Asus M4a79 deluxe up close and personal

[bingo13]

Current settings: 3.4 @ 1.47v (235x14.5) 940mhz cas 4 4 4 12 HT 2115 NB 2350 Pretty happy with everything except NB clocks. Any suggestions? NBv= 1.28 atm, CPU NBv @ auto HTv @ 1.24

Raise CPU/NB voltage up to 1.300V and try NB speed at 2585, keep raising CPU/NB (or lowering it) until that setting is stable. Then raise CPU/NB by .05 and try 2820, once again raise it more or lower it until that setting is stable. Having NB speed at or near 2800 is optimal without going crazy on volts.

I have five 940's and they all do 2800NB with CPU/NB around 1.350~1.400 with the CPU at 3800 (two will do 3100 at 1.450V, the other three lock around 3000 on air), this is under Vista 64 which will require ramping the voltages just a little higher than on Vista 32 or XP. If you take your CPU above 3.6, be sure to increase SB to 1.35V and HT to 1.28V when pushing the HT reference clock.

[chew*]

I have been lazy on this board, time to get serious, It made it to my main bench table, Ripping in a fresh OS now......I will start work on this tuesday as work is going to keep me tied up late tonight and most of monday.....I will use a retail 940BE to start with......

[mattkosem]

Are there any tricks to getting this board working with a ref clock over 245 or so? I was trying to do some cas 4 runs, which I know my memory is up to, but jumping the gap from 245 to 250 seems to bring on some nastiness. I get crackling over the audio when I try to switch it in windows. I tried the drive strength and clock skew settings that you posted on page 2, but they don't seem to be helping. I've got my HT multi down two notches, and my NB multi way down into the safe zone but cannot seem to get it running. I double checked the sticks on my GF's (intel) system and they still run DDR2 1040 at 4-4-4-12. I'm trying 4-5-5-12 and am not getting very far.

I'm going to stick memtest86+ on my thumb drive and see if I can find a bios setting that helps.

--Matt

EDIT: Very peculiar. I can set the settings in the bios and get into windows without any problems. Memtest86+ passes 5 passes of tests 3 and 5 with DDR2-1000 4-5-5-10. Playtime :-)

[chew*]

Matt those settings were memtest stable thats as far as I got and they were Micron D9 GMH, unfortunately now they are micron D9 half baked (my fault) .....back to using my OCZ's. Really don't want to invest much more money in ddrII.

[mattkosem]

Matt those settings were memtest stable thats as far as I got and they were Micron D9 GMH, unfortunately now they are micron D9 half baked (my fault) .....back to using my OCZ's. Really don't want to invest much more money in ddrII.

I edited my post. I can get into windows no problem if I set the settings from the bios, even with all of the drive strengths set to auto. It must just be a strange AOD bug.

--Matt

[mattkosem]

It seems to turn around some mildly impressive numbers at cas 4, for 2x2GB cas6 rated sticks at least. I can get them up to 1024, but it requires switching to unganged for a nice 1000mb/sec drop in read speed. Overall, I'll probably stick with this. It's a touch faster than 1200 cas 6 as far as I can tell, and the NB clock isn't too high to boot windows with CnQ enabled :-)

Also, it doesn't really seem to make any sense. But I'm able to achieve a slightly higher CPU clock with this ref clock and multi than I am with a much lower ref clock and multi. I'm pretty sure it's not my imagination either...

--Matt

[chew*]

Matt I think you will find that the happier you can keep the IMC the more performance and clock you will get out of the cpu

[iandh]

BIOS aside, which between this board and the UD5P would you say "feels" more stable while OC'ing?

Do you think you could do a really quick pro/con for the two?

Thanks for your time.

[mattkosem]

Matt I think you will find that the happier you can keep the IMC the more performance and clock you will get out of the cpu

Thanks! That seems to be the case! I'm not sure what the story is with AOD and high reference clocks. I tried it again and it reset my bios! I'll stay out of there for now! I also noticed that there isn't any option for setting the tRC with the DDR800 ratio. Is that common? What's the max safe voltage for the CPU/NB core? I've been sticking with 1.45, but I understand there are boards with a single power plane that lock vcore and NB vid. Is that one safe to 1.55 too? I did a bit of research and found that, in spite of the everest bandwidth scores being lower, unganged is faster than ganged. It seems like only single threaded apps benefit from ganged mode. I even did a little independent research and found it to be true.

--Matt

[chew*]

BIOS aside, which between this board and the UD5P would you say "feels" more stable while OC'ing?

Do you think you could do a really quick pro/con for the two?

Thanks for your time.

No clue yet, it seems stable at stock installing Sp3 seriously haven't got that far yet, the one time I was in windows it was on a OS geared for my 790GX, 90% of drivers were wrong...

Personally I like the layout of the gigabyte better for everyday use....I have 2 usable slots with this board when doublewide PCI X cards are installed. I had 3 usable slots with the gigabyte.

I can however run tri fire on the asus but with dual slot coolers I have no clue how they will stay cool......2 of the fans would have no breathing room....

I had to shove my post code reader in between the cards.....luckily it has an add on cable for an external mounted LCD as it's not viewable sandwiched in there.

On a plus side the ram slots are thought out better, but tall ram spreaders are still an issue in the inner slots.

Clear cmos button would have been a nice touch for a board that costs a little more....Would have liked them to place the sata ports all off the side of the board as well....and give more than 5 of them.

Enough of my rambling.

[chew*]

Here's the bios I was waiting for......

Improves system stability with certain memory modules......

1002 http://support.asus.com/download/dow...Language=en-us

[bingo13]

Here's the bios I was waiting for......

Improves system stability with certain memory modules......

1002 http://support.asus.com/download/dow...Language=en-us

There will be another one out next week or the week after. This one has been tuned with results from a dozen of the top selling DRAM kits in it, the next one will be tuned further and hopefully 4x2GB will be up to speed completely. However, several problems were addressed.

[chew*]

Cool, noticed the TRC bug is fixed in this, had to set 1 up to get the desired timing in windows.....etc 27 in bios equaled 26 in windows..guess I will have to start all over with drive strengths.......

[chew*]

Bingo,

Any thoughts as to why my port 80 diagnostic card is registering FF throughout the whole boot cycle?

It's installed in the available PCI slot visible in the pic in above post......

I really hate not knowing why a setting fails to boot..

seems I need more voltage to get my OCZ's to even boot on this board than the gigabyte......only needed 2.25 for 1180mhz on 790gx and they were 100% stable, they boot into memtest at 2.3v but are tossing alot of errors.....will have to tackle drive strengths tuesday.....

[_d_]

i will buy one

[HardCode]

Has anyone else noticed that loading the defaults completely screws with the vcore until its set over ~1.4v? I updated to the newest BIOS last night and loaded the defaults only to find my vcore was reading 1.50v in Windows. Setting the vcore manually to 1.35v resulted in 1.475 actual in Windows. I found that if I load the defaults I have to set the vcore at or over 1.4v, save, then reboot for anything under that to set correctly. It's not really a nuisance, just a really strange oddity.

I love this board though. On my Gigabyte mATX board I couldn't boot into Windows with the NB over 2.2GHz and it wouldn't remain stable after any type of NB bump...but this board allows me to run my NB at 2.8Ghz and has no problem with higher voltages.

I'll post some results tonight when I get home

[bingo13]

Bingo,

Any thoughts as to why my port 80 diagnostic card is registering FF throughout the whole boot cycle?

It's installed in the available PCI slot visible in the pic in above post......

I really hate not knowing why a setting fails to boot..

seems I need more voltage to get my OCZ's to even boot on this board than the gigabyte......only needed 2.25 for 1180mhz on 790gx and they were 100% stable, they boot into memtest at 2.3v but are tossing alot of errors.....will have to tackle drive strengths tuesday.....

I have no idea, still trying to figure it out. My POST codes are just fine actually. However, I am able to reach DDR2-1300 now but my NB speeds are limited to 2.6GHz on two CPUs that both did 3.1GHz on the 0903 BIOS. It is always something.... I have a meeting with them tonight.

[chew*]

I have no idea, still trying to figure it out. My POST codes are just fine actually. However, I am able to reach DDR2-1300 now but my NB speeds are limited to 2.6GHz on two CPUs that both did 3.1GHz on the 0903 BIOS. It is always something.... I have a meeting with them tonight.

try the PCI slot above the lower blue PCI X slot. Just solid FF during entire post even when it fails to post, will pull a card and try in another slot.

Seems I was wrong memory is passing now with less volts, had to give vcore a boost at 3080 mhz though......doesnt make much sense this cpu was tested prior does 3.2 fine at stock volts...

EDIT now all of a sudden port 80 is working.......changed DPS setting for memory..

[mattkosem]

I'm having terrible luck with 1002 as well. I couldn't get 3.7Ghz stable on the core with any NB or memory frequencies. The best I could do was half of an iteration of linpak. I also couldn't get DDR 1200 stable, even with 3.49Ghz on the core and a 2250mhz NB. I was running with the exact same timings, voltages, and skew settings that work perfectly on 801. This bios seems worse than 903 on my setup. I'm back on 801 and happy (and stable) as ever :-)

Bingo: What NB voltage are you running to attain 3.1ghz?

--Matt

[chew*]

Odd I just completed cinebench @ 3960/2860NB/586ram with my cheap OCZ golds.....not prime or linpack but still one of the hardest benches to complete with high cpu/nb clocks.

Probably prime stable around 3840 as that is about this chips max anyway. Even though ram passes memtest or wahtever your testing try running it an known previous clocks and then crank cpu/nb again and retry it...I know these sticks are 100% stable at this speed..... they pass higher but I decided to go with proven ground first...

Working on 4gig benchable, done it before with this chip...trying to see if can crack 22k 3d06 on this board as well My guess at this speed = 21,7xx

[bingo13]

I'm having terrible luck with 1002 as well. I couldn't get 3.7Ghz stable on the core with any NB or memory frequencies. The best I could do was half of an iteration of linpak. I also couldn't get DDR 1200 stable, even with 3.49Ghz on the core and a 2250mhz NB. I was running with the exact same timings, voltages, and skew settings that work perfectly on 801. This bios seems worse than 903 on my setup. I'm back on 801 and happy (and stable) as ever :-)

Bingo: What NB voltage are you running to attain 3.1ghz?

--Matt

I was using 1.425V on CPU/NB and 1.28V on NB and HT voltages for 3.1GHz. The board will not even POST with 1002 and NB set to 2.8 with my setup now. I rolled back to 0903, better NB speeds although memory speed is down from 0801. ASUS is working on it though....

[chew*]

Hmmm I'll try some prime......benching 2800+ NB doesn't seem effected by 1002 though. Fell asleep after that post. woke up to this, preety much where I expected it to be.

[Rolle2k]

If you guys haven't seen i. there is a New bios out (1002) with the following changelog:
1. Improve system stability with certain memory modules.

you can download it here - http://dlcdnet.asus.com/pub/ASUS/mb/...eluxe-1002.zip

Haven't tried it myself yet though.

[mattkosem]

Even though ram passes memtest or wahtever your testing try running it an known previous clocks and then crank cpu/nb again and retry it...

Still no go at 3.7Ghz, but I managed to get the sticks stable at 1200. It seems like they messed with the baseline settings for the clock skews. Pushing them to 300 instead of 150 as I use with 801 did the trick. Oddly, 801 is prime/linpak stable with 150 or 300. I've got a sneaking suspicion that something was done with ACC in 903 and 1002 that wasn't done in 801. I have the same clocking issues, that are corrected by +2% to the last 2 cores in 801, with both of those versions.

--Matt

[chew*]

Ahh I get it now matt....

here's that magical 4.0 22k 3d06.......no benifit with 4850's and 16x by 16x over my gigabyte with 8x by 8x.....Maybe 4850's can't bottleneck the PCI X lanes.....have to try with my 4870 slammed in there.

My 3d06 scaling calculator I made is preety damn nifty and accurate....