Printable View
If you manually set the speed OR the voltage for the memory... so they are on any setting OTHER than "AUTO" then the EPP settings/voltage will not be available.Quote:
Originally Posted by Phenomenalspeed
OR you don't have 1066 ram so it doesn't have an EPP setting.
G0ldbr1ck could I get your settings details for the Dominators on your DS5 board? Any settings that are different from optimized defaults would be good enough I think. Like for one are you using the Auto Xpress > EPP Mode or just setting the memory speed to 1066 and leaving auto timings? My dominators seem to like around 1.95v also how about yours?
Thanks
I have Auto Xpress > EPP mode enabled and EPP voltage enabled (For this to be available you must have all other ram settings to auto then reboot back to bios then will be available to enable). Changing anything manualy just dose not work. What that gives me is:Quote:
Originally Posted by Fulgore0rb
unganged (have tryed and benched both ways and not realy any diff. in performance but in ganged would get errors once in a wile)
DDR 1066
5
5
5
15
22
2T
@2.1v (There have been allot of revisions on the dominator ram and they have changed voltages, one of my sets is 2.2v and the other is 2.1 but both seem fine at 2.0 but I have read that the newer revisions use a lower voltage yet so not sure what yours is able to do.)
Sorry for the late reply, been out of town all weakend. Hope this info helps but feel free to ask if you need more.
Thanks G0ldbr1ck. I am narrowing the problem down it seems. I currently seem to be stuck at the increased DIMM voltage delimma of causing problems. I cannot get up passed 1.95v without many issues. I know some other voltage increases should resolve this problem. Goldbr1ck have you increased any other voltages and to what value? Northbridge and CPU voltages seem to be the most commonly increased to help this issue- and sometimes HT Recieve/Transmit voltages. FYI my Dominator version is rated at 2.1v.
I fear I may need to change my BIOS revision to get the NB overvoltage to work correctly. (still using F5- since the F6 seems to have a problem with phenom 9600's)
Im a bit confused to your problems, Things are simpler for my setup I guess. Maybe a diff between your B2 phenom and my B3 not sure, someone else might be able to answer that.
The bios is just so buggy that all my overclocking is done by software. My ram when set by EPP is 2.1v, Is your ram not setting itself to that voltage? I have no other voltage increases on anything other than a very small increase on the cpu vcore. As far as CPU overclock, your much better off using just the multiplier and no more than 210 on the base HTbus as its just to unstable to go higher.
If you can clarify your 1.95v problem a little better maybe I can help you better, but im just not sure why its that low to begin with.
@Phenomenalspeed: Have you got everything worked out with you DQ6? If your still having problems give us an update and Ill try to help you.
Quote:
Originally Posted by keithlm
Thanks. My RAM is 800Mhz and it does have an EPP profile. When I changed it back to auto they became selectable. I took RAM from another PC without a profile and the options weren't even there at all.
Even when I choose EPP mode and Voltage by EPP save the changes and reboot there is no increase in DDR2 voltage registered in the BIOS PC health screen or Windows. The only thing that does change is auto timings in DDR2 configuration screen. When EPP mode is off the auto timings are 5-5-5-18. With EPP mode and voltage on they change to 4-4-4-12. This is useless though as when I run CPU-Z , Everest, AMDOverdrive, and Central brain idenitifier the timings are still 5-5-5-18.
may be a long shot, but is it possable that you have software set to change your RAM settings on windows load up and forgot to turn it off? Your problem is one I havent seen yet with these boards.
also, are you sure that the EPP voltage is higher than SPD voltage, Most likely the SPD volts are fine for EPP at 800mhz.
Thank you for your reply.Quote:
Originally Posted by G0ldBr1ck
No I have not solved the problems with my motherboard yet. It is still doing weird things.
I have attached BIOS photos http://img365.imageshack.us/img365/9269/pict0511eb6.jpg
http://img365.imageshack.us/img365/9...8177fb97ea.jpg
http://img166.imageshack.us/img166/107/pict0512tt8.jpg
http://img166.imageshack.us/img166/2733/pict0513el5.jpg
http://img88.imageshack.us/img88/574...ilitiesbv8.jpg
You can see that despite changing the multiplier of the memory controller in the BIOS there is no increase in frequency. The same is true when you increase the DDR2 voltage. Even after you save to CMOS and reboot the PC health screen still shows no increase. It always remains at 1.856v which is rounded up to 2 decimal places and shows 1.86 in windows utilities.
http://img300.imageshack.us/img300/669/pict0510xm0.jpg
However when I change the CPU voltage the increase is registered in both the BIOS and Windows. http://img353.imageshack.us/img353/5311/pict0515yp8.jpg
http://img394.imageshack.us/img394/6668/pict0516jw1.jpg
I have tried each of the 4 DIMMS invidually in each and every slot as you asked me to do. There was no change whatsoever. I also tried 2 DIMMS of Hynix RAM that I took from another computer. Again it made no difference. If I leave the DDR2 voltage control as normal and go to Auto Xpress and enable EPP mode and set voltage to as EPP profile. When you save this and reboot the timings in the DDR2 configuration change from the auto of 5-5-5-18 to 4-4-4-12. No change is registered in Windows though. I am not sure if this is connected to the problem but I noticed in AMD Overdrive 2.1.0 that the slider for Mem VTT is greyed out and not selectable.
http://img389.imageshack.us/img389/4823/memvttcf9.jpg
I have noticed other peculiarities and instabilities with this motherboard. Any changes made to BIOS and then saved to CMOS even the most minor of changes like enabling (S.M.A.RT or fan warnings) can cause the motherboard to fail to POST. This has happened several times. Sometimes it will beep and POST successfully and then freeze at the screen where it shows memory size, CPU motherboard model and BIOS and hard drives. Clearing the CMOS and starting again and then booting to windows before restarting sometimes helps. Also sometimes after making changes the motherboard POST successfully but fails when it tries to boot windows. It gives the error message "Windows could not start because the following file is missing or corrupt
C:\windows\system32\config\system" I believe means that it cannot find a Windows installation on the drive. Even though the drive is recognised in POST and BIOS screens it is not reading it properly. I took the drive out and tried it in another computer and there is no problem with the drive. It's a Maxtor 120GB IDE drive that I am using. This problem is sometimes fixed by resetting the CMOS.
At the moment the motherboard/PC is stable and working but there are definite problems with it. These are just the ones that I have found so far.
Originally I requested an RMA because the motherboard would not POST at all. No beep code it was just a permanent black screen. The originally problem was fixed by Gigabyte but it seems as though the initial problem was fixed and it was sent back to me without examination of everything else. Originally I was very happy to have received it back so quickly (It arrived at Gigabyte on a Thursday and I received it back 4 days later on the Monday) but now there are more issues.
Please advise me of what I should do next.
Thanks.
1st question: Very unlikely. I do not believe that I have downloaded anything that does that. Do you know any software in particular that does that?Quote:
Originally Posted by G0ldBr1ck
Even if that was the case the BIOS should still show an increase as the the sofware would only be able overide the BIOS settings upon Windows startup.
2nd question: Yes definitely. EPP volatge is 2.1v. You can see for yourself in the screenshot of CPU-Z in my previous post.
Thanks.
#1 Mem Controller cannot be higher than HT link. If you try it will always set itself back to default of x10. <----Edit: I have this backwards
#2 The grayed out Mem voltage in AOD is normal.
#3 If you have tryed each stick one at a time and tryed another set of ram and still dont get a voltage increase when set. That realy leaves nothing but a problem with a PWM/VRM on the motherboard a physicaly defective part on the board unable to change voltage when requested. Before confirming this I would see if you can get a change of any kind ie...higher/lower switching from epp to spd and if still no changes in voltage of any kind it would have to be the board.
That may be what they replaced last time. I would research and see what the part was they replaced
oh ya, one more thing the PSU you ended up using, did you use a 24pin main and a 8 pin at cpu? if not this may lock out any voltage increases wich in turn the bios may be switching back to spd after seeing incorect epp volts.
Quote:
Originally Posted by G0ldBr1ck
Sorry I couldn't reply earlier. The site was down.
Did you mean memory controller frequency cannot be higher than Hypertransport frequency with this motherboard??
I think that I read somewhere that some AM2+ motherboards prevent you setting them at different frequencies. I did not realise that this was one of them.
If that is not what you meant then you have it the wrong way around. The Hypertransport frequency/multiplier cannot be higher than the CPU Northbridge(memory controller) frequency/multiplier.
Good news about about Mem VTT in AMD Overdrive. At least I am not the only one :)
About the PSU yes I used the main one with 8pin CPU plug. They both have 24pin power plug. The other has only a 4 pin powerplug. The motherboard didn't like it at all. I'm not sure why.
Everthing that I have tried the voltage has always been the same. It is not possible to decrease DDR2 voltage in BIOS. You can only increase it.
I tried decreasing the memory controller multiplier and the Hypertransport frequency to see if there would be any change but there was none.
I probably should have mentioned it before but increasing the HTT reference clock does increase the CPU Northbridge(memory controller)frequency.
The frequency is always 10x the HTT reference clock regardless of what I change its multiplier to in BIOS.
As for the defective component I think this is definitely the most likely scenario.
When I received it back there was a sheet enclosed with it. It think its more for their own internal use rather than anything else. Its location is given as U185. Being an inquisitve soul I searched the motherboard for it. It is underneath and in-between the two orange x8 PCI-express slots.
According to the sheet the part is called 10HL6-180477-20R CLK GEN ICS9LPRS477BKLFICS. It means nothing to me apart from CLK GEN which I'm guessing is clock generator. Long shot but perhaps that part number means something to someone. I also phoned Gigabyte the same day to ask them about it. The bloke on the phone had no idea what it was.
Sorry about that, you are correct. I rattled that off with it reversed in my head:shrug:Quote:
Originally Posted by Phenomenalspeed
Im guessing that the part replaced must be a PLL (Phase-Locked Loop) a clock generator as you suspected wich may have somthing to do with your Mem controller problem.
The Ram voltage is realy sounding like another faulty component like i mention above.
It seems like you have covered all the bases as far as I can think of. There just realy isnt anything else it can be but a faulty board.
Don't worry about it.
Yeah I think I have covered all the bases. I have been exchanging messages with Gigabyte for the past 4/5 days. They have no idea about the RAM voltage problem and dragging it out and making me jump through hoops and now starting to go round in circles.:shrug:
What's bugging me about the RAM is that I specifically chose this RAM as it is rated for 4-4-4-12 timings. Corsair have other PC2-6400 RAM at CL5 and PC2-8500 at CL5. I can't run it at EPP timings without an increase in voltage.
About the memory controller mutiplier. Gigabyte tried to increase theirs and they encountered the same problem. It stays at 2000Mhz regardless. They said they would get back to me after further testing. They got back to me today got and have given me beta Bios F6b which fixes this. I am very surprised nobody on here noticed this problem on their motherboard. I don't know if this BIOS has other changes.
The beta BIOS does seem to work.I was able to get the memory controller/CPU northbridge to 2200Mhz. :up: I tried changing the multiplier to 12 for 2400Mhz and I could not get to Windows. Maybe the memory controller needs more voltage. It's a shame we can only decrease its voltage.
If anyone wants the F6b for the DQ6 BIOS just say so.
Basically with EPP timings and EPP voltage (or manual 2.1v setting) I get quite a bit of Memtest86+ errors. The errors I get are significantly reduced after going below 2.0v. I have heard that especially the B2's CAN be very flaky with higher memory voltages- I guess my B2 is especially flaky. In fact this probably made even more complications with the G.Skill memory.Quote:
Originally Posted by G0ldBr1ck
I'll probably keep the Dominators and get a B3 (or B4) in the future- if every attempt fails to get a higher voltage. But I haven't completely given up yet.
Speaking of B4 is there a release date?
After my last message to Gigabyte they said I should ask for another RMA number.
I have sent my motherboard in again.
They said I didn't have to pay to have it sent back to me this time.
They also said they will give it priority.
I hope all goes well this time.
Hey!Quote:
Originally Posted by Phenomenalspeed
I've noticed it today :p:
The thing is, NB multi was working perfectly fine till I've put my new 2x2GB OCZ Reaper 1066 set. Then it stuck at 10x! (looks like 2x2GB is triggering this behavior, could also be related to 1066 mode - need more time to investigate, but today is F1 quali at the top priority:))
I'm glad you have beta BIOS. Can you share it with me? My NB is happy at 2.4GHz and 1.25V but I need 11x NB multi for that :D
Thanks!
BTW so far quick testing took me up to 1180MHz 5-5-5-15 2T at default settings (Auto timings, EPP voltage=2.1V)
PS. Regarding your memory voltage problem, I can only say it is working for me fine....
Yes it could be that 4GB RAM was causing the problem. I didn't think of that I was using 4x1GB.Quote:
Originally Posted by Lightman
So you undervolted your northbridge and you were still able to get it stable at 2.4Ghz? Interesting. Maybe overclocking northbridge is more stable by raising HTT reference clock rather than multiplier. I will try it when I get my motherboard back.
I have sent you the BIOS. Check your private messages.
I would appreciate some feedback. Please try Northbridge @ 2400Mhz (200 X 12) Also try and change the Hypertranport frequency in BIOS. I discovered another bug when I raised Northbridge multiplier to 11 and set Hypertransport frequency at 2.2Ghz to match it. I saved it to CMOS but when I went to Windows Hypertransport remained at 2Ghz.:confused::shrug:
As for my motherboard I don't know what's going on. I got a phone call Gigabyte on Wednesday. The man said it must be my RAM or CPU memory controller causing the voltage problem even though I have already ruled both of them out.:rolleyes: When he called back later he seemed to have backtracked from this.:rolleyes: He said he would talk to the technicians who are investigating.
Hopefully all will be resolved soon.:)
:up:Many thanks for a BIOS Phenomenalspeed:up:
It works perfectly for me! Memory controller x11 and 220HTT = 2.42GHz NB :) (and 1172MHz RAM :p:)
Quick Everest test.
Attachment 82401
PS. I had 12240MB/s in L3 cache test on lower HTT and higher (13x) CPU multi.:D
You're welcome.:)Quote:
Many thanks for a BIOS Phenomenalspeed
What about 200x12?? I couldn't get to windows at all when I tried it. I am wondering whether it's my CPU's northbridge or the BIOS/motherboard itself.Quote:
It works perfectly for me! Memory controller x11 and 220HTT = 2.42GHz NB
I never doubted that higher northbridge frequencies would be possible when you raise the HTT reference clock. F6b seemed to fix the multiplier but it's no bloody good if all it did was fix x11 and not all of them.
How about Hypertransport? I couldn't get it higher. It stayed at 2000Mhz regardless of what I set in the BIOS.
Interesting and strange. So you lowered HTT reference clock and by extension the northbridge and the level 3 cache increased?Quote:
PS. I had 12240MB/s in L3 cache test on lower HTT and higher (13x) CPU multi
I would have thought that the opposite would have happened. What did you lower HTT reference clock to?
Oh and your screenshot was that just to show the settings? Were you altering voltages/multipliers/frequencies using AMD Overdrive or BIOS?
:up:Quote:
Originally Posted by Phenomenalspeed
Here it is:Quote:
What about 200x12?? I couldn't get to windows at all when I tried it. I am wondering whether it's my CPU's northbridge or the BIOS/motherboard itself.
I never doubted that higher northbridge frequencies would be possible when you raise the HTT reference clock. F6b seemed to fix the multiplier but it's no bloody good if all it did was fix x11 and not all of them.
How about Hypertransport? I couldn't get it higher. It stayed at 2000Mhz regardless of what I set in the BIOS.
http://img206.imageshack.us/img206/8...ram1070wo8.jpg
All setup from BIOS and booted to Vista x64.
Regarding Hypertransport - it did not work for me either. 2000MHz or less regardless of BIOS multi.
PS. Look at the L3 cache copy test. Wired!!:confused:
NB 11x, HTT 212MHz but CPU multi 13x which resulted in 2.76GHz/2.33GHzNBQuote:
Interesting and strange. So you lowered HTT reference clock and by extension the northbridge and the level 3 cache increased?
I would have thought that the opposite would have happened. What did you lower HTT reference clock to?
Both :)Quote:
Oh and your screenshot was that just to show the settings? Were you altering voltages/multipliers/frequencies using AMD Overdrive or BIOS?
I've booted with HTT212 CPU multi 12.5x and then after quick check if NB multi works I've upped HTT to 220MHz. No voltages adjustment...
I will tweak this system tonight a bit more ;)
Thanks for checking that out for me. I did not realise you had updated your last post until just now.
As I suspected the Hypertransport issue is not unique to me but a BIOS problem. This isn't as urgent a problem as you can still use the multiplier slider in AMD Overdrive. I hope that they fix this soon. In fact I wish they would change it to a straight multiplier option instead of frequency selection.
Useful to know that 200X12 Northbridge worked for you. Especially considering you lowered your NB VID. That could mean it's my CPU's northbridge with the problem. I will have to try that again once the motherboard is returned.
Yes your level 3 cache copy test is definitely weird. Strangely high. Better unusually high as opposed to low I suppose.:)
I just thought of something the other day. Has anyone had any success with increasing SBcore voltage (either in BIOS or AMD Overdrive)?? Many say that SB600 limits overclocking potential. I just wondered if increasing its voltage would mitigate that somewhat.
well im on a TA770/SB600 chipset . I must say i was very suprized with my 9950 coupled with SB600.Quote:
Originally Posted by Phenomenalspeed
I Run NB/SB chipset Volts @ 1.31v in the BIOS(stock is 1.15v) TA770 Mobo.It may very well be the same for RD780/RX780/790/790FX with Sb600 chipsets. I stay away from AOD for now..
I may have had a bit of luck,but soon enuff i might have to byte on a 750chipset. My Mobo is Now maxed i think in the power department:D
Do you mean you had better overclocking results with increased voltages to northbridge and southbridge?Quote:
Originally Posted by gOtVoltage
If you look at Lightman's last post you can see in the AMD overdrive screen shot the defaults for this motherboard.
Yes (NB CORE) is the Mobo chipset which i set @1.31v.. Also my NB/SB is combined in my Bios and i have a seprate HTvolt setting which i raise to 1.3v when going above 250fsb.. Raising the NBcore(Mobo)helps with keeping the Ram stable while the Chipset Nb talks to the CPUNB...
The Higher You Oc the Phenon the more MoboNB chipset volts you may or may not need..
If i keep my NBchipset(Mobo) @1.15v i can only achieve a 3.3ghzmax and it also limits my Rams top Freqency..
I have tried numerouse setting with Ganged and unganged @ stock to 3553mhz now. Any more and My Mobo will go to the moon :ROTF:
Quote:
Originally Posted by gOtVoltage
Thanks alot. This is very useful information.
I have good news.
I have a new GA-MA790FX-DQ6.
I sent my one back a second time for RMA they had it for a month. I was patient, hoping that they would fix all the problems, quirks and instabilities that I spoke of before. I was also secretly hoping that they would send me a new motherboard instead if I waited long enough. I never really expected it to happen to be honest. I emailed them last monday asking for some news/ an update. I had heard nothing from Gigabyte since they phoned me 4 weeks ago. Within an hour I got a reply to my email stating they were sending me a brand new motherboard as a replacement! They said the person who phoned me was out of the office and would not be back for a few days. I can only conjecture that they thought I had been waiting long enough and didn't want to keep me waiting any longer. An excellent result considering the RMA specified REPAIR ONLY (NO REPLACEMENT)
I got the motherboard the next day. Not only did they send the motherboard itself but the complete box with all the paraphernalia. All brand new and in pristine condition. In fact it's in better condition than the one I sent them. The swapped the serial number stickers and left the new one on the outside of the box for some reason. This motherboard was made in 2008 and the old one was made in 2007. The software/driver discs are slightly different too. The old one is version 1.0 and the new one is 1.1
This new motherboard is great. All the problems, quirks and instabilities are gone. RAM voltage now works properly and I can now my RAM at its enhanced performance profile of 4-4-4-12 for the very first time.
I have had this motherboard over a week now and in that time I have had to reset CMOS once!! The other one I was resetting CMOS 10+ times a day.
The difference between the two motherboards is remarkable!!
Thanks to all those who tried to help me. (you know who you are)
:up:
Discovered a new problem. Detail here :--------> http://www.xtremesystems.org/Forums/...d.php?t=199447
I would really like to know if other DQ6 owners are seeing the same thing.
I'd be grateful if others could test their systems and let me know please.
No problem in Vista Business x64 SP1. BIOS F6b.Quote:
Originally Posted by Phenomenalspeed
Tomorrow I will try XP MediaCentre.
False alarm.
It turns out it was AMD Overdrive 2.1.2 enabling the TLB patch/fix on startup. I have uninstalled it until I can figure out how to prevent it from doing that.
Thanks for that Lightman.
For all those who have a 9850 Black edition and this motherboard.
Could you look at your Power on self test (POST) and please tell me what you see?
Mine shows:
<CPUID 00100F23 PatchID : 0065>
I'm assuming those with similar setups should see the same thing if all is working correctly.
I am trying to determine if this is related to the apparent BIOS bug which enables the TLB on AMD Ovedrive with Windows startup. This bug seems to be unique to me and I am at a loss to explain it.
Thanks.
Anyone put a Phenom II in this board yet?
In several reviews I've read today PhII clocks better on DQ6 than on DS4H 100+MHz higher which is superior performance when we know that the records for now are made on DS4H. :)Quote:
Originally Posted by G0ldBr1ck
No I think it's out of the production line. Shame, but it really was a good board, if the ACC bug hadn't pulled out of the market all SB600 mobos DQ6 would have had bright future in PhII OC. :)Quote:
Originally Posted by chew*
Do they even sell this board anymore? it's not listed on newegg.......
As of today Gigabyte released new BIOS for DQ6 with microcode update :)
F6D
I'm getting CPU hopefully tomorrow ....
PS. Yes, you can buy this mobo in UK :)
yes! :p:Quote:
Originally Posted by G0ldBr1ck
940 @ 3300 mhz stock voltage and stable.
http://www.xtremesystems.org/forums/...&postcount=153
Crank it up man, this is XS after all :D
If it goes to 3.6-3.7 stable im gonna hold on to this mbo.
nah i must remake a new cpu block first of copper and must get 3 strong blowers.
this quad gets very hot when oc'd. my comp turns into one big furnace.
maybe now Gigabyte will bring out a equivalent board.
how big of a radiator are you using ? 3x120Quote:
Originally Posted by BertM
maybe trying a koolance with 4x120 fans wouldn't be a bad idea.
the aluminum block your using looks really flat, love the pic of it.
I think you might need change where inlet hole is though, centering might gain something
yes i will design my next one with pillars and water entrance on top.
could take some weeks cause i have a huuuge pile of projects to finish first.
edit: waterblock was opened few weeks ago and its totaly spotless! no corrosion visable.
What do u mean? We still need to see how deneb will perform on DQ6 with proper cooling, and DS4H is also out, or is there something wrong with the latter?Quote:
Originally Posted by G0ldBr1ck
The DS4H is 4+1 phase and DQ6 is 8+2.............or am I wrong about the DS4H?Quote:
Originally Posted by -= rtz =-
Well first denebs were tested on that board and broke or came close to the 6 GHz mark, and afaik 8+2 is not needed for denebs, someone correct me if im wrong.Quote:
Originally Posted by G0ldBr1ck
I have 4 other 790fx/gx boards and the DQ6 voltage is rock solid as in virtualy no fluctuation where as my other boards have a good bit. This is helpful when using very high volts
BertM have you flashed your BIOS to F6D??
Regarding upgraded version of DQ6 I think Gigabyte will release something similar but for AM3 and best based around RD890FX! Then I'm moving up.
I'm really happy with this board, I've got it second on XS months before official launch and it's been rock solid, stable and well packed with extras. There are some shortcomings in BIOS department, but they still have chance to improve. Maybe this new BIOS finally give NB voltage increase option? I need to check it tomorrow ...:up:
yep im on bios F6D.
when i putted in the deneb, it booted with the 5000be settings XD (14 multi and other things where still intact)
edit:
@demonkevy666, im using pa120.3 with 2 lian li case blowers now. very weak cooling
let us know what you get it up to BertM
stock voltage according to everest and cpuz but i set it to 1.3375.
in bios voltage was set to auto i think. maybe it doesnt get accepted.
i just upped multi to 15.5 and htt to 220, thats all.
edit: i posted during the prime session and ran it 20 minutes more until i canceled it.
can you raise the NB speed with those clocks?
Very good clocks considering stock/under volts! According to Planet3D review our DQ6 achieved best stable clocks on stock voltages and best 2GHz undervolting using less than 1.0V :yepp:.
BTW difference between MA790GP-DS4H and DQ6 on stock volts was mere 124MHz ....
Link ---> http://www.planet3dnow.de/vbulletin/...#content_start
I think that can be contributed to the DQ6's power regulation. When you set the voltage it is solid and will not droop alowing for the lowest voltage possable.Quote:
Originally Posted by Lightman
I really have a feeling that its gonna be one of the best overclocking boards we have access to for now.
I really don't like the new bios, the old F6b was much better and more stable for OC. I hope the new bios performs better with PhII.Quote:
Originally Posted by Lightman
Gonna be? It is allready :) Im certainly glad I went for this board instead of the DS4 or DS5 a year ago, not regreting it one bit! Now all I need is the new phenom in it and its good to go :yepp:Quote:
Originally Posted by G0ldBr1ck
i have not tried it yet. time is up now and i can continue tomorrow or sunday.Quote:
Originally Posted by G0ldBr1ck
3500mhz was almost instant crash after starting prime, and it resulted in blanc screen and reboot.
f6d has problems booting almost each time. it resets my ram alot.
its insane that this board could be called the best since its the first one build for am2+.
i did some crysis benchmarks at 3300mhz. my automatic pump switch turned off the pump during the session because the cpu was below 32 degree. insane.
in windows it turns on the pump from time to time. perhaps every 5 minutes for 3 seconds.
lol ya they got suckered into building a top notch board back when the world thought Phenom was gonna be the second comming of conroe
board are being picky about this new phenom II I booted my nb at .28ghz with 1.450 volts stable got to windows tried higher and got a no boot redid the 2.8ghz nb and volts and got a NO boot after that >_< WTF.... it booted once with it.Quote:
Originally Posted by BertM
it seems its booting faster when you just left bios for a change.
i have the same with an asus a8v that allows me to set 2720mhz and is pretty stable but then doesnt allow me to reboot, ever!! :confused:
I think we deserve BIOS F7 from Gigabyte for our board. I've flashed F6D and still NB voltage is only allowing to decrease it down by 0.3V max. That doesn't bode well for NB clocking from BIOS at all. Good thing is that NB multi goes all the way up to 20x and CPU multi all the way up to 35x (or 36x??).
I will write an e-mail to their customer support. Who's with me?? What other things would you like to have enabled/fixed in BIOS?
cpu blower settings. temp starting point etc.
current settings make it never run max speed.
To be able to boot after power loss or power disconnect without ram in slots 1 and 2. (when you use only slots 3 and 4)
I have the problems with booting too, so they should fix that too. I have no complaints about something else, just the bios is not stable, last night it crashed several times with no reason and the clocks I reached were lower than with F6b. I hadn't had this kind of problems with F6b. I'm sure they will do something for this problems.Quote:
Originally Posted by Lightman
This morning I finally received my Phenom II X4 940BE!
Here it is:
http://img444.imageshack.us/img444/1...nomii01vk2.jpg
and here you can see how nicely it looks in a socket:
http://img444.imageshack.us/img444/6...te02bh7.th.jpg
I've spend some time running power consumption and benchmark comparison between Phenom 9950BE @3.0GHz NB2.0GHz RAM 1066MHz 5-5-5-15-25 2T and Phenom II X4 940BE @3.0GHz NB2.0GHz RAM 1066 5-5-5-15-25 2T
Note: For power consumption tests my HD4870 is clocked 625(500MHz desktop)/465MHz to save power. Everything else as in my SIG. RAM set in BIOS @2.05V
For game and application testing my HD4870 is set to 825MHz/950MHz
All tests are run on Vista Business x64 SP1 with all latest patches. Power options are set to High Performance which means C'n'Q is DISABLED!
Every test was ran at least 2 times and second test result was taken to minimize HDD activity influence on scores.
Cooling: AIR TT120Ultra + 120mm FAN 12V, ambient temp for first runs on Phenom 9950 around 16C-17C, later runs were 19C-21C.
Below are my findings:
Lets start form Power Consumption:
Phenom I 3GHz/1.32V after 15min. full load - 283Watt
http://img338.imageshack.us/img338/6...283wot9.th.jpg
http://img440.imageshack.us/img440/5...phi283web8.png
Phenom II 3GHz/1.36V after 15min. full load - 241Watt
http://img184.imageshack.us/img184/1...241wab1.th.jpg
http://img440.imageshack.us/img440/9...hii241wli1.png
Special for BertM :)
Phenom II 3GHz/1.32V after 40min. full load - 233Watt (Higher NB @2.2GHz/ RAM@2.15V instead 2.05 like for previous runs - I will re-run this test properly later)
EDIT re-tested with 1.32vCore NB@2GHz RAM@2.05V - 230Watt
http://img440.imageshack.us/img440/4...z233w13pb3.png
Phenom I 3GHz/1.32V idle - 238W
http://img80.imageshack.us/img80/641...ghz238wed0.png
Phenom II 3.0GHz/1.36V idle - 190Watt
http://img80.imageshack.us/img80/369...ghz190wwa4.png
Phenom I 1.2GHz/1.05V after 15min. full load - 206Watt
http://img338.imageshack.us/img338/2...206wfj5.th.jpg
http://img241.imageshack.us/img241/8...ghz206wii7.png
Phenom II 1.2GHz/1.05V after 15min. full load - 185Watt
http://img395.imageshack.us/img395/2...185wix5.th.jpg
http://img440.imageshack.us/img440/5...ghz185wvq4.png
Phenom II 1.2GHz/1.05V idle - 176Watt
http://img80.imageshack.us/img80/466...ghz176wka7.png
----> Undervolting<----
Phenom II 1.2GHz/0.88V after 15min. full load - 179W
http://img338.imageshack.us/img338/2...179wew6.th.jpg
http://img440.imageshack.us/img440/8...hz179w0ij7.png
Phenom II 1.2GHz/0.8V after 20min. full load - 177W
http://img184.imageshack.us/img184/3...177wll7.th.jpg
http://img150.imageshack.us/img150/8...hz177w0up9.png
---->Benchmarks<----
Applications:
Cinebench R10 x64 - look at highlighted and top line
http://img513.imageshack.us/img513/1...hr10x64kd2.png
:rolleyes: clarification - both Phenom I and II are using exactly same setting: 3001MHz core / 2277MHz NB / 1102MHz RAM
Cinebench 2003 x64 - look at the top 2 lines
http://img150.imageshack.us/img150/2...2003x64mj5.png
Phenom 9950 - KirbiBench City - 14.67FPS
http://img150.imageshack.us/img150/1...phi3ghzin5.png
Phenom II 940 - KirbiBench City - 15.69FPS
http://img150.imageshack.us/img150/6...hii3ghzte1.png
Phenom 9950 - KirbiBench Ultra - 4.08FPS
http://img80.imageshack.us/img80/478...phi3ghzul1.png
Phenom II 940 - KirbiBench Ultra - 4.24FPS
http://img80.imageshack.us/img80/994...hii3ghzby4.png
RealStorm2006 - Test1 640x480 - Ref. System --> Phenom 9950 @3.0GHz / Base System --> Phenom II 940 @3.0GHz
http://img80.imageshack.us/img80/711...rmtest1nq3.png
RealStorm2006 - Test2 640x480 - Ref. System --> Phenom 9950 @3.0GHz / Base System --> Phenom II 940 @3.0GHz
http://img89.imageshack.us/img89/225...rmtest2sm8.png
RealStorm2006 - Test3 1024x768 - Ref. System --> Phenom 9950 @3.0GHz / Base System --> Phenom II 940 @3.0GHz
http://img89.imageshack.us/img89/780...rmtest3qc0.png
RealStorm2006 - Test4 640x480 - Ref. System --> Phenom 9950 @3.0GHz / Base System --> Phenom II 940 @3.0GHz
http://img89.imageshack.us/img89/746...rmtest4bk3.png
RealStorm2006 - Test5 640x480 - Ref. System --> Phenom 9950 @3.0GHz / Base System --> Phenom II 940 @3.0GHz
http://img89.imageshack.us/img89/684...rmtest5nj4.png
Generic CPU Bench - Phenom 9950 - 21.366KB/s
http://img89.imageshack.us/img89/865...0090110yj1.png
Generic CPU Bench - Phenom II 940 - 21.489KB/s
http://img397.imageshack.us/img397/6...0090110xe1.png
Phenom 9950 - mBench v1.0 memory test
Access latency - 55.9 ns (146 clocks) wrong value calculated from 2.6GHz startup CPU speed. In reality CPU 3012MHz.
Read datarate (INT) 4702 Mb/s
Write datarate (INT) 3355 Mb/s
Read datarate (MMX) 6334 Mb/s
Write datarate (MMX) 4019 Mb/s
Read datarate (SSE) 6481 Mb/s
Write datarate (SSE) 5032 Mb/s
Phenom II 940 - mBench v1.0 memory test
Access latency - 53.7 ns (162 clocks)
Read datarate (INT) 5403 Mb/s
Write datarate (INT) 3773 Mb/s
Read datarate (MMX) 6986 Mb/s
Write datarate (MMX) 4427 Mb/s
Read datarate (SSE) 7063 Mb/s
Write datarate (SSE) 6485 Mb/s
KMB v0.57x64 Fractal Benchmark
Link to executable ---> Attachment 92669
Results ---> Attachment 92670
GAMES:
Phenom 9950 - Crysis x64 1280x1024x32 noAA - Avg. 26.93FPS
http://img340.imageshack.us/img340/8...1280x10xy6.png
Phenom II 940 - Crysis x64 1280x1024x32 noAA - Avg. 33.26FPS (sorry for wrong label)
http://img150.imageshack.us/img150/6...test168bq7.png
Phenom 9950 - Crysis x64 1680x1050x32 noAA - Avg. 25.45FPS
http://img340.imageshack.us/img340/8...1680x10er4.png
Phenom II 940 - Crysis x64 1680x1050x32 noAA - Avg. 29.26FPS (sorry for wrong label)
http://img131.imageshack.us/img131/9...test128xs1.png
Phenom 9950 - PT Boats DX10 High 1280x1024x32 noAA - Avg. 39.1FPS
http://img184.imageshack.us/img184/4...80x1024yr3.jpg
Phenom II 940 - PT Boats DX10 High 1280x1024x32 noAA - Avg. 45.7FPS
http://img184.imageshack.us/img184/5...80x1024zl0.jpg
Phenom 9950 - PT Boats DX10 High 1680x1050x32 noAA - Avg. 38.1FPS
http://img184.imageshack.us/img184/1...80x1050tz3.jpg
Phenom II 940 - PT Boats DX10 High 1680x1050x32 noAA - Avg. 43.1FPS
http://img444.imageshack.us/img444/1...80x1050nx2.jpg
Phenom 9950 - Quake II 1280x960x32 noAA
Demo1.dm2 / Demo2.dm2
Run 1 - 785FPS / 776FPS
Run 2 - 770FPS / 779FPS
Run 3 - 775FPS / 775FPS
Phenom II 940 - Quake II 1280x960x32 noAA
Demo1.dm2 / Demo2.dm2
Run 1 - 847FPS / 834FPS
Run 2 - 831FPS / 845FPS
Run 3 - 852FPS / 837FPS
http://img444.imageshack.us/img444/2...x960sr0.th.jpg
:rolleyes: Demo2.dm2
Phenom 9950 - Quake II 1600x1200x32 noAA
Demo1.dm2 / Demo2.dm2
Run 1 - 782FPS / 786FPS
Run 2 - 766FPS / 790FPS
Run 3 - 775FPS / 762FPS
Phenom II 940 - Quake II 1600x1200x32 noAA
Demo1.dm2 / Demo2.dm2
Run 1 - 839FPS / 845FPS
Run 2 - 848FPS / 826FPS
Run 3 - 824FPS / 850FPS
Phenom 9950 - Quake III Arena 1280x1024x32 noAA
Demo FOUR.DM68
Run 1 - 596FPS
Run 2 - 611FPS
Run 3 - 603FPS
Phenom II 940 - Quake III Arena 1280x1024x32 noAA
Demo FOUR.DM68
Run 1 - 680FPS
Run 2 - 672FPS
Run 3 - 680FPS
http://img444.imageshack.us/img444/9...1024ok3.th.jpg
Phenom 9950 - Quake III Arena 1600x1200x32 noAA
Demo FOUR.DM68
Run 1 - 586FPS
Run 2 - 596FPS
Run 3 - 601FPS
Phenom II 940 - Quake III Arena 1600x1200x32 noAA
Demo FOUR.DM68
Run 1 - 662FPS
Run 2 - 672FPS
Run 3 - 668FPS
http://img444.imageshack.us/img444/9...1050om6.th.jpg
----> SUMMARY<----
Attachment 92737
First impression is very positive! Just look at LOAD power consumption between Phenom 9950 and 940. What's amazing is that at 3GHz old CPU needs as much juice in IDLE as new one FULL LOAD!
Then there is undervolting and another big surprise. This CPU can reliably operate at only 0.8V 1.2GHz. I have tried 1.4GHz but this ended up by BSOD.
In my short performance review I've tried to use applications different than rest of my colleagues. But even old games like Q2/Q3 and applications improved IPC by 5-15%! Newer applications which are designed with SSE and larger caches in mind can extract even greater boost from tweaks AMD did to Phenom II!
I'm really happy with my new CPU and finally I can utilize this great motherboard in a way I was intending from beginning.
Quick few words about OC. HTT clocking is much easier with new Phenom II, I've struggled to get 230HTT on older CPU, but 267HTT was quite easy with new processor. AOD is working great, except a bug with false clock reading, but this can be fixed by loading BIOS defaults. I'm still exploring limits of my sample on AIR, but so far 3.8GHz CPU 2.9GHz NB looks to be stable. 3.93GHz Cinebench stable for now and finally 4GHz+ only for SuperPi or other single threaded apps.
I hope you enjoyed my tests and that it will help few people make their minds!
:toast:
argh 40 links :p
no problem with me if you show all the pics in this thread, im to lazy at saturday evening :p:
how much power consumption does it take less if you bench the deneb at 1.3 volt at 3ghz?
ah k your still editting. 3 watt less, didnt expect that. cpu probably takes 95-85 watt there.
here is another stock voltage from me :p:
( overdrive showed wrong values in its right colomn so i put cpuz over that to prevent confusion.
can you get all cores @ 3.7 stock v's stable?
no i cant get them all there but 3410 seemed good for 30 minutes prime
http://www.xtremesystems.org/forums/...&postcount=949
im interest for now only in stock voltage oc's or little above or below stock voltage.
atm im priming 3410 at 2200 northbridge. this resulted in 20.234 superpi.
edit: ah bah error after 4 minutes, enough for today.
:rocker:Quote:
Originally Posted by Lightman
:p:
I've done 1h OCCT @3.4GHz NB2.2GHz stock volts :)Quote:
Originally Posted by BertM
Attachment 92671
Attachment 92672
Now it's time to clock this baby! ;)
I will report shortly ...
did you use more voltage on the northbridge ?
i upped the nb core a notch, the nb vid seems stuck.
No, not yet. I've just changed NB multi to 11x in BIOS.Quote:
Originally Posted by BertM
ah yes that seems longer stable now.
i used 10 multi and higher htt 215 or something. that was not stable.
edit: i could be mistaken but cpu seems to be cooler at htt around 200.
im now priming 3535mhz with 2222mhz nb and 202mhz htt.
this was 19.6 seconds superpi.
it could also be that this mobo doesnt like to be high htt stable. i noticed that with the 5000be.
Before I went to bed I've did quick OC session and managed:
220HTT / 3740MHz(1.46V) cores / 2420MHz(1.225V) NB - 15min stable (played Crisis, CB2003x64, CB R10x64, SuperPi, etc...)
200HTT / 3820MHz(1.46V) cores / 2212MHz(1.225) NB - 10min stable (played same games as above :))
Then as a final move before leaving to bed I've looked for 1.46V clock limits and set CPU to 4070MHz - 5 sec. and reboot :p:
Looks like I will need more volts for 4GHz ...
Will see tomorrow :yepp:
Lightman, nice work, keep it coming :up:
Only one thing missing, I either dont see it cause its 2 AM or its really not there, your type of cooling? And some temps would be nice too :D
Thanks Lightman. Good work
Lightman, all I can give you is a :bows: and a :buddies: for the good work. :up:
P.S. Can you do some WinRAR tests with both cpus?
Now it's there :p:Quote:
Originally Posted by -= rtz =-
Cooling AIR TT120Ultra + 1 120mm FAN 12V (ambient temp 18C-21C, further down the page, warmer in my room :D)
Thanks G0ldBr1ck :)Quote:
Originally Posted by G0ldBr1ck
:toast: Stuen4yQuote:
Originally Posted by Stuen4y
Unfortunately P9950 is already on my desk. Beauty of this runs was exact same system and I have only one DQ6 :p:
So for comparing WinRAR too late, but I've selected tests which mAJORD and BenBrown not used in their great comparisons. They did WinRAR ;).
One thing I can add is CinemaR10 x64 under Vista. ---> Updated in review!
under construction ;)
----> OVERCLOCKING <----
If anyone is interested I've did quick NB/volt scaling test.
- 1.175V good up to 2.35GHz
- 1.225V good up to 2.5GHz
- 1.25V good up to 2.65GHz
- 1.275V good up to 2.7GHz
- 1.335V good up to 2.8GHz
- 1.4V good up to 2.94GHz for now ....
http://img242.imageshack.us/img242/3...ng28ghznd8.png
All tests done using 11x multi for NB because 12x multi is no boot for me. I wonder when Gigabyte will add positive voltage adjustement for NB in their BIOS. It's retarded to have 20x NB multi and no option to increase NBvCore :shakes:
Quick comparison of memory subsystem using mBench (settings as on the screen above):
You can compare scores to ones from my previous posts. Difference is massive!!Quote:
Access latency 45.7 ns
Read datarate (INT) 6737 Mb/s
Write datarate (INT) 4363 Mb/s
Read datarate (MMX) 8453 Mb/s
Write datarate (MMX) 4894 Mb/s
Read datarate (SSE) 8622 Mb/s
Write datarate (SSE) 8876 Mb/s
On the other hand vCore regulation is PERFECT! This should be brilliant board to do LN2 or LHe testing. Max. vCore possible is 2.05V!! To get that high just set +0.500V in BIOS and then use AOD to add another 0.200V! :clap: Because board has 8+2 PWM design and is using 8Pin EATX power connector to feed it, going over 300W shouldn't be a problem :)
lightman how are temps with that 2.8 northbridge?
a compilaton of the benches i did on 3616 stock voltage.
this clock is only stable with very low ambient or water chiller.
first bench cpu photoworks from everest shows difference between (tlb bugged?) 9500 and oc'd phenom II.
it makes it look like a single core now.
http://www.xtremesystems.org/forums/...&postcount=313
Very nice benchmarks and creative way of showing them! :up:
My CPU is almost identical to yours BertM! I mean, it can do 3.6GHz stock volts rock stable, but 3.65GHz needs 0.025V more to make it fully stable.
One thing to consider is that I'm running x64bit OS. I wonder what is the clock difference on new Phenoms between x32 and x64 now. Anyone with dual boot??
tx, :p: , it was some creativity with paint shop pro, simple and fast.
cpu crashes above 36 degree with 3616mhz, if i stay below it will remain stable.
this is not close to possible anymore in normal room temp.
it could be that my alu waterblock is realy bad at low temps.
i have a hd and windows vista 64 ready but no time yet to install it.
Awesome power figures, I would be curious to know the undervolting ability of the 940 at 3.0ghz. How low can you push that vcore, and how does it influence power?Quote:
Originally Posted by Lightman
Ok!Quote:
Originally Posted by ryboto
All settings like before and at:
1.25V full load - wall 223Watt - this is fully stable :)
1.20V full load - wall 218Watt - this is fully stable as well :)
1.15V full load - wall 214Watt - this is not stable (crashed after 25 minutes of stress)
Hope that helps!
Really can you share with us(me) what are bios settings for the actual NB/volt? I'm using right now -0.0375V.Quote:
Originally Posted by Lightman
I haven't took any SS but temps reported by AOD were pretty normal. 44C-48C full load depending on Core speed. Looks like between 2.2GHz NB and 2.8GHz there is only 1C-3C difference.Quote:
Originally Posted by Caveman787
Of course :)Quote:
Originally Posted by Stuen4y
BIOS set to:
HTT 218MHz
NB multi 11x
NB vCore (memory controller in GBT BIOS) Auto
RAM 4x (DDR 800 to allow hitting high HTT)
That's about it.
Next in Windows I needed to open AOD and crank HTT step by step (usually by 10-20MHz). At around 250MHz HTT I needed to lower HT multi from 9x to 7x for stability.
Everything else is on that SS :)
I'm sending e-mail to GBT with strong request to add BIOS NB vCore options at least +0.250V, ideally 0.5V!
BTW when lowering NB vCore by -0.0375 power consumption is almost same as on nominal, so I can't see a point in doing that. I've played with this option and on my KillAWatt meter I've seen 1W difference by going +0.125V @2.4GHzNB
Ok who is gonna be first to use LN2 with this board and 940?
awesome! Very good to know. So, that's ~15W savings from full load with 1.35v?Quote:
Originally Posted by Lightman
I really can't find NB VID in the bios, so you're using only AOD? And how are you saving this into bios, by hitting "Apply my last setting result when system boots" or?Quote:
Originally Posted by Lightman
You have no idea how awesome it is to an old Slipgater like myself that you benched a Quake game. Kudos to you, good sir.Quote:
Originally Posted by Lightman
Lightman, be sure to turn Cool'n'Quiet and - IMPORTANT - set energy saving profile in Vista to Balanced - your idle power consumption should go below 100W.
GA-MA790FXT-UD5P (Socket AM3 and 790FX)
http://img61.imageshack.us/img61/940...fxtud5bmr9.jpg
nice. another 8 phase.
It appears that this is one of 3 new Gigabyte AM3 models
With all the upcoming am3 mobos and 925 coming in February, in addition there are no 940's in BG, so all the facts(and maybe the fate) tell me not to buy and 940 now and wait :(. BTW is there any release date for the mobos? I really like that GB get rid of those 8x PCI-E ports, nobody builds quad crossfire with 4 4850... I thought it will happen like this and I won't buy from the first denebs but I was hoping that my mind will stand at background.
the 940 wont work in there, wright ?? :(
945 will come in april.Quote:
Originally Posted by Stuen4y
Thanks for all kind words guys!
@ryboto:
Looking at default vCore of 1.35V and power consumption of 241W full load, your saving 23W on 1.2V :cool:
@Stuen4y:
No, you can't save it into BIOS. You need to use AOD to tweak settings from what BIOS allows you. That's why I want positive voltage regulation for NB in BIOS. With it you would be able to save much higher settings and boot with them.
Here is a pic of my BIOS settings:
http://img176.imageshack.us/img176/8...ingssn3.th.jpg
And no, I'm not using "Apply my last setting result when system boots". I'm booting my comp on low speeds to save energy :). So far I don't need 3.8GHz quad to browse my websites and watch StarTrek TNG :p:
I'm usually idling @1.6GHz 1.1V (need yo find minimum stable voltage for that)
@THE JEW (RaVeN):
Oh yes! I spent loong hours at my school days fragging folks in Q2. First on P233MMX with Virge and then I've moved to K6-2 266 and Riva128 4MB SGRAM :shocked:! 800x600 OpenGL pownage!!
@xoqolatl:
It won't go below 100W! Not with all that stuff I've put into my computer :p:
Minimum I can go is 169W idle.
BTW C'n'Q doesn't work for me on Balanced. Power Saving profile is working as it should but the other one all the time is full throttle. I need to investigate that. That's for Vista, XP x64 works perfectly with C'n'Q :shrug:
Hi all,
F6 bios released. It says 2. Remove ET5, Smart CMOS function. Does anybody know what this is?
easy tuner 5, this was way more thrustwordy oc tool then aod for me, and it no longer works in f6d and i guess f6.
smart cmos i dont know.
Today I've received response from GBT support regarding positive voltage regulation for CPU IMC (NB) and couple of other useful tweaks.
Keep the spirit up! There is a light at the end of this tunnel :yepp:Quote:
Answer - 708063
Answer : We will check if options will be suitable for this motherboard BIOS.
Sometimes we don’t add options just because of BIOS size possibly occupied by other functions.
We will update ASAP.
Regards,
GGTS
good news indeed. Im thinking about moving my 920 to this board for some OC action but im afraid it wont be able to do the 306mhz HT that my Asrock board is doing.
BTW I just got a email from Gigabyte and they have updated the support on the DS5 model for the 920 and 940. Thats funny since this board would not support the 9850 and 9950.
Probably because they have less wattage :DQuote:
Originally Posted by G0ldBr1ck
Speaking about 920 on DQ6, without IMC voltage increase you won't push it very high. Assuming your NB will do 2.4GHz on stock volts and you want to keep it at 9x multi then limit would be 266MHz HTT.
Other than lack of that option, clocking on DQ6 is a blast! ;)