I dont think the temp measurements are accurate. Mine is reading 35*C idling at 1161 mhz with .8 vcore.
Printable View
I dont think the temp measurements are accurate. Mine is reading 35*C idling at 1161 mhz with .8 vcore.
Use EVEREST. AOD/Speedfan are becoming very inconsistent on different boards for temperature monitoring.
I was basing my speedfan temps off of PC Probe II since I know the speedfan normal temps are low and I figured asus compensated in pc probe. They dont compare though. I can calibrate em to the same, then come back 10 mins later and there is a 10* dif.
Everst and PC Probe show the same exactly. High, even at idle with very little voltage. I doubt it would be this high running passive unless the IHS wasnt touching the core good.
How high at idle? (CPU VID/VCore and NB VID?) [use AOD and AMD Power Monitor to find out]
Which fans/heatsink do you have and what's the ambient case temp?
Turn off CnQ, then see what you get.
Idle reads between 35-45. HS is an XP90c with a delta tri-blade (running on lowish).
0.8 vcore at a lil over 1ghz.
I am not able to fiddle with those other voltages in bios. K8 just has vcore, vdimm, sb voltage, and thats it.
AOD reads no temps for core 1 or 2. APM just reads the speed wrong and .8vcore.
Try at stock speeds/volts/vid like I stated. The temps might not be any different. But I don't know your heatsink ambient temps so they could be that high, if heatsink ambient is high .
Ambient in the room is about 20*C. Case has 2 120mm fans and 1 90nm for the drives. Also a tube that feeds the heatsink external air (doesnt line up perfect, made for stock hsf, not this big ol sucker. Heatsink doesnt usually get warm. Im about to go pull it off now and see if it was making good contact, then clean it up a bit.
As for CNQ vs no CNQ, makes about 5* difference. stocks speed vs OC speed made about no difference at idle (almost the same vcore).
Edit*
AS5 had only spread about as big as a dime. Re did the TIM. Didnt really seem to help temp readings but ohwell.
Now that is cool. Chinese Asus support sent me a new Beta 501.
That Bios as NB-multi settings and allows Prozessor and NB Volates up to 1,55V.
http://www.abload.de/thumb/cpuv3oo.jpghttp://www.abload.de/thumb/nbvgtp.jpghttp://www.abload.de/thumb/nb6xwj.jpg
Was able to boot with ref HT 242MHz and 8xNB multi. I'll start playing with voltages now. :)
WOW.
Suddenly you have a better BIOS than MSI 790FX! Far better. ;)
Also be very careful with that high NB volts and look at your power draw at bootup with those volts.
nice features.. hope you manage to clock ref clock higher.. afaik phenom ref clock upper limit arround 260+mhz. 270mhz should be enough to blow phenom ref clock record.
Tried to find the min volates for 214MHzx11 stable first. 1,25/1,2375.Quote:
Originally Posted by KTE
Upped to 1,2625/1,25, tried it with 216MHz and it failed.
Checked AMD PM, CPUT and AOD and it looks like those settings are still CPU/NB VID. As those are upward locked in my 9500 bios/cpu does this:
1,2625 -> 1,1625
1,3 -> 1,2
1,4 -> 1,2
I'd need an BE i think.
Bios CPUV 1,3V
http://www.abload.de/thumb/c13000n12375tt0.jpg
Bios CPUV 1,2625V
http://www.abload.de/thumb/c12625n1235-1txr.jpg
I'll go for max ref HT now. :rolleyes:Quote:
Originally Posted by tictac
Justapost,
Has it got the TLB disable switch?, also have they resolved the lower memory performance? Hope to see it on the ftp soon.
Have not found and TLB switch. Looking for max ref HT atm, seems to be 254MHz.Quote:
Originally Posted by Thesavage
259MHz max ref HT with ram 800MHz
http://www.abload.de/img/cpuz25930d.jpg
254MHz max ref HT with ram 667MHz
http://www.abload.de/img/cpuz254dch.jpg
Did not know my ram can go so far with an 1:2 divider. :D
What can you get maxed with unganged? :D
My 9500 did 270HT ref. max unganged mode at low multi. This new one I can't try due to problems booting if I reset.
270 max ref HT with 7 nb/ht multi
http://www.abload.de/img/cpuz270a0t.jpg
Not really stable but got it verified on the second attempt. :hitself:
wow.. you did it.. congrats.. !
Whats your max stable oc with the new bios? also is it possible to post the bios on the forum, would like to help test.
Bios still lacks a switch for (un)ganged. CAS 6 is the key here.Quote:
Originally Posted by KTE
It's gettin even better, nice bios. :rolleyes:Quote:
Originally Posted by tictac
275MHz max ref HT with ram @533 or @400
http://www.abload.de/img/cpuz275-2qcv.jpg
Should be the same as with 406 due to locked voltages.Quote:
Originally Posted by Thesavage
Not sure if I can post it here, can't read chinese
PM me an email and i'll send it to you.Quote:
achim, ??!
?
?!
Justapost,
Thanks just updated, done a few quick bench's seems about the same speed as the 0406. Very much easiers to oc though can now cold start with my oc settings and also use larger HT ref increments. One strange thing though is the lan disables after 210 ht. Aod does not like increasing the nb multiplier though, shows differently from cpuz.
LOL! That looks much better now.
How many cores can do that Achim? Which core is the weakest? Try them individually and max on each one. Have you managed to get those CPUZ verified?
Just keep in mind AOD is the best for verifying your ocs here, unless CPUZ verifies the clocks, which it won't do if they be bugged.
Looks like you might have a very high ocing 9500 there but you underestimated it, looking at your last ss, you've probably kept too low a tRFC at DDR2-800 which effects your HT ref. fully. Maybe you just need a better board and cooling because HT ref. is the limit with these usually. You should be able to get pretty damn close to 270HTx10 now, my 9500 was max 270HT ref. and it could do 265x10 pretty easy stable. Try >260x11, you should get close if not more. ;)
Single channel should get you higher oc on CPU/NB speed, usually ~2/300MHz higher. Mine did 150MHz higher at 11x multi this way. Seems the IMC really is the setback here.
Strange I don't have nic problems here running at ref HT 262MHz atm. I disabled the com port and the audio device however.Quote:
Originally Posted by Thesavage
All my CPU-Z screenies are verified. With the previous bios I had to repeat it a few times to verify, the new one has no problem with that.Quote:
Originally Posted by KTE
tRC you mean? I have only the four basic settings for ram in the bios.Quote:
Originally Posted by KTE
As for the ref HT. the NB speed was the wall here. Tried SuperPi with 265x9,5 and the system froze.
Priming with 262MHz cpux9 nbx7 htx7 and ram at 5-5-5-15-21-2T@800MHz right now. It failed after few minutes at 264MHz so at stock volts the limit is the cpu frequencie ~2350MHz right now. Will try single channel later.Quote:
Originally Posted by KTE
Using 2,25V for the ram most of the day and the phenom is still alive.
Nope, tRFC. It gets set differently at higher and lower RAM speeds and decides your RAM/IMC/HT ref. stability. Keeping a high tRFC gets you best HT ref. maximum speeds, I've mentioned it since early December just because of this reason because users will get a low HT ref. and say "that's it" whilst it was their settings which were holding back the HT ref. mainly.
My old CPU-Z files which I hadn't verified till yesterday (v1.42) don't verify anymore. :(
Ganged mode is single channel.
Bios allows tCL,tRTC,tRP and tRAS. I can find tRC (Row Cycle time) in Memset and AOD but no tRFC. :shakes:Quote:
Originally Posted by KTE
Damn, what did you get (unverified)?Quote:
Originally Posted by KTE
Hmm, CPU-Z reports Dual Channel Ganged mode with two Dimm's. Reports Single channel no mode with one Dimm.Quote:
Originally Posted by KTE
-Hmm.. maybe those settings don't appear in tools because your BIOS doesn't support it, but they can make a big difference (to perf as well).
This is tRFC in Memset->
http://img87.imageshack.us/img87/6312/trfctz8.png
Can you see that? Set it to 195ns when you oc HT ref/RAM/IMC.
-I disabled all cores but core 2 (3rd one) and it reached 2.8GHz before it crashed, on just multiplier. Never took any ss because I thought I'd see them in the valids anyway but they're too old to verify. I would try again but if I have to reset BIOS again it won't boot. So waiting for the new BE.
-Yep, the updated CPUZ shows exactly as you're saying. What I meant is, ganged DCT mode is synonymous to single channel and optimized for single threaded performance; in performance, IMC stress, RAM oc'ing and so on, it is like having a single DIMM installed. Unganged is synonymous to best performance typical of dual channel, harder to oc RAM in, much harder on the IMC and also restricts IMC speed oc'ing and HT ref. :)
I'm still thinking you can get more for some reason. ;)
Have you tried these settings? One of them might work for you.
http://valid.x86-secret.com/show_oc.php?id=279946
http://valid.x86-secret.com/show_oc.php?id=279944
The best setting all round for me was this (stable/benched/ran for 3 weeks):
http://valid.x86-secret.com/show_oc.php?id=278357
Lol, yeah if have those settings in MemSet here too. Will try if they survive an reboot with bios modifications.Quote:
Originally Posted by KTE
Now 2,8MHz with one core whould be pretty nice. With disabling you mean 5xmultis with aod I assume.Quote:
Originally Posted by KTE
Well I know the difference between the two modes but interchanged em.Quote:
Originally Posted by KTE
Unganged means both dct's operate independent.
Must do day-to-day business here now to fund my latest hardware orders. :DQuote:
Originally Posted by KTE
Thank you for the settings gonna try em later.
Settings do not survive a reboot.Quote:
Originally Posted by justapost
None of my cores is stable ~2,6MHz.Quote:
Originally Posted by justapost
-Those settings won't survive a reboot, yep. We set them using the BIOS options we have for them.
-The way to get only one core is simple>
1.
http://img228.imageshack.us/img228/8881/1corete0.png
2.
http://img111.imageshack.us/img111/9624/1coreka2.th.png
:D
We do this (platform independent) using the BIOS options we have for them. ;)Quote:
Originally Posted by KTE
http://www.abload.de/thumb/img_6234da3.jpg
The boot option does not let you choose the core you want to disable I guess.
Nope it won't let you choose specifically, but it works in numerical order: core 0, 1, 2, 3.
We don't have the option for disabling cores as such through the BIOS. :(
Got an english reply from Asus Support. They sent me the bios for testing due to my feature request so it's ok if I upload it somewhere and post a link here.
Download Link for Bios 0501.
Measured the mobo's power consumption yesterday:
System:
MB: M3A
Prozessor: GP-9500/BE-2400
Mem: 2x1GB G.SKill F2-6400CL4D-2GBHK
GFX: HD 2400 Pro (500/400MHz)
HD: Maxtor 7H500F0 (500GB/Sata II)
PSU: Zalman ZM460B-APS
OS: Windows XP Prof. 32Bit
No fans no opticals used. Cooling was done with Reserator 1p whom was not taken into account and pulls 4-5W AC.
CnQ and Desktop where idle measurements.
The load was generated with prime95 and Fritzchess running in parallel.
I used stock/auto settings in the bios. For 9100 simulation i use an 9xcpu multi together with an 8xnb/ht multi and 1,2V cpu-v 1,175V nb-v.
Ambient temps ~20°, CPU temps idle 36° under load 45°-47°, nb ~30° sb 25°.
Quadcore GP-9500:
CnQ: 95W AC / 75,6W DC (79,6%)
Desktop: 125W AC / 99,9W DC (79,9%)
Load: 173W AC / 140,6W DC (81,3%)
Triplecore GP-9500:
CnQ: 93W AC / 72,7W DC (78,2%)
Desktop: 122W AC / 97,4W DC (79,8%)
Load: 150W AC / 121,2W DC (80,8%)
Dualcore GP-9500:
CnQ: 91W AC / 70,0W DC (77,0 %)
Desktop: 111W AC / 87,7W DC (79,0%)
Load: 133W AC / 106,7W DC (80,3%)
Singlecore GP-9500
CnQ: 88W AC / 67,4W DC (76,6%)
Desktop: 98W AC / 76,5W DC (78,1%)
Load: 114W AC / 90,4W DC (79,3%)
Quadcore: GP-9100 (simulated)
CnQ: 93W AC / 72,7W DC (78,2%)
Desktop: 114W AC / 90,4W DC (79,3%)
Load: 158W AC / 128,3W DC (81,2%)
Dualcore: BE-2400
CnQ: 68W AC / 47,6W DC (~70%)
Desktop: 80W AC / 60W DC (75,0%)
Load: 100W AC / 78,3W DC (78,3%)
Difference between load usage with two vs. four cores is ~34W so a dual core whould have an ~60W TDP based on these numbers.
Difference between dual core k8 and k10 is 28,5W. Based on these the dual core K10 TDP is ~73,5W.
The accouracy for the psu's efficiency is low for 68W AC, so these measurements can not be used to estimate phenoms idle power consumption.
thanx for the bios, so far so good
Glad it helped.Quote:
Originally Posted by dball3
Pushed max ref HT with the BE-2400 a little bit.
http://www.abload.de/img/cpuz392kk0.jpg
392MHz can boot everytime, 393MHz never but I did not try all posiibilities yet.
With an 1x cpu and ht multi I can reach 282MHz max.
Max the cpu can handle are 1,55V. I can reach 3,33GHz+7MHz with all multis betweem 11,5-8,5, no way to get any further. Prime95 errors with 1,575V.
Did it help you get higher stably? or just let the HT go higher and give us more voltage?
Testing stability with 392MHz ref HT now. With higher cpu multis I got it stable for at least 30 mins with prime95.Quote:
Originally Posted by Jakalwarrior
It's no 24/7 setup placed the reserator 1p'2 reservoir outside the window to gain 10°.
EDIT: Seem's the board was just in a good mood max I can boot now is 382MHz.
It's stable at 380MHz with 8,5xCPU at 1,525V, Mem with 461MHz 4-4-4-12 at 2,2V and 2xHT.
I just gave 0501 a spin and my system didnt like it. CPU temps reported very high (50+ at idle) and in orthos I get errors very quickly with the overclock I was running previously.
Guys,what's the best and most stable BIOS for this board?I will have one by next week for testing :).
Thanks in advance
301 gives the best memory bandwith but the phenom support is on an initial state. It lacks cpu-multi settings.Quote:
Originally Posted by informal
406 Allows cpu-multi settings, max ref HT i could reach was ~235MHz
501 Allows nb-multi settings and allows higher ref HT's with lower nb-multis
So id go with version 501 if you use a phenom cpu.
Thanks a bunch justapost :).
I will use it with a Brisbane CPU,if it performs well i will keep it for my main rig :).
So with X2 still a good choice a 501 version?I don't plan on using Phenom anytime soon(i have no need for quad core cpu for the time being)
Thanks for the valuable informations in this topic :up:
I've tried manually to set NUMBER_OF_PROCESSORS value from advanced system settings\environment variables, but it didn't work after reseting (Vista 64bit) :(Quote:
Originally Posted by KTE
Where can I DL this util of yours?
On WinXP there was msconfig.exe for that issue. Not in the PATH locatet in C:\WINDOWS\pchealth\helpctr\binaries.Quote:
Originally Posted by Nedjo
If vista still uses C:\boot.ini you can also use a text editor and add /NUMPROC= manual.
This is the crap thing that throws me off! What the heck is happening with one high offs and then low and then higher later one and so on? :mad:Quote:
Originally Posted by justapost
Type msconfig in the run box in start menu, and under the boot tab, click advanced options.Quote:
Originally Posted by Nedjo
My brisbane much prefers 0406. 501 lowered my max stable overclock. Dont be scared to try it and see what it does for you though.Quote:
Originally Posted by informal
One thing Achim: for 9100E 1.8GHz, voltage/vid and so amperage won't be the same as 9500 so power consumption can't be simulated unless you have the unlocked 9600. In that case you can throw voltage to sub-1.1V (my 95W 9600 BE is 1.192V at 1.2VID stock).
So your volts/amps at 1.8GHz would have to be maximum 1.085V at 15A per core to simulate those 65W models and NB speed slower than CPU speed by at least 200MHz and thus low IMC voltage and VID. I should be able to test something similar soon but still not as low.
Tried low voltage @2.2 lowest I could get was about the 1.1 before the machine shut down, temps did drop nicely though nearly 8c.
Good point, i'll try to find the minimum voltages required for 5xmulti and add those results. Had used a 7xnb/ht multi together with an 8xcpu multi, can be the 9100 comes with an 6xnb/ht multi. You get an lower clocked ES from your uncle?Quote:
Originally Posted by KTE
Do one thing though; you might crash your system if your VID is too high for the low CPU speed you selected. So if you drop the VID to 1.1 at bootup and set low CPU speed, that would be better.
Nope, the only ES there were 9700 and 9900 and those I saw long ago. X2 sorta clockers, better than all the retail products I've seen.
EDIT: you can try PowerTOP to check which process uses most power at idle in your testing with Phenom if you have a distro using a kernel after 2.6.21. It'd be interesting although it won't give you any power measurements.
Thanks for the advice man,i will try the 0406 first.What is your preferred method of update on this board?Online while in WinXP or do it old fashion style-in DOS mode?Quote:
Originally Posted by Jakalwarrior
BTW,board arrived early ;).Got the sucker up and running ,no need for WInXP reinstall(for now).It reports some DLL on start was reallocated in memory but will try to google for solution.Previous board was nforce based one,just uninstalled its drivers and booted with AMD770 ones in Safe Mode.Will see how she goes now.ATM i am at 2.6Ghz while typing this(chip was capable of 2.7-2.72Ghz on previous board)
Thanks
Tried 0406 today, did not make a difference with the be2400 here. Max ref HT I could reach was 280MHz.Quote:
Originally Posted by Jakalwarrior
With the phenom I managed to capture this one at 1,25V cpu/nb with 501 bios. Could rund Spi1M once, system froze with 10x multi via AOD.
http://www.abload.de/img/ref276x9cop.jpg
With 277MHz the windows start freezes with 278MHz the disk makes some clicking but no keyboard lights blinking.
I keept the temperaturs low with environmental friendly cooling (2L-reservoir in 0° ambient outside the window). Max CPU temp was ~45° coretemps ~35°.
Ran prime95 for half an hour with 2375MHz(250MHzx9,5). With 255MHzx9,5 it failed after two minutes. Think I gain max 50MHz with 20° lower temps.
@informal: you can update the bios with ez-flash from the bios. ez-flash can read fat32/ext2 and iso images from hd's cd's usb-sticks.
Thanks justapost,i did it already through the WinXP update utility.It took some time(seemed like forever lol),but now i am on ver 406.I have one problem with Realtek Sound subsystem,refuses to load the drivers(tried new and old ones).WinXP had nforce board with older Realtek sound card integrated but i cleaned it beforehand.Now i might try in safe mode since i lost sound :(.
Didn't try the serious OC yet.I did notice the absence of very important tRC subtiming change in BIOS :confused: .
Thanks for your help,both of you,it's very appreciated.
Have you tried this?Quote:
Originally Posted by informal
M2A-VM had only a divider setting and no even the 2T field, but settings are still moderate.Quote:
Originally Posted by informal
Thanks for the link.Not quite like my problem ,but kinda similar.Quote:
Originally Posted by justapost
I had that famous Audio Device on High Definition Audio Bus yellow exclamation problem.It happens on some intel based mobos with this Realtek chip too,and who else but Microsoft is to be blamed(they screwed up with some update before).
I resolved it(excl. mark) with the KB888xxx(or smtng like that) update but i am DL-ing the latest Realtek drivers to try to make the sound work again(before it went thru the 99% of install process and fails recognizing the Realtek chip).Now i will try the latest(slowish download) and see if i can make em work now that i resolved the yellow exclamation mark in Device Manager.
Also one more question for M3A owners:is it usual for the heatsink on the x790 chip to get VERY hot even while idling?I am a lil bit worried about it.
UPDATE:Solved the stupid and VERY annoying problem with audio device !
Here is what i did thanks to the ppl in that majorgeeks' forum topic.Without their helpful tips i wouldn't have done it.Gosh,it's 4:33 :S:,gotta go sleep some.
So if anyone has the Audio Device Problem,please refer him/her to this post of mine and to the majorgeeks thread.
Another thing,i solved the crackling sound coming from the speakers while mouse is moving(yeah that was happening too...).
Solution: Double click the Speaker Icon is taskbar and choose Advanced View.Then scroll to top right of the list of controls and MUTE the stupid PC Beep control.It's this lil bugger that creates some interference with the onboard chip or its driver.
So the problem can be caused eigther by the hotfix or by the realtekdriver. In both cases the device must be disabled and uninstalled first.Quote:
Originally Posted by informal
In case of a failed hotfix the uninstall registry key must also be removed.
Then eighter the hotfix or the realtek driver works. In M3A case it's the latest realtek driver. Checked it here had the Realtek and not the Microsoft hda driver installed.
Never had this problem because i had this hotfix integrated in my install cd with hfslip and also had the latest driver pack sound package integrated as well. But i remeber some time ago that 888111 hotfix made trouble getting interated in the install medium.
I'll add a note and a link to majorgeeks to post #1.
The nb stays cool if the heatsink get's a little airflow. With just watercooling it's so hot i can't even touch but the temps speedfan/pcprobe reports are ~30°. The Stock cooler produces nearly no airflow in the direction of the nb so for overclocking better cpu cooling or an extra fan is needed.
Thanks.
Udpate: passed 1 hour AOD stability test with 286x9.5,mem @ddr2 900.
So far so good.Sound is working great since i solved the issue with UAA.
Going to try some more HTT to see how far will she push it :D.
Cheers!
sorry,forgot the picture:
http://aycu22.webshots.com/image/416...6483481_th.jpg
isn't that a pain in the ass, even on vista it happens, the last four boards i got have had that, first thing i do is boot up ,run firefox and scroll up and down to see if i get the feedback out of the speakers. I thought it was just me.Quote:
Originally Posted by informal
Yeah,a true PITA!Quote:
Originally Posted by dball3
From now on,try the "solution" i used with my M3A mobo and see if it works out for you.The PC BEEP function is not needed anyway,so mute it away.
Cheers
PS My board needs 2.25V for my ram in order to make it work @ 900Mhz cas4.Damn you geil :)
Justapost - Kudos on a nice thread.
Tried two Windsors and a Brisbane on the mobo.
Reached 3GHz with the 3800+ EE F3 Stepping-2 at 1,525V
http://www.abload.de/img/cpuz300x10mi0.jpg
FB Stepping-2 only reached 2,7GHz at 1,5V
http://www.abload.de/img/cpuz305x981l.jpg
3600+ Brisbane FB Stepping-1 reached 3,09MHz at 1,525V
http://www.abload.de/img/cpuz330x95982.jpg
For comparison the BE-2400
http://www.abload.de/img/cpuz392kk0.jpg
Here is a diagramm with the voltages required for different frequencies
http://www.abload.de/img/powerly8.jpg
Very nice. What could they each hit stable or 3D benchable Achim? That's mainly what we'd be interested in TBH. ;)
Also those volts 1.55-1.57V are very high and not runnable in a system past suicide shots. Hence why I wouldn't feed high volts to my 9600 BE chip, its pointless doing so for me since the temps/power/stability was so bad. I mean, what would 1.45V give you on those CPUs and what temps would you get as well as power draw?
Was looking for volts/frequencies today, so I just ran prime95 for 5-15 minutes.Quote:
Originally Posted by KTE
Yep 1,55 is so far the max the chips can handle same settings with higher voltages do not work.Quote:
Originally Posted by KTE
Also I had the watercooling reservoir placed outside the window. That way the temps where ~60° with 1,55V.
I'm amazed about the difference between the two windsors, must check if they where both EE's. Will add more specs and info's tomorrow, I noted the power draw and temps.
OK.Quote:
Originally Posted by justapost
So they were all water runs and not air?Quote:
Yep 1,55 is so far the max the chips can handle same settings with higher voltages do not work.
Also I had the watercooling reservoir placed outside the window. That way the temps where ~60° with 1,55V.
I'm very interested in the stock-> overclocked power and temps. idle/load if possible. The BE and the EEs usually sky rocket in power when oc'd.Quote:
I'm amazed about the difference between the two windsors, must check if they where both EE's. Will add more specs and info's tomorrow, I noted the power draw and temps.
EDIT: chips limits aren't decided by the oc amount of headroom in business but by the TDP/voltage/frequency/scaling. Naturally it is paramount to keep at decent TDPs for the chips whilst still scaling decently with added voltage and frequency with respect to performance, operational temperatures and power. Some CPUs do have a lot of headroom within acceptable TDP/power/temps, sub 150W, and those are like the C2D and C2Q range (not including the early QX). Once you oc them from stock, they still don't break the power/temps barrier that much compared to most other CPUs until around 3.3GHz (G0), meaning they have headroom left. Whereas the QX didn't have much headroom to begin with, it was too hot with high power demand right from the beginning being the highest bin SKUs, so they couldn't improve on 3GHz for 14 months until Penryn. In the same way, to measure power/temps influences with added frequency/volts will show you the headroom of the IC process, design, material and technology. If higher decent bins were not possible, the chip wil jump very quickly from being a low power/temps chip to a high power/temps chip. ;)
Like the C0 QX9770/QX9775 exactly like the top bin B3 QX6700 were when released Nov '06. But in stark contrast to the Q6600 G0 which will go to 3.5GHz stable very easily at lower volts/temps/power than they do.
Test system:
MB: M3A
GFX: HD2400pro
MEM: 2x1GB GSkill F2-6400CL4D-2GBHK
HD: IBM Deskstar 20GB IDE
FANs: 1x12" Fan
PSU: ZM460B-APS
Cooling: Zalman Reserator 1p (reservoir outside ~5°).
Prozessors:
3800-1 (red):
X2-3800+ EE (65W)
F3-2 F43 JH-F2
ADO3800IAA5CS CCBYF 0633MPMW
3800-2 (yellow):
X2-3800+ (89W)
FB-2 F4B BH-F2
ADO38001AA5CU LEDBF 0647GAL
3600 (green):
X2-3600+ (65W)
FB-1 F6B BH-G1
ADO3600IAA5DD CAALG 0710SPMW
BE-2400 (blue):
BE-2400 (45W)
TODO
I used 1,2V as a minimum and did not look for lower stable values.
For stability testing I ran Prime95 for 5-15minutes, so expecialy with higher voltages those are inaccurate.
For example the 3600+ ran stable with 1,45V and 320MHzx9,5 for 5-10 minutes. For longer prime95 stability I need 1,5V (currently running ~1,5hrs).
I used BE-2400 results from a prior measurement session in the f/v diagramm, gonna run the measurement in the current testing system today.
Frequencies / Voltages
http://www.abload.de/img/f_vl29.jpg
Frequencies vs. Temperatures
http://www.abload.de/img/f_cdie.jpg
Frequencies / Power AC Idle
http://www.abload.de/img/f_pifsg.jpg
Frequencies / Power AC Load
http://www.abload.de/img/f_plw1h.jpg
Voltages / Power AC Load
http://www.abload.de/img/v_plbnd.jpg
Voltages / Temperature
http://www.abload.de/img/v_cnfn.jpg
During the testing i found that the bios uses a 5x multi if I choose one below 9x with the windsors.
Tried to find the max ref HT with the two windsors. 3800-1 reached ~310MHz, 3800-2 reached ~350MHz. Gonna find the 3600+'s max ref HT now and test the BE-2400 afterwards.
EDIT: 3600+ reached ~360MHz.
Running the BE-2400 measurement now with an 11x multi and looking for lowest possible voltages in the below 1,2V region. Preparing for an GP-9500 comparison in the 2200-2400MHz regions.
Great, thank you for the results, what cooling and cpuv did you use?Quote:
Originally Posted by informal
Thanks, positive feedback is alway welcome. :up:Quote:
Originally Posted by JumpingJack
Oy, I just found the 0501 link. THank you. I'd like to give these to G2s a run on air.
What do you guys use to measure power consumption?
Thank you Achim.
I'd like to respond to a few things yet after running some caclulations but I'm a little busy for a few hours at least. :)
Hi justapost,i used HyperTX and 1.45V for vcore.The g1 i have is not some particular OCer,have achieved 2.75Ghz max stable until now(the wall for the chip is somewhere around this freq.)
Board runs just fine,the only thing is the buzzing noise coming from around the cpu socket(high freq. one,probably from coils).But it can be barely heard if at all.
NP, second I use a normal power meter build for domestic appliance measurement.Quote:
Originally Posted by noee
http://www.abload.de/thumb/img_6206e5r.jpg
Thanks, gonna try mine on air later as outdoor cooling seems to improve stability alot, even at low core temps.Quote:
Originally Posted by informal
With my X2 6000+ i can reach with the Bios 0501 3420Mhz @1.50 Primestable Custom FFT.
I tested also with the DFI Lanparty 590 Board,I can't retry the results....!!
The Asus M3A is a very nice Board!
But the Beta Bios for the Phenom is not so good,NB Multi is now in the Bios,but I'can't increase the Vcore! ( It still at default Vcore!)
I finally received an account activated email, about time! :up: Justapost I had sent you a pm on tpu did you get it? Asus gave me the email address for the engineering department. I emailed them about the voltage, ganged mode, and a TLB fix switch.Quote:
Originally Posted by justapost
Their support team ain't as bad as I thought. I found that with my BE2400 the max voltage I can apply to the cpu is 1,55V. If I apply 1,575V via bios it is 1,475V.Quote:
Originally Posted by jpierce555
The M2A-VM allowed voltages up to 1,55V but also had an +100mV field, so the max appliable was 1,65V.
I have only 0,02v overvolting under 100% processor load!
EDIT: The Everest Memory Benchmarks are very,very good! ;-)
I think this boad is just a good clocker (for X2s). Settings that were unstable on my MSI are stable on this board IF the chip is very cool. Overall this board seems to be able to take chips to their ragged edge without pushing them over due to board related screw ups so when you hit your max it is definetly the chip and not the boards fault. One of the best boards ive ever owned as far as that goes.
Just my inductive guess. I could be wrong, but that is what it feels like to me.
I agree 100% :).Quote:
Originally Posted by Jakalwarrior
On X2's this is true, it is coming out to be a great board. The Phenom needs more volts and it should be an o.k. board on it. I can hit 2.48 100% stable with the Phenom now. I can also now get my ram to 920mhz(all it can take). The Phenom will go a little higher but at the cost of to much ram performance to be worth it.Quote:
Originally Posted by Jakalwarrior
Justapost, I found that up to 1.35v on the nb helped, and 1.4 the system would not boot. For initial playing a nb multiplier of 6 worked and nothing lower, going lower required using the cmos jumper. I have not tried lower than 6 after changing other settings. I also see the system was stable down the an ht frequency of 600, and ~at 200!! That surprised me, the board shows promise on the quad if the bugs get worked the rest of the way out.
So the nb voltage is not locked upwards? I'm currently trying to find the minimum voltags for given frequencies with different cooling systems.Quote:
Originally Posted by jpierce555
It's a little time consuming took me two days with the be2400.
Could you apply more than 1,55V to an k8 cpu? Seems to be another lock here.
Can be there is another one at 0,850V. At 1,8MHz my BE24000 required only 0,875V to be stable everything below resulted in an CPU-Overvoltag BIOS message.
EDIT: Tried 1,265V via bios for the nb but AMD power monitor shows 1,165V. Seems I can run triple core with 222 ref HT (2442MHz) stable and ~230 ref HT (2530MHz) as single core with 1,25V cpu/nb.
Seen from the hardware side I agree. But the bios lacks the +100mV feature the M2A-VM had. With good cooling I could run an 3800+ stable at 1,6V that is not possible with the M3A as the max I can apply is 1,55V.Quote:
Originally Posted by Jakalwarrior
Does the 1.5v on the x2 work with you? Another thing I noticed was the 1.5v setting dropped voltage back down, but at 1.525/1.55 it did work. I never tried higher.
As far as actual nb voltages mine don't show on anything but the bios. I never thought about AMD power monitor.
It does show my voltage at 1.237 on the nb, but it only shows 2.2 on the cpu speed so I don't know if it is accurate.
Strange it worked flawless up to 1,55V beyond it's locked.Quote:
Originally Posted by jpierce555
Yep, AMD PWM does not read the current ref HT from the clockchip and assumes 200MHz.Quote:
Originally Posted by jpierce555
a new Beta 0601 is out!:up:
But still the Vcore Bug with the Phenom....:shakes:
Thank's found it on their ftp sites. So still no vcore beyond 1,25V hope they will fix this asap, starts to annoy me, that this board can not oc a phenom as far as it's ancestor.Quote:
Originally Posted by planet456235
Justapost ,can you please post the link for the ftp site? :)
Anyhow,board is running great and stable for now,all with 0406 bios :).
The link is on the first page of this post :)
Thanks mate and bad me => :slapass: :D
Here is direct link to version 601, to make reading flawless. :)Quote:
Originally Posted by informal
That 1,55V upward lock does not annoy me but 1,25V max for a phenom is something they have to fix.
Can someone tell me where I can find the superio chip on that board? I have an osciloscope here for a week and want to measure that chip.
Thanks for the link man :).
As for the location of SuperIO chip,it's right next to PCI3 slot(near the edge of the board,normal/90 degrees oriented to PCI3 slot).Look at page 2-3 of manual for M3A.
Gotta slap my a$$ now. :slapass: cuz I did not look into the manual.Quote:
Originally Posted by informal
Thanks located the chips. Damn those pins are tiny.
has anyone here used the onboard raid controller on the m3a yet? im sellin my current m3a rig and am contemplating using the m3a for my next build again, but im gonna be going raid 0 with two seagate 320's. just curious if anyone has had any experience with the onboard raid so far.
Found the layout of a pin compatible super-io chip.
http://www.abload.de/img/superior6e.jpg
Specs can be found on the ITE homepage here.
I'll stop measuring now, I already caused a short circuit between some of the VIN pin's. Pins are a little darker now but reading still works. Don't wanna challenge fortune.
@crazydiamond: Have not tried raid0 here but no one is complaining about it on the asus form.
i've got two wd 320's running in raid o and i have had no problems at all, very stable. But i haven't done any overclocking since i set up the raid so i can't give any advice on how it might react to higher clocks or stress.Quote:
Originally Posted by crazydiamond
Any benefits from the new bios, if not I will leave it.
thanks guys, unless an nvidia am2+ board magically hits the shelves in the next few days i think im gonna go back to the m3a. so far it has been very stable with my x2 5000 be.
the only things i would like to see changed as far as X2's go is: more memory timings/voltages in bios, nb volts in bios, better fan control in bios. other than that this board is a very nice budget x2 board IMO. hopefully asus can fix these minor things in updates, but i think they will have thier hands full for awhile with the phenom bios'
That new Biostar board whould interest me.
Tried the 601 bios. Can not find a difference between 501 and 601. Still no unganged and tlb-fix field. As planet said before the cpu and nb voltage is locked upward at 1,25V. Have not tried the multi issue i had with the windsors.
Read the VINs pin 96-92, one of them will be the VDIMM. ;)
Yep, be careful as they are tiny and small so don't short them out.
I don't want to touch those pins the space between is to slim for a peace of paper to fit between.Quote:
Originally Posted by KTE
I measured a few resistances around those pin's and found one with an amplitude of ,2V. That should be vcore i think, or are there other 1,V strains at the VIN's?
If it's vcore it's allways 1,2V, tried it at 1800MHz with 1,075V applied and at 2350MHz with 1,25V.
Gonna check the chips specs now.
EDIT: Hmm I measured the 12V VIN, did not take the 10xmulti of my probe into account.
Achim, check the Vcc pins and ones right near it, one of them will be VCore.
There are many 0.8-1.5V pins there usually, you have to test 'em to find out which is which by fluctuating Vcore. Don't keep probe on them for long bursts.
The PSU voltage readings will be the most inaccurate through the HWM IC BTW.
Did you try ASUS AI Gear for your board? http://dlsvr02.asus.com/pub/ASUS/mis...ar2_V10018.zip
Which ITE chip do you have, version I, J or K?
Have you tried using Smart Guardian to get all temp/volts feedback? It should work perfectly for your board since that's the HWM IC you have: Attachment 71289
Yep VIN0 and VIN1 are the recommended pin's for vcore, i expected absolute values here as those volatges are in the 0.8-2.0V range the chip can handle.Quote:
Originally Posted by KTE
I assume I can measure those on the 24pin power plug on the mobo much easier.Quote:
Originally Posted by KTE
That power consumption optimisator you mean, yes I tried it at the first day i had the mobo and it froze my machine, but that was with the first bios version.Quote:
Originally Posted by KTE
It's an IT8716F-S 0737-DXS whom is only pin compatible with the IT8712.Quote:
Originally Posted by KTE
Just gave it a try does not read correct with the 8716. :shrug:Quote:
Originally Posted by KTE