I see, I dont have my UD4 yet still waiting for it to arrive.. probably tomorrow, you recon normal voltage + DVID is a better way to OC that manual volts?
Printable View
I see, I dont have my UD4 yet still waiting for it to arrive.. probably tomorrow, you recon normal voltage + DVID is a better way to OC that manual volts?
For 24/7 with all C states on, idle 1600@~1V. Do you see better option?
F3B still the best as of now ?
IMHO F3 stable version is the best.
it seems a bit more stable than F3B beta and behaves a bit better with Elpida Hyper RAM - still having trouble running super-pi 32m BTW...
Post your settings here Mafio. All of them in Advanced Memory Settings section.
uhm, i've tried quite everything but really nothing change.
with the following settings i can do everything but CAN'T do a single super-pi 2M (yes, 2M, not 32M....).
and this happens with 4 or 5 kits of Hyper i've tried and with every mobos and CPUs.
http://upload.centerzone.it/images/2...2942_thumb.jpg
Try RK Inter. Disabled, 2T , RTL and tRC Auto. tWL at 7 too.. just to see if it works.
Try 1.1v vtt and 1.82v vddr. I get contra-effect with increasing vtt over 1.1v and vddr over 1.82v.
already tryed with Command rate 2T and bank\rank interlaving disabled and quite every different combination of VTT\SA voltage\VDIMM but nothing change.Quote:
Originally Posted by Bullet92
even use loose timings and subtimings (9-10-9-28 etc etc) doesn't make any difference, super-pi 2M freeze after half second.
seems that everything i do doesn't make any difference for super-pi stability and i'm quite sure that here the real problem is this crappy BIOS, not settings or something else.
but gigabyte seems more focused on that Touch-BIOS crap and Z68 motherboards instead of make some REAL BIOS for P67 stuff...
plus, with all the gigabyte p67 boards i tryed (2x UD4-B3, 1x UD4-B2 and 1x UD5-B3) there is no way to boot at high BCLK, if i want 104 MHz BCLK several boot are needed (bios 1005 --> bios 1010 --> bios 1015 --> and so on)....
i'm a long time gigabyte fan but this time i'm really pissed off, this is one of the worst motherboard i've ever used.
BTW, thanks for your support, i really appreciate it :)
Have you tried to run 7-7-6-21 @ 2133mhz ?
I have had the same problem with SPi. I could run everything but SPi 2M+ but after days of sitting and trying I founded the right settings I've posted above. That made it stable for 32M.
Keep playing with subtimings and don't use much of them on Auto except those which I posted.
Playing with Hypers is hard but it worth if you want to do some crazy run ;)
yes, sure, tried also at 2133 MHz, cas7, 8, 9 but same results...half second and then freeze. :DQuote:
Originally Posted by Bullet92
never used AUTO subs and i'm quite sure that they aren't the problem, usually too tight timings give you not exact in round, not freeze...and here i've never seen a damn not exact in round, only freeze.
IIRC you use corsair dominator, have you tried something else? eg SuperTalent, Kingston HyperX, A-Data?
Noup, only have these GT's. But don't think I like this board. I hate it :D
i'm asking that because till now i see ONLY guys with OCZ Blade and Corsair Dominator doing super-pi 32M at high clock with sandy :banana::banana::banana::banana::banana:.
i also read about someone that flashed his SuperTalent with GTX2 SPD but nothing changed...bah
hell, i'm still using this crap because is the only motherboard that can unleash the real power of da :banana::banana::banana::banana::banana: in 3dmark 01.
i'm hoping that the new EVGA mobos will run hyper without problem.
Quote:
Originally Posted by Mafio
I just tested this on my UD5 B3. F3G BIOS, main and backup are both F3G:
Set 104.0 BCLK in BIOS, saved, board powered off after a few seconds, powered back on, booted at 104 BCLK, first try.
Board powers off and on before setting high BCLK, powers off and on again when setting BCLK back low again.
I have NOT tested this with PLL overvoltage.
Are you sure your main and backup matched each other?
Were you using pll overvoltage?
Vccio = QPI/VTT ?
I ran Kingston HxperX 2000 CL8 with Elpida Hyper DDR3-2133 CL8 with 1.65 on UD4 B2 F8i. I crossflashed to B3 F3 and Hyper dont work anymore. :rolleyes:
SB got an idea if F3d got LLC levels - not just on/off - i read it somwhere but since i been testing a P8P67 Deluxe in the meantime i haven't tried myself F3d.
I read somwehere F3d should be best Bios for UD4 B3
It is confirmed that GA-Z68X-UD4-B3 got 10 LLC levels.
It is available in a german shop - the owner coops with gigabyte bios developers. Many bios fixes been done due to his involvement.
He got GA-Z68X-UD5-B3 and GA-Z68X-UD7-B3 in Stock too !!!
http://www.jzelectronic.de/jz2/index.php
Regards
Where can I grab a LinX thats for SB?
yep.Quote:
Originally Posted by Demo
http://www.mediafire.com/?prlre1k1rb30tdzQuote:
Originally Posted by Demo
Thanks, whats been changed from the F3 to F3D?
F3d is a beta - i was told Gigabyte got two different teams working on beta and final code.Quote:
Originally Posted by Demo
F3d supposed to be better - have not tried myself yet - it does not have LLC leveles - which sucks.:down:
F3 seems to have VCore jumps during system start up with overclocked cpu.
So JZ_ (the guy wiith contact to bios developers) says you better should use F3d.
yes, sure same bios on main and backup epromQuote:
Originally Posted by Falkentyne
and this BCLK weird issue occurs on at least 3 boards (2x UD4-B3 and 1x UD5-B3) each one with different CPUs and RAMs.
definetly not a problem of settings or a faulty board but ANOTHER bios issue that gigabyte don't want to fix.
if i set 104 MHZ of BCLK the board do a fake boot and then starts but hangs on post error code 56 showing overclocking failed message.
honestly never tried with PLL disabled.
awesome work gigabyte...awesome...Quote:
Originally Posted by IronAge
where the measured voltage, I mean the points that do exactly the measurements of the digital meter
On the left chokes/caps legs.
I've been reading on other forums, cant really get a solid feedback on the F3D, I might as well try it or just stick to F3.
Sigh....Quote:
Originally Posted by Mafio
I just tried 104x50, 5 ghz, PLL overvolt on (directly from a previous setting of 4 ghz 100.03x40), and it powered off, powered on and booted right up at 104x50, first try...(this is on a UD5 b3 with f3g....)
I'm not sure why your UD5 B3 couldn't do it...
F4 BIOS is out.
thats what i was waiting for :D
i hope the llc options are now there...
change log?
-fix bug with DVID (LLC disabled)
Any link to f4 bios?
Link in my sig.:DQuote:
Originally Posted by Calathea
Im using F4 now I dont really know if its good or bad since its the first bios I tried!
Now this is odd, If i enable HT. CPU-Z sees 2 cores 4 threads, if i disable HT i get 4c/4t.. what seems to be causing this?
Windows task manager also shows 4 cpu graphs only.
With HT on device manager shows 8 processors listed.
thanks :) I might need a hand with 1600 c6 GTs on this board later on... SB isn't exactly boot--> get beer :O
hyper are still a no go...
anyway it seems a bit more stable than f3 and still strong in 3dm01.
F3 97200 :: F4 97276
http://upload.centerzone.it/images/2...6298_thumb.jpg http://upload.centerzone.it/images/4...5462_thumb.jpg
Wow, 1060 7-7-6. I can't post with 1:8 divider... maybe it's a cpu thing?
play with VTT and System Agent voltage (i used 1.2 and 1.005 volt), i'm able to do this kind of frequency with all of my Hyper kits and all my CPUs.Quote:
Originally Posted by Calathea
the problem is that on this board there is no way to make a damn super-pi 32m, after 1 or 2 seconds windows freeze. :D
still waiting for a bios that finally fix this issue, but i'm quite sure we will never see it........gigabyte is focused on touch bios crap, they have no time to fix the :banana::banana::banana::banana:load of problems that their p67 have. :rofl:
good to know f4 didn't kill 01 score lol
After shorting the clear CMOS swtich is it expected for the board to automatically power up ?
Also spamming alt+f12 should flash the backup with the main bios right? I dont seem to get that option.. and how do I verify the bios version on both chips?
you should have the board off..and the power supply off...before you clear the cmos...
so: it shouldn't automatically power up
Yes make sure the board is off (you already know that) and the power supply is also turned off or unplugged. If you want to do a proper clear,Quote:
Originally Posted by Hondacity
wait 60 seconds after turning the board off, and press the power button after you unplug the power supply, to clear the +5 vsb.
Yeah thats what I do, the Stider Gold is a little pain though it has no power switch at the back.. why silverstone left that out is beyond me!
I using manual voltage right now and 1.371 volts seems to be sticking even if i just feed it 1.320v
Using F4 bios.
EDIT:
Setting the bios to normal with an offset of -0.015 still gives me a VID of 1.371v under realtemp 3.67 but it gives me an error in just a min with the smallFTT test.
enhance built strider gold series , enhance built antec tpq1200, no switches...lol
Im really confused about setting the voltages on this board, no matter what i choose I always see 1.381 VID under ReatlTemp 3.67.. How do I determine what voltage is actually getting fed to my processor?
I've mimicked the automatic voltages defined at 4.5Ghz which is 1.360 under bios, it fails if I manually type in voltages even with LLC.. I've also tried a +0.010 to +0.025 offset using "Normal" voltage and im still getting errors in prime95 smallFTT, pure auto makes it work.
actual voltage: multimeter
et6: reads voltages..not actual but close
cpuz: reads voltages..same as et6
aida64: reads voltages..same as et6
Right now I have Vcore on my bios at "NORMAL" and the VID it gives is 1.360v
Booting into windows and checking with realtemp 3.67 shows me a VID of 1.3811, putting load on the processor moves it to 1.3861
I have the DVID under the bios at +0.050, so that means 1.360 + 0.050 = 1.41vcore ?
I seem to be instantly getting errors on Prime95 SMALLFTT test..
I have PLL over voltage enabled, went as high as 1.200v on VTT/QPI
RAM is at 1333Mhz 9-9-9-24-2T
And this is just 4.8Ghz, Im at a loss on what to do.. I've tried to make the chip POST and enter bios upto 52x and it can post.. I must be doing something wrong ? Im using bios F4.
All of this voltage is seen by CPU-Z as 1.2000v
Having the voltage set to auto also gives me 1.360 in bios and 1.3811 in realtemp but it shows about 1.068 something on CPU-Z
You're using the wrong cpu-z. There is an updated 1.57.1 something that reads the vcore properly.. Your CPU-Z is showing the VTT. I don't think the updated cpu-z is on their page since it's beta...if you use the search, you may find it, but I know a link was posted a few weeks ago in the UD7 thread.
You're right about the DVID, it SHOULD be 1.41v, since it's VID+offset, but you're also getting vdroop.
Try setting a DVID of +0.08 instead of +0.05, and see if that helps. That should compensate for the vdroop.
Of course the proper answer would be to use LLC, but seems like you can't use LLC with dvid anymore....
Realtemp does NOT show vcore at all. VID is not the same as vcore.Quote:
Originally Posted by Demo
It shows the VID which the processor tries to request as an "Auto" voltage.
Look for the updated cpu-z 1.57.1 beta, or use hwinfo32 to see the voltages.
I'll get a hold of that new CPU-Z, under the bios LLC can be enabled with DVID.. by using 0.080 i might hit the mid 80's since with +0.050 im getting 77-78cQuote:
Originally Posted by Falkentyne
EDIT:
Any power saving feature on this board that I can easily enable w/o any stability issues?
Gaaaah so frustrating. This mobo has to be the least user friendly board ever. Got pi32m to loop 16 before it froze, now I can't boot with 2133 again.
lol, welcome to the club :DQuote:
Originally Posted by Calathea
I just got my replacement harddrive back from samsung which is a 1TB F3 drive ( had my 2yr old 1TB F1 replaced ) and Im thinking of going RAID0 on this board, can anyone give me some tips on how this XHD works from Gigabyte, also I was looking at the drivers page and I saw drivers for Marvel and Intel raid, I suppose the one on the UD4 is Intel based?
Would using the auto configure from XHD automatically create the raid0 stripe for me or do I have to reinstall everything from scratch ? Also the installation on my windows was done NOT in AHCI mode if this matters.
And hows the performance of the RAID0 on this board, total good for an onboard solution ?
and whats the best block size?
Just buy SATA6 SSD already and then use your 1TB for storage :)
Too expensive as of now, maybe mid 2012.
-NVM- found answer.
today i had a 3 hours bench session with this mobo, a 2600k and the crappiest GTX280 ever.
anyhow managed to get some quite good results.
hardware used:
3dmark 01 :: 120633Code:i7-2600K # L044B675
Gigabyte P67A-UD4-B3 - bios F3
KHX 2k cas8 MGH-E-Hyper
PCP&C 1200 Watt
GTX 280 + VGPU + CAPs mod
DMAX 200 GB sata
XP SP3
Single Stage Rotary 1.2 Hp - CPU
Guglio's GPU pot rev 1.0 - VGA
http://upload.centerzone.it/images/8...6348_thumb.jpg
Aquamark3 :: 411392
http://upload.centerzone.it/images/7...3395_thumb.jpg
here are some pics:
http://upload.centerzone.it/images/8...9829_thumb.jpg http://upload.centerzone.it/images/4...9318_thumb.jpg http://upload.centerzone.it/images/9...1081_thumb.jpg
http://upload.centerzone.it/images/1...2061_thumb.jpg http://upload.centerzone.it/images/7...9250_thumb.jpg http://upload.centerzone.it/images/9...7412_thumb.jpg
i must say thanks to Guglio for the VGA and this awesome GPU pot.
whats the gold string for lol?
eeeek no heatsink of mosfets...
strong am3!
nice to see rams at least work in 3d...
Nice score anyhow! If you have problem with cold slow there's a simple fix you might know of, disable the chip that manages temp. Here's how a friend of mine did it, we benched his card at 1150MHz @ -145C nature :) http://www.techsweden.org/forum/view...?f=15&t=225644 (in swedish)
Also OCP mod for VT1165 is now available.
thanks honda :)Quote:
Originally Posted by Hondacity
the gold string was used to keep the POT on the GPU :D
i've no mounting stuff to fit this POT on this VGA, so had to do it in the ghetto way. :D
calathea!!! ILOVEYOU :DQuote:
Originally Posted by Calathea
wait where is ocp mod for vtt1165?
Pin 31 on VT1165M is R_SEL 4. You set your desired current (and slave setting) based on a resistor to gnd. This doesn't entirely diable ocp but you can easily raise the limit 10A/phase.
This is confirmed working on 9800gx2 using 1165 slaves. I set R_sel4 to 5.11kOhm and have no problem with ocp anymore :)
http://imup.se/i/SaNLSmkdQl.png
Volterra docs can be found here, thanks to TiNs überleet search skills http://www.xtremesystems.org/forums/...d.php?t=264080
thank u so much!
my 280#1 died days ago...but 280#2 will be tortured :D
GT200 gets another chance to shine with SB :D
I've been getting the error code 124 (BSOD) with NORMAL voltage and NEGATIVE 0.025 , this is about 9hrs into Prime95 BLEND.
So I decided to try something else...
I tried to manually volt it at 1.360 w/o any offsets this time
Ran 30 loops of LinX ( I dont really like this as ppl say it degrades the processor so fast )
Right now at stock 1.050VTT
Memory at 1333Mhz 9-9-9 1.5v
HT-OFF, all power savings off
All other voltages in AUTO, PLL over voltage OFF
http://i1184.photobucket.com/albums/.../quest_4_5.jpg
Error code 124 reffers to VTT/QPI do I really need to increase it away from stock considering this is just 4.5Ghz or I just lack the raw Vcore?
A day ago I've been trying to start with 4.8Ghz at 1.404v loaded VTT at 1.2, PLL overvoltage and Im also getting the same error code.. I must be missing something or doing something very wrong..
Bios is F3B
yesturday updated UD3 to UD4 and i say big fail.
unreal overclock on UD4 (bios f3) on five s attept getting 4,5GHz and more oc like 5+GHz onreal to get its alltime RESTES RESETS RESETS, even big volts not helping. cpu can do 55x. on last UD3 MB was no problen getting even x56 with this cpu (only lowered bclk)
maby to tray F4 beta biose?
E: flashed to f2 back ...all seme,i f¤¤¤ing hate this mb :cord:
sometimes even with all def. settings cant restart computer, it will not boot up..its trys something reset again.
Next week will buy msi.
Use BIOS version F3D and use alt f12 at beginning of POST to copy the main BIOS to the backup BIOS. Your problem is F4 is not stable (people have had various problems with it, including unable to overclock past 4.4 ghz without 1.2 VTT (confirmed bug), and you need to copy main bios to backup BIOS but flash F3d first. (after flash, and after setting your settings once first and successful boot, hold alt f12 before you see CPUID strong appear at beginning of post).
thanks but i already have the DS of VT1165. :DQuote:
Originally Posted by Calathea
here seems that there is no coldslow or OCP problem, anyway i will do OCP and mess with that temperature related chip, maybe i can get something more out of the card. :)
i can bench at -100 or -180 but no gain in frequency, probably the problem here is that this card just sucks. :D
ah, did your card scale with voltage?
mine here not very well, from 1.4 to 1.65 volt there is really a small gain in frequency. :(
GA-P67A-UD4-B3 - F3d
this? its beta?
bookmark these 2 links:Quote:
Originally Posted by Mask
1- http://forums.tweaktown.com/gigabyte...beta-bios.html
2- http://forums.tweaktown.com/gigabyte...tml#post270423
F3D - http://www.fileden.com/files/2009/12...67aud43.3d.zip
We used around 1.4 vgpu IIRC. 920 core is weak. Maybe your problem is pot mounting ;)Quote:
Originally Posted by Mafio
On topic: Gigabyte failed hard with bios on this board. I'm surprised cause my last Gb mobo, h55m-ud2h was a pleasure to use even with extreme oc.
When I got my board I started with the F4 right away and both 2600K's I had could not do 4.5Ghz w/o the 1.2VTT.. I was like going crazy then decided to try F3b, whats the diff from F3b to F3d?Quote:
Originally Posted by Falkentyne
Also how low can the PLL voltage go w/o really messing up anything, 1.720 ?
can anyone identify either litter picture, where the measured voltage at the ud4?
I tried most of the legs of the capacitors and coils at all times display the 1.092v on my multimeter, the tubes showed me around 3v and 4v, what to do, where to measure?
vcore? did u try this?
Big thanks, I checked and it works, the bios, 0.01v tile work on more, it's not bad, I use the f4 works fine bios'es
offset +0.110V / LLC: Enabled
Linx: CPU-Z 1.464V ~ 1.476V / DMM 1.48V
Prime95: CPU-Z 1.452V / DMM 1.46V
GA-P67A-UD4-B3 F4 MOD
P67A-UD4-B3 F4 Mod.zip
Change log included, BIOS updated with Intel RAID ROM 10.5.0.1034 and Realtek LAN 2.39
LSD:,
can you make a mod for F3G on UD5?
Ya, give me a second and I'll post it to that thread :up:
How low can a PLL voltage go till it really makes the system totally unstable. 1.50v? or this is purely CPU dependent ?
1.3 sometimes!
Alright, I got some issues that I need help with though.
Using Bios F3B
4.8Ghz , 100.33Mhz BCLK
Mem at 1333Mhz 9-9-9 2T
All voltages in auto except for;
Vcore, Manual at 1.455v CPU-Z reads it at 1.416 and 1.428 depending in whats going on with prime BLEND
CPU PLL Internal overvoltage OFF
Powersaving off,HT off
PLL is at 1.70v
I get a BSOD in blend test either error code 123 or 400 (???)
What should I try next, lower PLL ? maybe hit 1.600v?
EDIT:
Right now im seting PLL 1.600v and the CPU temps dropped a few c! I should probably try to see if my CPU can do 1.3v
can they do 32M with 7-8-7-21 ?Quote:
Originally Posted by Mafio
I had similar probs on the M4E and UD5, turns out to be a lesser stick. 3D is no sweat, but 32M locks up , errorred out... Not even at 1.8 does it do 7-7-7-21, the other two do 2133 7-7-7-21 at 1.7Vdimm, 2200 CL7-8-7-21 1.74...
On other platforms the bad stick works great...
Demo, you may need to enable CPU PLL Overvoltage since you are on a high multiplier.
Probably not needed since you can boot that multi, but it would be worth testing to be sure if it helps with stability or not.
What version of this Pi testing are you guys using ?
I'll give that a try, the chip I have now can only boot at a max of x49 w/o PLL over voltage ( which IMHO sucks )Quote:
Originally Posted by Lsdmeasap
I have the PLL running below 1.80 would that PLL over voltage still help even at lower than stock PLL?
What I hate about Prime blend is that it takes ages!
EDIT:
After 47mins Core 2 failed
http://i1184.photobucket.com/albums/...Z088/error.jpg
Right now im clueless on what to change.. its more of a guess work on my end..
-double post-
llc on?Quote:
Originally Posted by Demo
Pll and PLL overvoltage have nothing to do with each other, but I don't know how that would affect using low PLL and PLL overvoltage enabled.
Yes LLC is on, Ill try PLL over voltage as well.
rounding error im guessing is memory...what is vdimm?
STOCK 1.500vQuote:
Originally Posted by Hondacity
1333Mhz 9-9-9 2T
G.Skill 6-8-6 1600Mhz 2x2GB kit.
Oh yeah the UD4 has some LEDs going on when in POST manual says its the phase led but goes out after post.. do we need to install some software for those lights to show up again in windows?
is that p67 kit? did you try xmp profile?
Yes the P67 Kit, XMP profile also gives it 1.500v and VTT at 1.050 , For now Im not using the XMP profile just to eliminate the memory as one of the variables.Quote:
Originally Posted by Hondacity
http://www.gskill.com/products.php?index=352
Quote:
M/B Chipset
Intel P67
CAS Latency 6-8-6-24-2N
Capacity
4GB (2GB x2)
Speed DDR3-1600 (PC3 12800)
Test Voltage 1.5 Volts
you haven't ruled out memory by not using xmp...
use xmp to rule out memory...
I dont see how running the memory faster will rule it out of the equation but I'll do that since im the one seeking help.
I've rebooted and its not running at XMP profile 1
Would it matter if the memory profile was under standard/turbo/extreme?
Will now run blend test.. and gonna wait for hours or so..
The voltage read out photo you posted on the previous page, is that the only read out pin or any of the 2nd pin on that entire column will work ?
if you're talking about the side of the inductor...yes they're all the same point. im not sure if you can access between inductors...
the lower part i think is for vccsa but you can still measure the voltage from there..just insert your ground probe to psu ground(molex).
as for rams..i've seen two kits go bad..from the xs forums...both from gskills..so i still suggest test rams first if they can do xmp. if they fail memtest...they're worth rma'ing
if the mems pass then its time for other tests...
http://i1184.photobucket.com/albums/...88/readout.jpg
Just to make sure I got it, all the points I've marked with green can be used for voltage read outs right?
As for the RAM's what app do you recommend I use to test them with, Mem86+ ? How many hours/any specific test to loop? and should i test it at stock 3.4Ghz speeds with XMP profile?
1.yes .....im not sure about the bottom one...just look at the capacitor...the blue marking is grnd...other side is read point..
2. me86 is ok
3. yes..id test it at 3.4ghz ..with xmp ...
For Mem86 how many hours would you recommend any specific tests to loop ?
Test ram with HCI Memtest (you need to run multiple instances with eg 512 or 1024mb set, till you run out of ram ) in windows, Memtest86 will not discover errors as fast ( if it discovers errors at all )
1pass is ok
5passes takes forever lol
I'll try both. 1 pass for Memtest86+
how many hours for HCI memtest?