Retail Abit IN9 32X-MAX - Pictures, observations and testing...

Now that I have left my system up for a while, well, more than 24 hours, it seems very stable.. am running with essentially the max stable OC I could get with my cpu/mem on the p5w dh..


3.5ghz @ 1.5vcore (but I think on this board it might need less, still testing that)
fsb @ 1556, 9x multi
pci-e locked @ 105mhz
mem linked + sync'd for 778mhz effective @ 3-3-3-9 1t (the p5w dh would only do 2t tho)

CPU VTT @ 1.50
NB @ 1.45
SB @ 1.60
HT @ 1.30

Does the CPU VTT on this board equate to CPU "HT" from the 680i tips thread? in that thread it said to run at @ 1.50v (max) for best stability.. I didn't see ny excess heat or anything when I turned it up, so I left it there..

I played LOTRO for like 6 hours yesterday with these settings and there were no issues what-so-ever.. but my PWM is getting into the 80s with the stock sink.. so I am going to fall back to my aftermarket setup when I yank my psu on Sunday to swap it for the newer revision one. The sound coming from the PWM seems to be a little quiter, else I am getting used to it. heh.. I'm going to camp on my current settings to let the board burn in a little.. then try some more OCing when I get the new psu.. I think this cpu has 3.6ghz stable in it.. ..(think/hope).. I could bench with that on the p5w dh but was not prime/orthos stable at all.. I think it might have been the mobo tho.. didn't seem to behave well when I dumped the multi to 8x and ran @ 400 fsb.. anyways, I want to try this mobo @ 8x to see what she'll do. :)

guys.... I can't get more than 400MHz fsb :(

is it mandatory to increase SB voltage? I doesn't change it.

even with VTT 1.5, NB 1.5, HT 1.4 - I can't hit more than 400MHz.

actually, I think the reason is the low HT voltage and it's needed to increase SB voltage to get it more stable on high fsb's.

with both Asus 680i boards I was able to hit 500MHz easily with default SB voltage, 1.52v HT, 1.5 VTT, 1.5 NB.

my bios is official 1.0. So, I see that there are no 1333 strap and if I set it as "by CPU" I'm able to boot with 450MHz but it's unstable even in windows idle. I see that mem latency goes slower +20ns -> it means that 1333 strap enabled, but it looks like not all chipset settings was set correctly since it's really unstable.

Does anyone has beta4 bios? Could you please share it?

I'm going to try beta1 and beta2 today.

ps: am I right with USB keyborad settings like OS and not BIOS?

revenant, please note that with low multi - your NB works on higher clocks. (NBCC freq)

i didn't notice anywhere but do any of you know if it has problems with sli and sound cards like asus and evga have on theres ?

Quote:

---

Originally Posted by Nilson

revenant, please note that with low multi - your NB works on higher clocks. (NBCC freq)

---

yep... I'm wondering about the lower multi creating a hit on memory BW also.. FYI, no beta4 bios yet.. unless it's leaked somehow.. keep an eye here for it: http://forum.abit-usa.com/showthread.php?t=119222

Quote:

---

Originally Posted by Nilson

guys.... I can't get more than 400MHz fsb :(

is it mandatory to increase SB voltage? I doesn't change it.

even with VTT 1.5, NB 1.5, HT 1.4 - I can't hit more than 400MHz.

actually, I think the reason is the low HT voltage and it's needed to increase SB voltage to get it more stable on high fsb's.

with both Asus 680i boards I was able to hit 500MHz easily with default SB voltage, 1.52v HT, 1.5 VTT, 1.5 NB.

my bios is official 1.0. So, I see that there are no 1333 strap and if I set it as "by CPU" I'm able to boot with 450MHz but it's unstable even in windows idle. I see that mem latency goes slower +20ns -> it means that 1333 strap enabled, but it looks like not all chipset settings was set correctly since it's really unstable.

Does anyone has beta4 bios? Could you please share it?

I'm going to try beta1 and beta2 today.

ps: am I right with USB keyborad settings like OS and not BIOS?

---

b01 will be easier for you to overclock with this board...:)

Anyone thought of grinding down these ends so that you could use AS5 instead of that thick thermal pad?

I'm tempted to try.

Quote:

---

Originally Posted by kiner

b01 will be easier for you to overclock with this board...:)

---

beta1? can't find it :( just tested beta2... no luck... 400MHz stable and thats all... any voltage - can't boot with following fsb: 430, 440, 450, 460, 470, 480, 490 :( (volts: 1.4v HT, 1.5v VTT, 1.55v NB and 1.65v SB)

my E6600 able to run 32M with 515x9 on Asus boards. So, no issues with CPU limitation. PWM temps goes around 50-60C.

we need at least 1.55v HT voltage! even on Asus 680i boards I was not able to hit 460-500MHz stable on 1.4v HT...

furthermore... I found a bug in beta2. It's regarding strap...

so, on common 680i motherboards you have nice latency and NB settings till 425MHz fsb. With beta2 bios (can't boot with 420MHz on bios v1.0) we have BIG delays even on 420MHz. Strap tested: "By CPU" and "1066". Look at this:

It proves hipro5 words about "bios v1.0 faster in terms of memory bandwidth"

Correct:
Attachment 55601

Not correct:
Attachment 55602

@Nilson: I agree that HT limitation of 1.4v is something Abit has to seriously consider in their next bios.
has been a key voltage for high fsb's and stability in 680i platforms.
I had the p5n32-e sli and the striker extreme and ht's were 1.5v or above for 400-500 fsb's. although i have to admit those boards where nowhere close to abit's superb voltage control and accuracy.

Another thing i noticed is that 680i's don't like low multi's and high fsb combinations, that showing in memory benchmarks (bandwidth-latency) and pi times.So please check your high fsb oc's cause you're probably turning your 680i beast into a slower pc.

@Neorecon: do yourself a favor and read back in this thread. it's one of the best threads i've seen and pretty much everything about this board has been covered here.

@momomo6789: sli is not a problem with this board. doing quad 2x gx2 stable @ 3.4g w/ qx6700.
raid has no problem either.
no sound issues (cracks-pops) with the onboard audio. haven't tried an audio card.(no space!!) had pops and crackles with both asus boards and the analog devices hd audio chip.(lowering output levels helped reduce peaking-digital distortion.)

@revenant:cool to see you're back tweaking again!

LOL. I thought I had already but I guess I missed that.

Well my board is rock solid with my E6600@3.6Ghz 1.45v and memory at 800Mhz but anything above it it crashes with Orthos at about 8 minutes. However I was very gentle with the SB voltages as well as HT etc so it could be a reason since I did not increase those. It boots at 3.8Ghz and could also complete a 3dmark06 for 15800 3Dmarks but it was at 1.72volts for CPU.

Anyway my problem is that i need to put a fan on the PWMs since at 1.7v they hit 100c and my southbridge gets really hot since I am only having the cut off stock piece (I cut the stock cooling in pieces). My NB is watercooled so no issues there.

I will retest.

Quote:

---

Originally Posted by Nilson

beta1? can't find it :( just tested beta2... no luck... 400MHz stable and thats all... any voltage - can't boot with following fsb: 430, 440, 450, 460, 470, 480, 490 :( (volts: 1.4v HT, 1.5v VTT, 1.55v NB and 1.65v SB)

my E6600 able to run 32M with 515x9 on Asus boards. So, no issues with CPU limitation. PWM temps goes around 50-60C.

we need at least 1.55v HT voltage! even on Asus 680i boards I was not able to hit 460-500MHz stable on 1.4v HT...

furthermore... I found a bug in beta2. It's regarding strap...

so, on common 680i motherboards you have nice latency and NB settings till 425MHz fsb. With beta2 bios (can't boot with 420MHz on bios v1.0) we have BIG delays even on 420MHz. Strap tested: "By CPU" and "1066". Look at this:

It proves hipro5 words about "bios v1.0 faster in terms of memory bandwidth"

---

http://xyz2007.myweb.hinet.net/IN9_32X_MAX_11.B01

try it... good luck... ;)

Should I bother with watercooling the SB or is a TT Extreme Spirit II enough?

The funny part now is that I have a PA120.2 for the NB only - a bit overkill but CPU and GPUs have also their dedicated loops and radiators.

i hope beta 4 comes next weeek :(

@ JargonGR: A re Patrida!..Will the TT fit in for sli? have you tried it? if yes it should do the job.
ideally i would put pwm nb and sb on the pa120.2 loop.. if i had the pwm and sb blocks that is.
1.72v for vcore is very close to the edge... did you play with vdimm? it should help 3d benching.

Quote:

---

Originally Posted by OKsoVRomoSkylo

@ JargonGR: A re Patrida!..Will the TT fit in for sli? have you tried it? if yes it should do the job.
ideally i would put pwm nb and sb on the pa120.2 loop.. if i had the pwm and sb blocks that is.
1.72v for vcore is very close to the edge... did you play with vdimm? it should help 3d benching.

---

Hehe Patrida ole!

Well I managed to fit it using the screws from the Alphacool Waterblock I had and it fit perfectly. I will order the Blocks from MIPs if after the new BIOS I can't overclock more than 3.6ghz. This chip could do 3.6GHZ on air but not even boot any further. Now I can boot and work in windows at 3.8ghz but was no go for Orthos after 13mins. Ofc I did not play with the voltages at all since the SB was badly cooled. Now I will play around and see. I don't think the 1.7v is really needed since it also boots at 1.65v.

Will check tomorrow since I have to go out now.

Thnx for the tips.

Pou vrethikane tosoi Ellines re? :D

Is there an aftermarket heatsink for the PWMs? Not for watercooling, only the CPU will be watercooled.
Any other useful hints/tricks for this "hot" area?


*Also, does this board suffer from the multiplier issues of the QX6700/AW9D combo?
http://img50.imageshack.us/img50/6349/abitwtfmg3.th.jpg

Quote:

---

Originally Posted by momomo6789

i didn't notice anywhere but do any of you know if it has problems with sli and sound cards like asus and evga have on theres ?

---

Im running 8800GTX SLi and an X-fi, not noticed any sound problems at all with this board!

They skipped to beta .B05.

Quote:

---

Originally Posted by Yellowbeard

They skipped to beta .B05.

---

That doesn't bode well...

Quote:

---

Originally Posted by Yellowbeard

They skipped to beta .B05.

---

what ? why the canceled B03 and now B04 ?

When will B05 available ?:confused: :confused:

Quote:

---

Originally Posted by Chosen.

Pou vrethikane tosoi Ellines re? :D

Is there an aftermarket heatsink for the PWMs? Not for watercooling, only the CPU will be watercooled.
Any other useful hints/tricks for this "hot" area?


*Also, does this board suffer from the multiplier issues of the QX6700/AW9D combo?
http://img50.imageshack.us/img50/6349/abitwtfmg3.th.jpg

---

haven't found an after market solution yet, and thats why i hacked the otes.. but i found this: http://forums.hexus.net/showthread.php?t=99096
As u can see the little brother QuadGT, has a copper pwm stock sink.
If abit are selling these, they would be a good solution.

also found a similar pwm layout on a foxconn 975 board over at O2:
http://www.overclockersonline.net/im...e/layout-2.jpg
looks like aluminum, worse than the abit one.

hopefully there will be other aftermarket solutions since the pwm trend is here to stay.

About the multi problem, can you provide a link?
is the problem real or one of cpuz's occasional bugs?

Well I am all set with the NB (watercooled) and an Extreme Spirit II on the SB (had to play a bit to fit it with SLI). For the PWMs I just used the stock heatsink and folded the thermal pad in the middle as Johny Bravo did.

But if I used an extreme voltages on the CPU 1.68 -1.72 then the PWMs get close to 100C in Orthos. With 1.45v they hover at around 55c-60c on load max depending on room temperature.

So now I will check if I can stabilise it @ 3.8GHZ by playing around with the VTT and the other voltages a bit.

Will post results soon.

http://mips-computer.de/ these guys sell waterblocks for the PWMs/SB/NB althouth a bit expensive ones.

Quote:

---

Originally Posted by JargonGR

Well I am all set with the NB (watercooled) and an Extreme Spirit II on the SB (had to play a bit to fit it with SLI). For the PWMs I just used the stock heatsink and folded the thermal pad in the middle as Johny Bravo did.

But if I used an extreme voltages on the CPU 1.68 -1.72 then the PWMs get close to 100C in Orthos. With 1.45v they hover at around 55c-60c on load max depending on room temperature.

So now I will check if I can stabilise it @ 3.8GHZ by playing around with the VTT and the other voltages a bit.

Will post results soon.

http://mips-computer.de/ these guys sell waterblocks for the PWMs/SB/NB althouth a bit expensive ones.

---

These Waterblocks are a dream :) i got waterblocks for my W64 Professional from them and i live just an hour away from Mips Computer :cool:

Quote:

---

Originally Posted by OKsoVRomoSkylo

About the multi problem, can you provide a link?
is the problem real or one of cpuz's occasional bugs?

---

It's one of the early spotted issues with the AW9D, discussed in depth here.
several tests have been done to prove it:
1) CPUz 1.33 or earlier shows the different speed of each core.
2) If you run 4 SuperPi's assigned to each core individually (affinity=0, etc.) you will get 2 high and 2 low scores.
3) In Cinebench the ratio between 1 CPU and x CPUs (=4 CPUs) varies from 3x to 3.5x, depending on whether one of the slow cores have been used for the "1 CPU" part of the test.
4) wPrime detects the speed of the slower cores, so CPUz 1.38 will tell you 12x300 and after you get the score, wPrime will tell you 10x300.

My only concern is if the IN9 suffers from the same issue, cause mine is on its way...