M3A overclocking

[muns786]

Hey instead of making a whole new thread I will ask it here:
I'm planing on building a new PC with this board (AMD 6000+, HD3850, 4GBs or RAM etc.) And i would like to know if there are any real problems with this board or is it good to go?

[justapost]

Hey instead of making a whole new thread I will ask it here:
I'm planing on building a new PC with this board (AMD 6000+, HD3850, 4GBs or RAM etc.) And i would like to know if there are any real problems with this board or is it good to go?

Some people experienced problems with the onboard network and audio devices. These problems seem to go away after a complete bios reset (by removing the cmos battery, the clear cmos jumper alone did not work).
It's a solid board with and X2 proc and it is very power efficient. I get lower temps and require less volts than on the M2A-VM.
You can expect a max ref HT ~350MHz with an 6000+, check the thread at legit i linked in #1.

[justapost]

@ justapost, still no sign on and I get any emails to the forums back as undelivered mail. So the cross forum talk continues, lol. I did mention in my email to Asus on the memory performance, together we should get things through to them. Do you know what the other memory choices are for, I have never seen them before?

Also have you watched the mosfets, they seem to be cool on my board despite not having h.s. or fans. I don't think the MSI firecracker event is likely.

Tell noee to unistall aod and try it with the BE, that is what I had to do.

Can a moderator explain why jpierce55 can not join the forum please?

@jpierce55 did you get any response from asus? I did not even get a mail saying my request arrived.

Mosfets don't get too hot here. They used the same stuff as on the M2A-VM.
With the stock cpu cooler they got pretty hot there but that board allowed up to 1,65V for the CPU. 0,1V more that on the M3A. I allways used three additinal fans (mem, mosfets, nb) on the M2A-VM. With the M3A i only use the ram fan somtimes. NB and mosfets get enough airflow from the cpu fan now.
On the M2A-VM the mosfets man some clicking noise under stress sometimes did not experience that on the M3A also.

[justapost]

@justapost No Asus did not respond to me either, not even a "we received you email" message.

Damn Asus techsupport is hard to reach. Last time i tried to contact them via theit webteam because their webform did not work.
Got this reply.

Dear Sir/Madam,

Thank you for supporting ASUS products.

We apologize that you have encountered this problem.

Please use Internet Explorer.
We are web site no suport good with Firefox
Sorry for the inconvenience!

Yuck.

About the Memory Options.

Bank Interleaving is a different way to organize memory-pages on the dimm's. If it's enabled it should improve memory performance.

Memory Hole remapping must be énabled if you want to use 4GB with 32bit wxp together with the /PAE boot option.

I don't know what the MEMCLK/Tristate Option does.

The last option is a power saving feature, if i remember correct, the phenom can turn off dct's if they are not needed.

214MHz is stable here too. Can boot with max 235MHz.
Tried 220MHz with a lower NB multi (8x) but it was not stable (that's what DaMulta recommended you).
M3A does not have a bios option for the nb multi. But the multi can be tweaked with wprcedit in the range ht-multi<=nb-multi<9.
Hope to get 235MHzx11 stable once we get vcore options in the bios.

[Thesavage]

Justapost,
Can you check your 5v rail to see what it is reporting, my rail is @ 4.9 at the moment on my previous board it was 5.3. Also I have been monitoring cpu usage through my G15 keyboard and it seems higher on the new bios in ut3 and stalker .

[justapost]

@TheSavage: The 5v rail reports 4,92V in idle and 4,87V under load.

I also noted the power usage. The new bios needs ~10W more in idle. Under load there is no difference.

Bios 301: idle/load

133W/220W

Bios 406: idle/load

144W/220W

[KTE]

Use a DMM, don't even look at software rail voltage values. There is no way some of those rail voltage are close to correct, i.e. your +5V and your +3.3V.

[Jakalwarrior]

Of interest to the K8 people. Max HT speed I could squeeze out of mine. CPU-Z reports this HT sometimes varying as much as +5 though. Not sure how stable the HT is up here.

[justapost]

Use a DMM, don't even look at software rail voltage values. There is no way some of those rail voltage are close to correct, i.e. your +5V and your +3.3V.

Hope a friend of mine with more more background in electricity will help me doing that this week.

Of interest to the K8 people. Max HT speed I could squeeze out of mine. CPU-Z reports this HT sometimes varying as much as +5 though. Not sure how stable the HT is up here.

Can you boot with 371MHz ref HT. ?

[Jakalwarrior]

I didn't get that exact with it, I went up in bumps of 10 per. 380 doesn't post, 370 was stable enough to take that screen shot, upload it, make the post, run orthos for a minute. Then I went back to priming my main OC (Currently 2.9ghz @ 1.3vcore.... which made an error after 30 mins just as I typed that lol)

[justapost]

I didn't get that exact with it, I went up in bumps of 10 per. 380 doesn't post, 370 was stable enough to take that screen shot, upload it, make the post, run orthos for a minute. Then I went back to priming my main OC (Currently 2.9ghz @ 1.3vcore.... which made an error after 30 mins just as I typed that lol)

Playing with lower cpu multis and the 9500 right now. Will go for max ref HT with the BE2400 later. Looks like the 406 bios allows higher ref HT's
I also have an X23800+ EE. Could reach up to 2.7MHz on the M2A-VM. M3A should give me ~100MHz more as it did with the 2400.

[Jakalwarrior]

See im having weird issues where 3ghz will get into windows at bellow stock voltage.... but even at 1.5 it makes errors (When it gets >50*C). Theres got to be some spot in the middle where it wont get hot enough to make an error, but its enough voltage to not just make an error on its own eventually lol. Its somewhere between 1.3 and 1.4. If I cant get this thing settled in soon though ima just buy a new HSF or lap it or something.

Seems the lower the voltage I use the less heat it makes, but also the less heat it takes to make an error. Very annoying balancing act.

[justapost]

See im having weird issues where 3ghz will get into windows at bellow stock voltage.... but even at 1.5 it makes errors (When it gets >50*C). Theres got to be some spot in the middle where it wont get hot enough to make an error, but its enough voltage to not just make an error on its own eventually lol. Its somewhere between 1.3 and 1.4. If I cant get this thing settled in soon though ima just buy a new HSF or lap it or something.

Seems the lower the voltage I use the less heat it makes, but also the less heat it takes to make an error. Very annoying balancing act.

With X2's i did it like that. Up the ref HT till it fails in orthos. The i upped all voltages one step (sometimes two steps for the cpu at once). If it's stable i try to lower single voltages till if find the minimum stable settings for the given ref HT. I increase the ref HT in steps of 2-5. It's time consuming but results in a save ref HT pathway.

I'm tempted to lap my X23800+ too, because i never did it.

BTW: I need watercooling for 3,3MHz with the BE2400.

[KTE]

If you have a DMM Achim, I can guide you through what you need to measure. Very easy, you'll see.

[justapost]

If you have a DMM Achim, I can guide you through what you need to measure. Very easy, you'll see.

Lol, thanks unfortunately my roommate forgott him at her ex's place.

[justapost]

System with ref HT 220MHZ, cpu 10,5, nb+ht 9, vdimm 2,15V 800MHz 4-4-3-5 everything else stock is prime stable for 7 hours.



CoreTemps where ~48&#176; at the end. Used the reserator 1p for cpu and 2400pro cooling and had three more fan applied.
It's unstable with a 10,5 cpu-multi at 222MHz even with an 8xnb-multi.

System with ref HT 229MHZ, cpu 10, nb+ht 9, vdimm 2,2V 800MHz 5-5-5-15, sb at 1,25V everything else stock is prime stable for ~1 hour.



System froze with stock sb voltage, did not after i increased it 0,05V worked as stock settings later. Used save ram timings.
With 230MHz the system does not boot. A lower ht-multi and lower ram settings did not help.

CoreTemps where ~45&#176; because i placed the reserator near an open window.

System with ref HT 235MHZ, cpu 9, nb+ht 9, vdimm 2,2V 800MHz 5-5-5-18, sb at 1,25V everything else stock is prime stable for ~1:45 hours.



CoreTemps where ~46&#176;.

System with ref HT 239MHZ, cpu 9, nb 9 ht 8, vdimm 2,2V 800MHz 5-5-5-18, sb at 1,25V everything else stock is prime stable for ~1 hour.



Settings beyond 235MHz did not work with 667MHz but with 800MHz for the ram.

[justapost]

These where my results on the M2A-VM:

Setup:
Motherboard: M2A-VM
Prozessor: GP-9500 @ 2200MHz - 200MHz x 11
Ram: 2x1GB DDR2800 GSkill 2xF2-6400CL4D-2GBPK
HD: Western Digital Caviar WD800 80GB PATA
Optical: none
GFX: OnBoard(X1200)
HSF: Stock AMD Cooler
PSU: Zalman ZM460B-APS
OS: Windows XP 32Bit Prof.

My resullts with 1:2 ram (800) and 5x HT multi:
It makes no difference if i switch between Desktop and Energy Saving Profile in idle usage. For load i ran prime95 for 10 minutes and build an median between max and min. Temps are CPU and Board Temps during the prime95 run.

200HTT, CPU 1,25V, NB 1,2V, RAM 1,8V
Idle: 77W AC / 54,7W DC (71&#37
Load: 151W AC / 122W DC (80,8%)
Temp: 56&#176;/38&#176;

205HTT, CPU 1,25V, NB 1,3V, RAM 1,8V
Idle: 78,5W AC / 57,5 W DC (73,2%)
Load: 154W AC / 123,5 W DC (80,2 %)
Temp: 59&#176;/39&#176;

210HTT, CPU 1,3V, NB 1,3V, RAM 1,8V
Idle: 83,5W AC / 62,6 W DC (75%)
Load: 165W AC / 134,0 W DC (81,2%)
Temp: 63&#176;/40&#176;

215HTT, CPU 1,325V, NB 1,3V, RAM 1,8V
Idle: 87,5W AC / 66,5 W DC (76%)
Load: 178W AC / 144,7 W DC (81,3%)
Temp: 67&#176;/41&#176;

220HTT, CPU 1,375V, NB 1,4V, RAM 2,0V
Idle: 93W AC / 72,5W DC (78%)
Load: 197W AC / 160,8W DC (81,6%)
Temp: 70&#176;/41&#176;

Added a few fans

220HTT, CPU 1,375V, NB 1,4V, RAM 2,0V
Idle: 109,5W AC / 86,0W DC (78,5%)
Load: 204W AC / 166,5W DC (81,6%)
Temp: 67&#176;/30&#176;

225HTT, CPU 1,4V, NB 1,5V, RAM 2,1V
Idle: 110W AC / 86,4W DC (78,5%)
Load: 218W AC / 177,9W DC (81,6%)
Temp: 68&#176;/33&#176;

230HTT, CPU 1,45V, NB 1,5V, RAM 2,1V
Idle: 123W AC / 98,3W DC (79,9%)
Load: 238W AC / 194,0W DC (81,5%)
Temp: 71&#176;/33&#176;

235HTT, CPU 1,525V, NB 1,5V, RAM 2,1V
Idle: 138W AC / 111,4W DC (80,7%)
Load: 274W AC / 222,5W DC (81,2%)
Temp: 80&#176;/34&#176;

Can boot with 238HTT but system freezes, same if i reach 239 HTT with ClockGen.

All those tests ran with the stock cooler, could not get better results with watercooling later.

With stock cpu settings i could reach 205MHz ref HT. Had to increase Vcore +0,75V to get it stable at 215MHz ref HT.

220MHz ref HT was the max I could do prime stable (12Hrs).
System was stable up to 1,45V vcore but failed with 1,475V vcore on the same settings.

Max bootable ref. HT was nearly the same. System can boot sometimes with 238MHz, sometimes it does not post.

EDIT:
Here is a review of my PSU they measured those efficiencies

75,1W 73,1%
137,6W 80,5%
262,6W 81,3%
332,2W 80,6%
471,4W 78,3%

I used a curve template for interpolation on scale paper.

[justapost]

Did anyone reach more than 2350MHz stable with the 9500 on the M3A yet? Or boot with more than 235MHz ref HT.
I'd like to know your results, because i'm curious if it is a board or a cpu issue.

[KTE]

My resullts with 1:2 ram (800) and 5x HT multi:
It makes no difference if i switch between Desktop and Energy Saving Profile in idle usage. For load i ran prime95 for 10 minutes and build an median between max and min. Temps are CPU and Board Temps during the prime95 run.

200HTT, CPU 1,25V, NB 1,2V, RAM 1,8V
Idle: 86W VAC / 77W DC
Load: 161W VAC / 151W DC
Temp: 56&#176;/38&#176;

205HTT, CPU 1,25V, NB 1,3V, RAM 1,8V
Idle: 86W VAC / 78,5W DC
Load: 164W VAC / 154W DC
Temp: 59&#176;/39&#176;

210HTT, CPU 1,3V, NB 1,3V, RAM 1,8V
Idle: 92W VAC / 83,5W DC
Load: 176W VAC / 165W DC
Temp: 63&#176;/40&#176;

215HTT, CPU 1,325V, NB 1,3V, RAM 1,8V
Idle: 96W VAC / 87,5W DC
Load: 187W VAC / 178W DC
Temp: 67&#176;/41&#176;

220HTT, CPU 1,375V, NB 1,4V, RAM 2,0V
Idle: 101W VAC / 93W DC
Load: 204W VAC / 197W DC
Temp: 70&#176;/41&#176;

Added a few fans

220HTT, CPU 1,375V, NB 1,4V, RAM 2,0V
Idle: 117,5W VAC / 109,5W DC
Load: 212W VAC / 204W DC
Temp: 67&#176;/30&#176;

225HTT, CPU 1,4V, NB 1,5V, RAM 2,1V
Idle: 118,5W VAC / 110W DC
Load: 225W VAC / 218W DC
Temp: 68&#176;/33&#176;

230HTT, CPU 1,45V, NB 1,5V, RAM 2,1V
Idle: 132W VAC / 123W DC
Load: 246W VAC / 238W DC
Temp: 71&#176;/33&#176;

235HTT, CPU 1,525V, NB 1,5V, RAM 2,1V
Idle: 146,5W VAC / 138W DC
Load: 278W VAC / 274W DC
Temp: 80&#176;/34&#176;

First thing I noted:
-Your ambient heat sink temp is high, hence your high idle/load temps.
-You don't need to add a single mV to the NB which is the chipset, you need to add for the CPU NB which is different.
-What was your HT voltage? (makes a difference)
-Your DC wattage is too high, it should be much lower. That looks like apparent/real power (AC) your stating rather than DC power you've quoted.
Even the latest & best low power optimized 81-85&#37; efficient SMPS only do <77-79% at sub 145W AC loads, like this one in a review.

For instance, the 3 Seasonics I've tried, of them the 380W is the most efficient at low power draw and it draws 145W AC at 78% efficiency (113W DC).

[justapost]

First thing I noted:
-Your ambient heat sink temp is high, hence your high idle/load temps.

Lol,i just cut and pasted those results from your thread.
It was all done with the stock heatsink.

-You don't need to add a single mV to the NB which is the chipset, you need to add for the CPU NB which is different.
-What was your HT voltage? (makes a difference)

Had no special HT voltage setting on the M2A-VM. Had to increase what was labeld NB-Voltage as it was not stable as stock.

-Your DC wattage is too high, it should be much lower. That looks like apparent/real power (AC) your stating rather than DC power you've quoted.

You are right it where the VA and W AC values.
The PSU has a pretty constant efficiency between 130-330W 80.581.5%, gone update those now.

Edit: Fixed above

[KTE]

That's much more accurate, thanks.
Yep, the SMPS units work on a typical de-rating curve. Peak performance is usually at the "typical" load which is 50&#37; of the rated DC wattage under rated temperature conditions and either side of that usually decreases output efficiency. So a curve is exactly how the PSU performance is usually observed and those figures look very accurate now.

Can you just add full system specs (hardware) in that quoted power measurement post too please? It will show everyone better the hardware and the MHz/volts connected for those power loads.

[justapost]

That's much more accurate, thanks.
Yep, the SMPS units work on a typical de-rating curve. Peak performance is usually at the "typical" load which is 50&#37; of the rated DC wattage under rated temperature conditions and either side of that usually decreases output efficiency. So a curve is exactly how the PSU performance is usually observed and those figures look very accurate now.

Can you just add full system specs (hardware) in that quoted power measurement post too please? It will show everyone better the hardware and the MHz/volts connected for those power loads.

That's true, added specs above.

[KTE]

I'm going to have to nag you a bit more.

You missed the hard drives/opticals.

[justapost]

I'm going to have to nag you a bit more.

You missed the hard drives/opticals.

NP, haven't slept the night so I'm in a patient mood.
Added HD and had no optical connected.

I already did better estimations for most of the components as i posted in your thread in the first place. must review the ram only.

[KTE]

I'm content enough.

It gives the end users reading the whole data to look at and make their own conclusions up without any guessing of if and buts.