the CPU was at a 2.55Ghz which should be a breeze, and the memory was lower than normal. So what gave me the error there?.

Other tests (and this is really annoying, i.e. 1650 with an 8x multi, cranked up the vcore to 1.42, HTT to 1.4, and then it locks at the BIOS screen (needs a CMOS clear). This goes for various other settings also. Yet at other times it will simply give me 'safe mode' and drops the FSB and allows me back into the BIOS.

Might this be a BIOS issue rather than hardware? Do you think it's worth flashing to a different version?

*edit* just noticed that uGuru in Windows will give me an 'overclocking failure' at +1 or -1 over my current 375FSB setting, as I said, before it even tests it.

**edit** flashing to BIOS 1.1 and reinstalling uGuru did absolutely nothing. ANY suggestions for this? FSB wall on the chip? wait for a new BIOS? bad mobo? all takers welcome :)
 
Violator,
Are you cooling the NB at all? Is the breeze from the Tunique cooling the PWM HSF? What does uGuru report PWM temps as? Note: a 40x40x10 NB fan does alot to cool both the NB and the PWMs as both Heat Sinks are connected with the heat pipes.
1:53 on orthos seems like either a heat issue from the highly OC'd FSB or a voltage stability issue (less likely).
Also, what OS?

For info, I run BIOS 1.1 no problems, but if I was running a Quad, I'd use 1.2b3
 
1:53 on orthos seems like either a heat issue from the highly OC'd FSB or a voltage stability issue (less likely).

I am using the beta3 with a 60mm Stealth on the NB and until I did what was described in my previous post I got tons of 2 minute pulls. Over 3 minutes and I was in heaven. CPU would be fine, but when I tried to stress the RAM it would quit every time. After endless voltage tweaks I made it to 17 minutes and then 3 hours and that was the end of it and that took quite a few late nights. After enabling the extra goodies I was able to bring my RAM voltage down some, I was thinking of trying less voltage on other stuff, but I might just leave it alone because it works now. After I get the other 2 gigs stabilized I have to do a clean XP install because it thrashed everything. Part of me wants to try 3.6 again after my discovery, but I'm not sure it's worth it.

Also, here is some interesting news.
http://h20000.www2.hp.com/bizsupport/TechSupport/Document.jsp?lang=en&cc=us&objectID=c01020735&jumpid=reg_R1002_USEN
 
@ Buff,
My mistake, he's getting a Quad now that prices are dropping. His current OC is 4 GHz is on an E6600 (445MHz Bus) 1780MHz QDR running 4 GBs Crucial Ballistix PC2 6400 1:1 @ DDR2 890MHz and fairly tight 4-4-4-8 timings. He's still benching.
MM
 
Violator,
Are you cooling the NB at all? Is the breeze from the Tunique cooling the PWM HSF? What does uGuru report PWM temps as? Note: a 40x40x10 NB fan does alot to cool both the NB and the PWMs as both Heat Sinks are connected with the heat pipes.
1:53 on orthos seems like either a heat issue from the highly OC'd FSB or a voltage stability issue (less likely).
Also, what OS?

For info, I run BIOS 1.1 no problems, but if I was running a Quad, I'd use 1.2b3

I am not actively cooling the NB at the moment. Temps seem fine, the PWMs get up to mid 50s max when running Orthos full load for a few minutes, typically they're in the mid 40s.

I don't think the NB overheating would explain the hard-lock in the BIOS either. Although yes it may explain Orthos bailing out after less than 2 minutes when the FSB is high(er). Voltages in BIOS and uGuru are rock-steady, even under load.

*edit* Windows XP btw
 
officermartinez,
hanging around Abit Forums and the VideoCard section of this forum;) too busy with work too spend much time otherwise....I'm hooked on playing S.T.A.L.K.E.R. SoC.:p
 
Any more suggestions guys? I'm running low on ideas here.

Is it normal for this board to totally lock on the BIOS screen if it has a setting it doesn't like? If so, all I can say is thank God for the CMOS clear switch at the back.
 
Any more suggestions guys? I'm running low on ideas here.

Is it normal for this board to totally lock on the BIOS screen if it has a setting it doesn't like? If so, all I can say is thank God for the CMOS clear switch at the back.

Yes that is a tad unusual. Normally if you power down and wait 30 seconds it will revert to working settings.
 
Any more suggestions guys? I'm running low on ideas here.

Is it normal for this board to totally lock on the BIOS screen if it has a setting it doesn't like? If so, all I can say is thank God for the CMOS clear switch at the back.

Yeah.. My IN9 did that.. I one tried to set the FSB to 500 and it locked in the bios.. I had to hit the CCMOS switch in the back of the I/O panel
 
Well found out something new. The mobo refuses to POST if I enable SLI Memory (I have EPP enabled RAM). If I select any profile, nothing, it powers up for a few seconds, clunks away a bit then the monitor loses the signal.

Odds on memory problem?? Memory worked fine though in last motherboard

*edit* motherboard stops on C1 code, which is 'detecting memory'.
 
Got my other 2 gigs of Gskill last night. I was stable with 2gb at 4-4-4-12-1T at 2.050 vdimm. The extra 2 gigs did not want to stabilize at all and caused crashes, including Bios going into safe mode(haven't seen that before), even though Memtest showed no errors. I had to back down to 2T. After a good number of CMOS clears and three hours later it decided to cooperate. But it was unstable in Orthos at anything other than 2.00 vdimm. It seems hard to believe, but after the slight lowering of voltage, it ran blend for 8 1/2 hours with no errors. I was not expecting the RAM voltage to be that touchy given some of the overclocked voltages you see some people running on these sticks. And after I first fired them up I noticed they were a little warm, so I ran out to get a fan, and amazingly Staples had a low db 80mm that I mananged to wire tie in to blow on them . I'm so ready for this to be over, I've lost way too many hours and wire ties, it will be great when I finally get to use it instead of pulling out my hair ;) .

I've heard nothing but bad things about SLI Memory and EPP on this board. When my Bios went into safe mode it would take much longer to post, you'd see all the codes go by on the readout, and then it would stop me after post telling me to fix the settings or continue. Only way to make it stop was a CMOS clear.
 
It seems hard to believe, but after the slight lowering of voltage, it ran blend for 8 1/2 hours with no errors. I was not expecting the RAM voltage to be that touchy given some of the overclocked voltages you see some people running on these sticks. .

The IN9 for some reason requires lower memory voltages than other boards. Rule of thumb is lower it .075 to .1 volts from other motherboards..

If you run it too high this board does have issues.
 
So any chance, given my postings, this could be a bad board? I really feel like I'm hitting a brick wall here and if the board is duff, I'd rather RMA it and possibly get something else.

*edit* done full extended memory test now, no errors. Before this, the board locked up on me in the BIOS while I was changing some settings.
 
Thanks for the tip Spazilton.

Violator, I'm not quite sure what to tell you, but if you don't have that northbridge fan I would get one pronto. I used a 60mm Stealth and a handful of wire ties to get it on there, because the bracket is for 40mm, but I wanted more cfm and less db. I've been at this for god knows how many nights now, and it's been very frustrating, and time consuming to get stable at 3.2 with a 6600. I'm running a vacuum on my case courtesy of my radiator mounted to the top of the case, so it pulls air through the backplane grill onto the pwm, I would recommend a fan on these as well unless your doing what I am. You might also try pulling the heatpipe assembly and using Ceramique on the NB and SB and putting a little more tape on the pwms. I haven't done this, but it has helped others. Check the thread at Xtremesystems in the review section.

The only time I use Uguru in Windows is to monitor temps and tweak voltages when I'm getting close. Almost everything is done in the bios. I had nothing but problems trying to use Uguru for anything else. And remember, there are plenty of FSB holes to fall into, and this is a tweaker board so pack a lunch. I've read every thread on this board twice, made notes of setups that worked for others, and in the end still wound up feeling my way in the dark. Linked and synched with linkboost enabled has worked for me. Letting the HT go up to 1.4 and letting the board overclock the pci-e etc was also a big help.
 
Thanks for the tip Spazilton.

Violator, I'm not quite sure what to tell you, but if you don't have that northbridge fan I would get one pronto. I used a 60mm Stealth and a handful of wire ties to get it on there, because the bracket is for 40mm, but I wanted more cfm and less db. I've been at this for god knows how many nights now, and it's been very frustrating, and time consuming to get stable at 3.2 with a 6600. I'm running a vacuum on my case courtesy of my radiator mounted to the top of the case, so it pulls air through the backplane grill onto the pwm, I would recommend a fan on these as well unless your doing what I am. You might also try pulling the heatpipe assembly and using Ceramique on the NB and SB and putting a little more tape on the pwms. I haven't done this, but it has helped others. Check the thread at Xtremesystems in the review section.

The only time I use Uguru in Windows is to monitor temps and tweak voltages when I'm getting close. Almost everything is done in the bios. I had nothing but problems trying to use Uguru for anything else. And remember, there are plenty of FSB holes to fall into, and this is a tweaker board so pack a lunch. I've read every thread on this board twice, made notes of setups that worked for others, and in the end still wound up feeling my way in the dark. Linked and synched with linkboost enabled has worked for me. Letting the HT go up to 1.4 and letting the board overclock the pci-e etc was also a big help.

Thanks for the response Tnoggin :)

Let me summarise my experience so far :

System will POST/Boot with FSB set to 375, linked, 3:2 divider (CPU 3Ghz, mem 1Ghz). Will pass Prime/Orthos at this speed.

System is not stable at any other setting other than default, particularly if changing from the setting above. An increase or decrease of even 1 on the FSB will result in instability.

Unstable settings frequently hard-lock the BIOS, requiring a CMOS clear (powering down for 30 secs does NOT resolve the issue)

Hard-locks in windows require a reboot, the BIOS will then more often than not hard-lock (as above)

System will NOT post when SLI memory is enabled (in any way). Mobo sticks on C1 error (detecting memory). Memory has been tested using Extended Memory Diagnostic (inc cache) and passed 100%. Also been reseated, and tested with only one module with the same result.

Adjusting any voltages does not seem to make any difference

Temps are absolutely fine. CPU idles low 30s and under load high 30s/low 40s. PWMs are mid 40s idle and low 50s under load. Voltages are rock steady on all rails. The Tuniq Tower is set up so that it pushes hot air towards the rear-exhaust.

Have tried BIOS 1.0, 1.1 and 1.2 all with same results


It's incredibly frustrating. This board should be a monster and yet a thing of beauty also :), but it's just not happening for me.
 
Get that fan, if I remember correctly, the board is only good for stock settings without the fan.

"...In conclusion I would only like to add that you should try to avoid not only the FSB frequencies between 400 and 450MHz, but also 325MHz FSB, and 360MHz-380MHz interval. Too many limitations for the top-of-the-line chipset, don’t you think so?..."

"...The results of our benchmark suggested that the mainboard worked fine only at FSB frequencies that were multiples of 50: 300MHz, 350MHz, 400MHz, 450MHz, and at frequencies that were multiples of 25 we observed unexpected performance drops..."
http://www.xbitlabs.com/articles/mainboards/display/abit-in932x-max_10.html
 
Get that fan, if I remember correctly, the board is only good for stock settings without the fan.

"...In conclusion I would only like to add that you should try to avoid not only the FSB frequencies between 400 and 450MHz, but also 325MHz FSB, and 360MHz-380MHz interval. Too many limitations for the top-of-the-line chipset, don’t you think so?..."

"...The results of our benchmark suggested that the mainboard worked fine only at FSB frequencies that were multiples of 50: 300MHz, 350MHz, 400MHz, 450MHz, and at frequencies that were multiples of 25 we observed unexpected performance drops..."
http://www.xbitlabs.com/articles/mainboards/display/abit-in932x-max_10.html

Well, firstly I don't think I can fit a fan on the NB with the Tuniq in place. Second, the board is OK at 375FSB. It's any other setting near enough that it doesn't like, and that's higher or lower. If the NB was overheating it wouldn't explain the BIOS lockups as they happen almost immediately. I also checked the NB today and while it's warm, I can certainly hold my finger on there without any discomfort.
 
Well, firstly I don't think I can fit a fan on the NB with the Tuniq in place. Second, the board is OK at 375FSB. It's any other setting near enough that it doesn't like, and that's higher or lower. If the NB was overheating it wouldn't explain the BIOS lockups as they happen almost immediately. I also checked the NB today and while it's warm, I can certainly hold my finger on there without any discomfort.

Its not the NB overheating its the VRM and SB. The SB on mine without cooling got so hot I actually burnt my finger. VRM at anything over 70c which I have seen at idle in the bios hard locks this board.

That is why I will never again buy a motherboard that has no active cooling.
 
If the NB was overheating it wouldn't explain the BIOS lockups as they happen almost immediately. I also checked the NB today and while it's warm, I can certainly hold my finger on there without any discomfort.

That's not entirely true. Back when I overclocked the A7N8X and AMD Barton I had to lap the northbridge chip and put on a copper Iceberg to get it stable, it would do all sorts of strange things until I did that. Heat is an enemy when it comes to overclocking, plain and simple, it only takes one part of the chipset to get hot or have poor heatsink contact to cause all sorts of problems.

Remember, thanks to that heatpipe, all three sections are connected.
 
Spazilton and Tnoggin, thank you for the replies :)

I can understand your points about the overheating, and Spazilton you mention this can cause this mobo to hard-lock in the BIOS. That's fair enough (and my next comment is in no way doubting you, it's just odd)

Based on that, can you explain why dropping the settings, and by that I mean FSB/Multiplier/Memory from a previously stable 375 would also cause the board to hard-lock. I'm talking maybe down to 370, or 350, or even 300. I've tried them all and they all have the same effect.

By the same token, if I use uGuru to change the FSB (as I've said previously), from 375 to anything else it reports an overclocking failure (but will set itself to the new setting, and will hard-lock after a short period of time), if I then reset it back to 375 it does not give an error.
 
From my experience this motherboard does all kinds of odd things. So starting with the basics like heat control will help you narrow it down. This board should have come with active cooling to begin with. I thought I had mine figured out. It ran 8 and a half hours on blend last night without a problem. First boot today and it wouldn't run 3 minutes with the exact same settings, dropped the RAM voltage to 1.95 and I'm off again, it's ran 90 minutes on blend so far. I've easily been at this for the last two weeks and everytime I get close, it throws me a curveball. I've pretty much given up on attempting 1T with 4gigs because my patience is now wearing thin at 2T.
 
At least with my quad, the most critical setting for me is CPU VTT voltage. I have to leave it at the lowest (1.2v) or else I cannot get into Windows. The only time I can set it higher and get into Windows, actually, it added some stability in this case to, is when I want to run my ram 1:1 1-T with higher than 333fsb. I am able to bench at 360fsb but I do have to add VTT but my system is not 100% stable at that fsb. Below 333fsb and anything above 1.2 VTT=NO Windows.

Incidentally, I also have 4 1GB sticks and I can run them at 3-3-3-5 1-T 669MHZ 2.250 vdimm. Any faster and I have to go to 2T.

Also, I lose about 200MB/s in memory bandwidth read tests going from 330fsb to 333fsb, so there is definitely some fsb holes to be aware of with this mobo. I'd imagine the particular type of cpu may cause variances in where the holes are, but just something to definitely be aware of.

As for PWM temps, I have a 60mm fan strapped to the heatpipe, have replace the NB and SB pads with AS5 and have doubled the heat tape on the VRMs and I think that dropped 20-25c from my PWM temps, but even still, my PWMs idle at 50c get get up to 80-85c with 4 instances of P95 testing small FFT's. When I run the cpu at 3.5MHz, 1.49vcore, PWMs get to about 115c doint the same tests. Needless to say, I don't let that run too long, but before I doubled the heat tape and applied the AS5, I couldn't clock this chip above 3.3GHz without the PWM temps tripping a shutdown.

I really am starting to feel as comfortable with this mobo as I was with my old Max3, but I still miss playing with my DFI Lanparty!

Running bios 1.2B3 and hopefully abit will do something about this low fsb with quads soon, but for now, I am very happy with the performance and stability.
 
Spazilton and Tnoggin, thank you for the replies :)

I can understand your points about the overheating, and Spazilton you mention this can cause this mobo to hard-lock in the BIOS. That's fair enough (and my next comment is in no way doubting you, it's just odd)

Based on that, can you explain why dropping the settings, and by that I mean FSB/Multiplier/Memory from a previously stable 375 would also cause the board to hard-lock. I'm talking maybe down to 370, or 350, or even 300. I've tried them all and they all have the same effect.

By the same token, if I use uGuru to change the FSB (as I've said previously), from 375 to anything else it reports an overclocking failure (but will set itself to the new setting, and will hard-lock after a short period of time), if I then reset it back to 375 it does not give an error.


Uguru never sucessfully overclocked(or changed FSB settings at all) this motherboard for me once. I find the application to be quite useless.
 
At least with my quad, the most critical setting for me is CPU VTT voltage. I have to leave it at the lowest (1.2v) or else I cannot get into Windows. The only time I can set it higher and get into Windows, actually, it added some stability in this case to, is when I want to run my ram 1:1 1-T with higher than 333fsb. I am able to bench at 360fsb but I do have to add VTT but my system is not 100% stable at that fsb. Below 333fsb and anything above 1.2 VTT=NO Windows.

Incidentally, I also have 4 1GB sticks and I can run them at 3-3-3-5 1-T 669MHZ 2.250 vdimm. Any faster and I have to go to 2T.

Also, I lose about 200MB/s in memory bandwidth read tests going from 330fsb to 333fsb, so there is definitely some fsb holes to be aware of with this mobo. I'd imagine the particular type of cpu may cause variances in where the holes are, but just something to definitely be aware of.

As for PWM temps, I have a 60mm fan strapped to the heatpipe, have replace the NB and SB pads with AS5 and have doubled the heat tape on the VRMs and I think that dropped 20-25c from my PWM temps, but even still, my PWMs idle at 50c get get up to 80-85c with 4 instances of P95 testing small FFT's. When I run the cpu at 3.5MHz, 1.49vcore, PWMs get to about 115c doint the same tests. Needless to say, I don't let that run too long, but before I doubled the heat tape and applied the AS5, I couldn't clock this chip above 3.3GHz without the PWM temps tripping a shutdown.

I really am starting to feel as comfortable with this mobo as I was with my old Max3, but I still miss playing with my DFI Lanparty!

Running bios 1.2B3 and hopefully abit will do something about this low fsb with quads soon, but for now, I am very happy with the performance and stability.

Hey thanks I always ran the VTT at 1.3 and couldn't get over 300 FSB on my quad. I'm going to try it at 1.2.

I am at a crossroads. Either get a P35 motheboard or get the DFI 680i. The DFI is getting 430+ FSB on quads with tweaked Vgtl settings. I have one coming in next Thursday. I just want 3200 out of this quad and I would be more than happy.
 
@Violater, et al.
Well I can't figure the main board either. Check this in my specs - when I'm OC'd to 3.2 GHz on this E6600, the memory switches to 1T Timing...even though I have 2T set manually in BIOS (on 1.1)
Settings when OC'd
CPU Operating Speed 'Unlinked'
FSB<QDR>1425MHz (Bus 356MHz)
FSB:MEM Ratio Auto (but grayed out cause I'm unlinked) yet CPU-Z and Everest report 1:1 - don't know if this is a default because I'm running the memory bus 'unlinked' from the FSB, and CPU-Z reports it this way (didn't notice it before)
MEM<DDR2> Manually set to 800MHz
uGuru BIOS Reported 791MHz (But I can increase the above to around 814 MHz to get reported 800MHz , however, if I leave it at the BIOS reported 791MHz, both CPU-Z and Everest report it spot on 800MHz once I'm in OS)
FSB STRAP by PSB1066MHz
Timings manually set to 4-4-4-12 2T (reported 4-4-4-12 1T when OC'd:confused: )
VCore 1.42 manually set for stability (Reporting 1.38 in CPU-Z)
VDimm 2.1V (where I've always kept it for tight timings)
OC'd Orthos stable and looped 3DMark06, but can't believe 1T CMD timings, especially when running 4 x 1 GB - got to believe it would be stressing the shit out of the memory controller on the NB (when I run Stock, with same memory BIOS settings, the CMD timings report correctly at 2T). Yes this is EPP memory - two matched pair.
Go figure....:eek:
 
Hey thanks I always ran the VTT at 1.3 and couldn't get over 300 FSB on my quad. I'm going to try it at 1.2.

I am at a crossroads. Either get a P35 motheboard or get the DFI 680i. The DFI is getting 430+ FSB on quads with tweaked Vgtl settings. I have one coming in next Thursday. I just want 3200 out of this quad and I would be more than happy.


Yes, definitely give dropping VTT a shot. Also, not sure if I was clear on this, but I cannot go above 333fsb unless I run my memory 1:1 (Linked and Synced), stable up to 340fsb, benching at 360fsb until BSOD :)

For you and your multi, I can see where you'd want to try a diff mobo, but in my case, it's really just all in my head, and of course synthetic benches. The real-world performance gains I'd see if I had a mobo that could run 400+fsb wouldn't really justify the cost, especially since I'm really just a big gamer for the most part :D
 
Back
Top