Printable View
stasio you have to put new settings and clear bios , then if you know your load stable voltage , pick a vcore you had before and work your way thru the llc s on load to you hit your same load vcore again , llc1 and llc2 or not the same anymore , like sin said 5 is closes to old llc1 but its still offQuote:
Originally Posted by stasio
Did some more testing on LLC + offset issue.
Any LLC level + Vcore [Normal] + DVID [Auto] - works
Any LLC level + Vcore [Normal] + DVID [Normal] - works
Any LLC level + Vcore [Normal] + DVID [+0.000] - works
Any LLC level + Vcore [Normal] + DVID anything but [+0.000] - infinite reboot loop till CLEAR_CMOS.
Everything else Optimized Defaults.
Yes,I agree with you.Quote:
Originally Posted by the_real_7
BIOS is different,but NOT defective,as they attack me on "hardwareluxx.de" forum.
Yeah,Quote:
Originally Posted by the_real_7
notice this a few hours ago also on TT forum:
http://forums.tweaktown.com/gigabyte...-bios-454.html
Did you try LLC disabled?Quote:
Originally Posted by nijel
For me no point LLC1-10 and DVID together.:confused:
Like you have full tank of fuel in your car and want to put some more. :shocked:
Bios is great Stasio , seem you made a simple mistake with not clearing the bios no biggie , if they attacked you they're a lame forum , as we are here to help each other not bash each other :)Quote:
Originally Posted by stasio
Yeah,they in Germany had his own "Guru" and they beleive only to him.:rofl:Quote:
Originally Posted by the_real_7
They had been surprise and sceptic,how I post BIOS before GB server and act like half closed eyes horses,waiting always for "Jockey" command.
i didnt clear when flashed to F8 new bios for B2 and had higher volts damn now its perfect searching the right LLC now :)
First off what do you mean clear BIOS?Quote:
Originally Posted by stasio
Second Stasio you do a ton for our community, we have no one else that notifies us and makes it possible for all of us to have access to so many different BIOSes and programs so quickly. I love how you always have the latest and greatest BIOS, and I think JZ is just a work of art lol. Did you beat him to it, is that why he said they are defective? lol.
Stasio maybe you want to show them the chart to explain trends for the new LLC as well as difference from software to real measure.
To be honest, i think they should have first implemented the levels into a public beta earlier, but i think they wanted to revamp LLC to give the user more control. Its a great thing to say hey my board has 10 levels of LLC! lol. Levels 5 and 6 are great for less difference between idle and load, Level4 is great for 24/7, level 5 and 6 for normal mild ocing, and level 7-10 with lower vcore set the higher in LLC level you go is a great thing. it coudl all together replace DVID if you have a DMM(hint hint) If we are seeing .1v increase at LLC level10(theoretically) and you have a DMM, then you can kinda make that your DVID, it could reduce your ripple.
For instance with DVID you set a 0.1v offset, with lets say LLC level4
Your stock VID is 1.2v and so at load your processor is give that extra +0.1 and then LLC level4 drops that voltage slightly.
Take another case with LLC level8 or 9 and lets say it increases your vcore 0.1v like LLC level2 did on previous BIOSes(i haven't looked to see what it is on any of the new tested levels, i was so tired from doing all that). So idle you can be at 1.2v and then load 1.3v without any droop. The only difference here is that you would set something like 1.15v for your stock VID because you need to account for the offset that comes with LLC>
Those are hypothetical situations, but there are many cases where a very high Vrise can replace DVID.
Thanks Sin,
where I post "clear BIOS"?
He,he no time yet,to test like you,as always provide to us valuable info.
well im lost with so many LLC..
i was stable at 5005 HT with CPUZ LOAD 1,416 - and CPU IDLE 1,452 now i cant seem to find to WHAT LLC i pair with :( any help ?
For me, its level 5. Try level 5, andre.
I tried level 4 at the same bios setting and got INSTANT restart when i pressed "start" on prime blend. I swear it restarted in 1 second...didn't even give the cpu a chance to warm up...went back to windows at level 4 again and had enough time to run prime blend to see that the load vcore had dropped to 1.380v at 5 ghz.
Went to level 5 and no problem...(1.404-1.416v load) and no problem :)
Sin, do you still have the old standard, level 1, level 2, along with the 30 different LLC's, after you cleared the CMOS?
well i went back 2hours on this an cant figure out :shrug:Quote:
Originally Posted by Falkentyne
im stable 5005 HT on this settings
BIOS SET 1,47 LLC1 B2 BIOS F8f
BIOS READ 1,452 / on chokes 1,464
CPUZ IDLE 1,452 / on chokes 1,468-69 ~ 1,47
CPUZ LOAD 1,416~1,428 (most of the time 1,416) / on chokes 1,443 ALWAYS
tried LLC5 and LLC6 and not only got more vcore on idle but BSOD on PRIME :shrug:
its load vcore that is changing. try level 7. Its odd though i don't get droop on levels 5 and 6 at 1.45v set.
Guys, I know this is XS but let me ask. What's the point running fixed Vcore for 24/7 nice moderate < 5 Ghz cklock?
When I was done testing my chip I sattled down and optimized the Vcore with LLC1 and DVID for stable 4.7 GHz (my cooling max) but I was getting random BSODs in idle Vcore ~0.9V. I disabled LLC and upped DVID which gave me higher ~1.1V idle Vcore for the same full load Vcore. If you guys remember I asked back then why GB have only 2 LLC levels when other boards have 5. Now we have 10 for much better control and I'm prety sure I can even better optimize my idle/low load voltage ~1V and still have a stable system. If only new LLC + DVID was working... It will I'm sure in GB.
You can't get same Vcore scaling when fixed with LLC only and no DVID. We need both working togeter to have full control over the entire 0=100 load range Vcore.
Correct me if I'm wrong.
will try thanks sin...i flashed back cauz it was driving me crazy now im stable again on F8fQuote:
Originally Posted by sin0822
I swear ive cleared the cmos with power unplugged
Im on LLC9 bios 1.54v, ET6 idle 1.584v, load 1.524-1.536-1.548v. Most of the time BSOD once it fluctuated to 1.524v.
That was on latest LinX dated few days ago Exactly the same settings running Prime95 26.6 x64 custom blend 3100mb min fft 128, load vcore stays at 1.536-1.548v, never gone down to 1.524v which will cause the bloody BSOD. 30mins now...
I'm not sure why yours is drooping when it should be rising.
Even Sin's one is rising.
What I find more shocking is that some people (Andre, Sin) are reading the vcore hardware as much higher than software, while real_7 is reading it only 0.01v higher. And my software hwinfo32 matches his software and hardware too so I think my volts are about correct with hardware. Even though he's using UD7 and I'm using UD5.
And my UD5, 1.450v bios llc4 5 ghz, INSTANT reboot on prime as soon as I pressed the start button. Got back into windows again and managed to start prime blend again without immediate restart and noticed the load vcore showed 1.380v. No wonder.
Set it to LLC5 (this time it showed 1.404-1.416v on the first 15 minute loop) and ran multiple loops successfully. Even when on the second loop (15min-30min) (the one that gives higher temps) it showed 1.390v in hwinfo32...
You definitely need a DMM to see what's going on. Seems the same boards, some are reporting accurate voltages with software, and some are way off...
LinX puts more load on CPU but not constant and when this happen you get 1.524v. You need just 0.005V more to not go under 1.536V. At least that's my experience.
well maybe its because B2 mate or that im not using F3 modded for B2...dunno...still cant pass mins of prime till LLC6...i can pass 1h+ easy on BIOS F8f on B2 with the values measured i mentionedQuote:
Originally Posted by Falkentyne
will keep trying tomorrow as i liked F8 bios
important to read load one must run prime i dont know what real tested but i tested that way
cheers !
Sergio
Yes exactly what is im facing right now. Anything lower than 1.536v will result in BSOD 124. In prime load vcore as well as max temps wont fluctuate so much.Quote:
Originally Posted by nijel
Have you tried increasing qpi/vtt maybe you are trying to over compensate vcore for low qpi/vtt?Quote:
Originally Posted by owikh84
Andre,Quote:
Originally Posted by andressergio
Is Sin using a B2 modded board to B3 BIOS? Do you happen to know?
Guys,try to compare latest LinX with newest Prime95 v26.6 Build 3.
Nothing can beat 4 threads LinX AVX on 2600K 4c/8t. Still puts 135W, 105W with this prime95 version.
No i have a P67A-UD7-B3.
very wrong mate you can pass linx easy on sandy but NOT primeQuote:
Originally Posted by nijel
Tried but dont think it helps. Normally i will touch vtt when im changing new ram.Quote:
Originally Posted by acebmxer
Im using the version. Custom blend min fft 128 3100mb which will stress most of ram amount.Quote:
Originally Posted by stasio
Yes for sure, since extra threads hv to run on the same core amount. Making each core overloaded.Quote:
Originally Posted by nijel
Whatever works for you.Quote:
Originally Posted by andressergio
Never had problems with prime if I pass 25 LinX AVX 4 threads max ram first.
LinX AVX 4 threads catches low Vcore on 2600K 4c/8t instantly, BSODs or errors 5 runs max.
Same thing for idle/low load Vcore with LinX AVX 1 thread. Get that stable as well and you never get idle BSOD for 24/7 system with DVID.
I can pass both very easily:DQuote:
Originally Posted by andressergio
but really running linx with 4 threads and SP1 on w7 is very stressful and even the newest version of prime does not raise temps or wattage as high as running linx with 4 threads selected and not 8
sergio its complicated only if your starting new in the new bios .. . but tomorrow if you show me your f8f settings ill make them work in new bios :up:Quote:
Originally Posted by andressergio
thanks bud :up:Quote:
Originally Posted by the_real_7
I agree that linx is way more stressful a , but prime proves long hour stability cause you can run it longer without degrade and ive hardly ever seen people these days who pass 20 linx ever pass prime 6 hours or more and also i can show you , that linx will pass easly with less volts as prime need more volts . . but its all to the user , if your a gamer and bencher linx is more than fine , but if your like me a heavy decoder multitasker prime is the best stillQuote:
Originally Posted by l0ud_sil3nc3
Hey guys, i'm seeking guidance with some USB3 fun.
So i have P67A-UD7-B3 /w F3 BIOS & Win 7 64bits
I picked up a USB 3 hub :
D-Link DUB-1340 (http://www.dlink.co.uk/cs/Satellite?...GB%2FDLWrapper)
It seems there is no pilot for this piece of hardware and it's plug&play.
When I plug it, on whichever USB 3 (or USB 2) port I plug it, it is not recognized properly.
I get an "unknown peripheral" in Device Manager.
(I've installed USB3 renesas drivers from Gigabyte website btw)
How can I make my USB Hub work ? :/
I tested it on another computer on windows XP it works just fine
Thanks a lot :(
EDIT extract from D-Link's website :
Quote:
The DUB-1340 works with any Windows, Mac or Linux system. Your computer will detect the USB hub and install the drivers automatically, so there is no need to carry the drivers on a CD or a memory dongle.
guys got some good news and photos , with Falkentyne's help found a software program that matches vcore , to the actual DMM test , Im using HWiNFO32 v3.72-1207 http://www.hwinfo.com/download32.html
see here in the photos my fluke meter to the program on the screen , yes volts are high im just testing llc now ;) but you can see its a perfect match
Im at 1.480 in bios with llc6 at the moment
My Meter
http://i84.photobucket.com/albums/k7...c/IMAG0662.jpg
My Screenshot with loaded prime and HWiNFO32
http://i84.photobucket.com/albums/k7...pc/desktop.png
My meter in front of program HWiNFO32 on the screen
http://i84.photobucket.com/albums/k7...c/IMAG0661.jpg
you have the B2 board correct?
I will try that out with my board and see the results.
EDIT: i just checked, and the B3 board, no go. Its teh same as easytune. Please compare your results against easytune from hwinfo to easytune. Actualy vcore is 1.279.
i hate hwinfo lol.
Pretty sure he's on a B3 board, not a crossflashed B2.
then hes the only one where software matches hardware ;) exactly
Pretty sure that my sensor readings are close to hardware as he and I have almost the same droop range, although I'm using a UD5.
Your droop with LLC disabled is like 0.025v lower than mine, just comparing software to software. (not sure about real_7's though).
with 1.25v BIOS, I got 1.176-1.164v load with hwinfo32 with LLC off. Yours is 1.138'ish (guesstimating since you had 1.23v set, and 1.18 showing at sensors load).
I'm wondering if maybe GB changed some hardware on the board earlier or later in the B3 rollout? Some people had stickers over the revision, while some had the new revision directly painted over.
Real_7, when did you get your UD7 B3?
And Sin, when did you get yours?
I got mine early in march from newegg (UD5, though)
Sin, I did mention this before,Quote:
Originally Posted by sin0822
But one way to tell if the vcore is .1v higher or not is from the temps.
That will also increase the power draw as well.
.1v is about +10C on air (maybe slightly more).
But two cpu's on two different systems can't be directly compared, since default VID can also play a role in the temps (my 1.21v chip is 5C hotter than my 1.235vid chip at the same voltage (vid at default speeds).
Er, scratch that last comment about your 1.23v drooping to 1.108; you had that on an earlier test.
But I just looked at your droop....you're going from 1.23v BIOS to 1.218v at load with NO LLC....that's like your board has automatic LLC7 or something. But the sensors won't show it....
Sin....are you SURE your DMM is calibrated correctly?
Can you read the DDR3 or the VTT to compare?
It seems like maybe your board is defective or some calibration is way off....
Owi,Quote:
Originally Posted by owikh84
Can you check your idle/load vcore with LLC9 at 4 ghz and 1.25v set in BIOS?
I want to know if you're rising at load or drooping
You should be rising considerably at 1.25v for sure.
so far it seems like this board needs more vtt than my M4E, but other than that everything's good.
falk- with LLC disabled, i was getting better load voltages and idle voltages than levels 1-3. It is pretty odd, i noted that.
I got my board, and it has UD7-B3 directly on it, and i received it last week. No stickers.
The initial boards could have just had the chipset just replaced, but then again that is what all these boards have. I have two DMMs, but cost me 50/80 bucks each, ill make some pictures for you, but you guys should measure off the MLCC cap ont eh back of the board, like Tin did here:
http://xdevs.com/kpc/sandy/ud7_osc/ud7_sense.jpg
LLC#1 on mine had more vdroop in prime than disabled (1.15x v vs 1.164v). But had an higher idle vcore in the BIOS....
anyways I revisted with two DMMs this time going at teh same time. One that I was using was showing a .03v difference, an the better one i , its a $80 one says its more like a 0.02v difference. i will redo my charts, but this has become much better. I don't know why yesterday i was seeing 0.05v difference :( I will fix up my charts. But its not dead on at all, there is no wway he got it to be exact, at least, at a minumim without any LLC there is a 0.015v difference.
BTW both my meters are rated for a 5mv difference in accuracy. somthing like 5%
yea i told you LLC disabled seems to have LLC there! LOL
Really? I've seen several people reporting stickers, like a lot...Quote:
Originally Posted by sin0822
And a post over at Gigabyte .de forum talks about it too, don't remember what was said though. Also, see Dino's comments here about labels (Which makes sense and was my expectation - same as with 5870 OC/SOC cards, labels on OC cards covering SOC PCB)
http://www.xtremesystems.org/forums/...d.php?t=268078
interesting I though tthey all had UD7-B3 written on it, mine has UD7-B3 labeled on it, no sticker. The box is different than teh first, but the board has P67A-UD7-B3 on it directly.Quote:
Originally Posted by Lsdmeasap
thanks ed
Now, as far as models go I am not sure where the line is drawn on new PCB/Old PCB with labels, but I do know some of the older PCB's were used on some and labels applied, but not sure if that applies to all models or only certain ones?
Rev. 1.0/1.1 was also a part of the discussion in regards to this same subject, here is the .de discussion I mentioned (Can't find the one about labels/PCB though)
http://translate.google.com/translat...%26prmd%3Divns
I thought UD5/UD7 are stil revision 1.0? my UD7-B3 is rev 1.0 with new PCB label. There has been no hardware change, i paid close attention to resistor ratings around the PWM and they are 100% the same as teh old UD7-B2.
ED- its hard for me to read that type of english! LOL
Sorry about the last PM, i didn't want you to give your hopes up like i did.
I am seeing the same differences from hardware to software as this guy is now:
http://lh4.ggpht.com/_CJpbLrog_T0/TY...age%5B4%5D.png
he is on B3 i still cant find the LLC that i need, and i would like stasio, dinos or lsdmeasap tell me if the crossflashed is better than F8 for my board or its the sameQuote:
Originally Posted by Falkentyne
plz ?
cheers to all
Sergio
1.45 bios set on LLC5 i getQuote:
Originally Posted by sin0822
1,452 hwinfo/cpuz/ET6 and 1,454 on chokes
i can game whatever but prime fails...bsods 101 and 124 its all my world now lol
So, andre, your DMM readings are accurate with software (at load) just like The_real_7's readings?
I am using LLC level 7 with 1.42 set in the bios, that gives 1.44 idle and 1.416 under load according to hwinfo32, testing under the latest build of prime 95 64 bit on a UD7 B3 (f3 bios). I tried using LLC 6 like most recommended and did not like how it was under/over volting at the wrong times, I was getting bsod 124 left and right, so I tried level 7 and I was able to reduce vcore quite a bit. I still think it can go a bit lower as well.
[IMG]http://i51.tinypic.com/23mwygp.png[/IMG]
I tried to cross flash my UD7-B2 but it wouldn't work? Id like to know how to do it as well.Quote:
Originally Posted by andressergio
Try LLC level 6 or LLC level 7 those results are justing showing DMM and software. Its has nothing to do with out board, but that board uses the same hardware monitoring chip.
is HWinfo showing the same voltage as easytune? easytune has real time voltage monitoring too.Quote:
Originally Posted by l0ud_sil3nc3
why don't you go buy a 20 dollar DMM? That way you can tell us how your UD5 handle LLC.Quote:
Originally Posted by Falkentyne
As i showing in the picture before, here is idle is the lower voltage, and load is the higher voltage, this is LLC level10.
See there still is a difference, but its minimal, my other older DMM and its cheaper too, it showing 5mv more than this DMM.
Also as i showed before hwinfo32 the version he posted for download=easytune. You gotta set eh poling interval to 1second.
honestly i forget what voltage i set, but this is level10. I think its either 1.3v or 1.32v i am unsure. I am redoing my tests with this DMM, even though its showing the same trends.
correct G :up: this is on my UD7 B2 with last F8 matches your volts i dont know if this is stable or not on this BIOS i was stable on F8f but volts match as you saidQuote:
Originally Posted by l0ud_sil3nc3
http://img864.imageshack.us/img864/731/llc7ax.png
Uploaded with ImageShack.us
seems u enlighted me guys Thanks loud and sin LLC7 seems to work DMM Shows LOAD vcore
http://img823.imageshack.us/img823/2...primellc71.jpg
Uploaded with ImageShack.us
Your readings are pretty much like mine, i didn't think there woudl be much difference between the two different revisions, and there really isn't.Quote:
Originally Posted by andressergio
Does BIOS f8f for the B2 have the 10 levels?
I did test against F3H and there is a huge difference in the difference between what software and hardware show. I think they improved it, i did tell GB that there was a huge difference that could potentially degrade chips with LLC level2, and i think that is why they incorporated all the levels, so that the user has a bit more control. OR maybe its that these levels are closer to software readings.
I am going to redo my chart tonight, so i will have finalized readings for you guys hopefully by midnight, if the GF allows.
BTW if you are wondering about the DMM, mine at least about the accuracy its as follows:
400mv +/- 0.5% of readings, +/- 4 in the last digit
4v-40v,400v +/- 0.8% of reading, +/- 3 in the last digit
600V +/- 1.0% of reading, +/- 4 in the last digit
its auto ranging so there is no calibration(modern day DMMs don't need calibration, some do most do not)
My original stuff was done with my very old trust velleman DMM, and its not auto ranging, its just i didn't have batteries for my better DMM, but i went out and bought some. anyways they are pretty damn close, but my old one shows voltage higher than my other.
yes exact the same :up:Quote:
Originally Posted by sin0822
sin can you confirm this ?
ALL levels of LLC when you are gamming (tired all games) vcore increases more than idle choke reads 1,477 gamming, load prime i show on the pics
so far tried 5 levels ALL do the same
3dm11/crysis2/moh/bc2/shift2 all load more acting like OLD LLC2
such a weird thing...
cheers :up:
Might get one next month. I want the one real_7 has. Over my budget right now since I have to buy my friend a UD4 and give him my first 2600k.Quote:
Originally Posted by sin0822
Andre:Quote:
Originally Posted by andressergio
Your vcore at load in prime is 1.438, but in 3d game it's 1.477 ??
What about Super Pi?
Ill check it for you, but that is VERY weird. Can you tell me exactly what you want me to check?Quote:
Originally Posted by andressergio
Ehh fluke is ok, they are WAY overpriced. Extech makes really good ones, kiethly really makes some of the best. Fluke you just pay for the name, don't get me wrong they are GREAT meters, but for measuring DC voltage under 40v there is no difference, you could use a 10 dollar meter and it would give you an accurate measurement.Quote:
Originally Posted by Falkentyne
True RMS meters are also NOT needed at all and wont make any difference. All our measurements from DC are from a sin wave, so true RMS allows measurement of other types of waves, which is VERy important when measuring AC current, BUT you wont be doing. Fluke meters also come with 10 year warranties, but DMM tech has come such a far way you can buy a 10 dollar one and when it starts to goto :banana::banana::banana::banana: you can buy a new one. So don't get me wrong fluke makes great DMMs, but not necessarily any difference between its readings and a much cheaper brand. They prob will last much longer though. The funny part is that fluke says people love their meters is because they measure both AC and DC voltage and frequency, well i am sorry but a 10 dollar DMM does the same now(well not frequency).
Something to look for in a meter is accuracy like i posted above, those specs are available for every meter you buy.
Id buy from these companies though not in anty order:
Fluke Extech
Klien
Kiethly
milwakee
Kiethly and milwakee being more expensive. The problem i have with very expensive DMMs is that you never use their other functions that come with that price. You could also buy a scope for $200 bucks lol and that would beat out any DMM, of course a scope takes skill you gotta know more than just stick it here and there.
just put the DMM on the choke and test VCORE while u game or 3DM11 you will see vcore goes upQuote:
Originally Posted by sin0822
test it im sure it will do the same
kinda worried of this as i game a lot...
naaa unpack da beast :DQuote:
Originally Posted by Skakruk
no man, if Dino said its fine then its fine, i changed my statement because of that info. The PCBs had already been made and they used them.Quote:
Originally Posted by Skakruk
so during gaming your vcore raises? shouldn't it raise?
much better than my MIVE B3. It is recycled from my B2 board which is rev 1.02. it came back after 3 weeks together with TIM residue left on heatsink some more. plus a bent system panel connector pins. :(:mad:Quote:
Originally Posted by Skakruk
i think the question is for me ? :confused:Quote:
Originally Posted by sin0822
why vcore should go up when load prime goes down ? so weird
so crap man i would be ungry too...but im on South America so no chance to return notha...so far B2 runing strongQuote:
Originally Posted by owikh84
i have a b2 commin in soon tooQuote:
Originally Posted by andressergio
but i cheaped out n got a 2500k:p:
hopin these gigabut boards are good:p:
i don't think you will be disappointed is this your first Gigabyte board?Quote:
Originally Posted by zsamz_
nice zsamz :up: they are great and strong and also 2500k are greatQuote:
Originally Posted by zsamz_
have fun :up:
no lol check the signature he tends to destroy them :ROTF::rofl:Quote:
Originally Posted by sin0822
lol those are pre X48, i think GB did a turn around then.
Haha, I know what you mean, I have a tough time deciphering it sometimes too. I think only the boards with new codecs are rev. 1.1Quote:
Originally Posted by sin0822
No worries about the PM, I'm happy to get it so at least someone can share the pain with me :yepp: :up:
There is modified BIOS a few pages back (And on Hwbot), with B3 BIOS that has board ID Changed to B2, so you can flash B2 board to B3, then reflash both BIOS with official B3 BIOSes. If you want it, and can't find it let me know and I'll point it out for you.Quote:
Originally Posted by sin0822
Kevin, you're kidding......Quote:
Originally Posted by owikh84
Ok..you're not kidding... :(
Can you contact your credit card company to file a claim ? Because that's basically fraud...(after of course trying to resolve this with whoever processed your RMA).
They're supposed to ship you NEW boards, not frankenstein'd boards... :(
If I saw this happening I'd be on the phone with my bank faster than the speed of light....
i been seein good things with the 2500ksQuote:
Originally Posted by andressergio
didnt wanna try my luck on a 2600k yet
i bought 3oh6's board its ready for my phase:D
i'm goin straight to -55:D
lol even the x48 was crap:rofl:Quote:
Originally Posted by sin0822
i went through 6 gigabut p35s in a year and an x48 in a week:rofl:
i did have a p45 die:rofl:
not like i do it on purpose lolQuote:
Originally Posted by andressergio
but it is fun to watch a crap board die;) especially when it takes the ram with it:mad:
had 1 die @ stock in my sons rig lol
i was hoping for a gigabut free home:ROTF:
jaja this are stong boards and yes 2500k tend to OC very goodQuote:
Originally Posted by zsamz_
I crossflashed to F3h and update with F3 and so far it's OK.Quote:
Originally Posted by andressergio
Also updated Backup BIOS to F3.
Any time you can crossflash back to F8f,as per my post here:
http://www.xtremesystems.org/forums/...postcount=1672
Sorry Sergio I missed your question.
I crossflashed a few weeks ago, no problems at all, and yes it does seem like the newer B3 BIOSes are better in several ways (Plus I don't think B2 BIOS updates are going to continue, at least as often as B3 BIOSes are released)
anyways so i have some tips for some guys and a recheck on my DMMs. Seems like the one that was off really wasn't bad at all, they were hookedup to different ground. I don't know how it made a difference, but it did. And before hen my readings were very far from my newer readings, its because i used a different ground. You want to use teh 8 pin 12v cpu power connector as your ground.
Here are some shots of LLC level10 1.23v set. Idle is 1.282 and load is 1.324. Two DMMs is better than one! that is for sure a great backup of my readings.That is what i am saying falk, get two cheaper DMMs and if you want have them check each other, different manufacturers would increase the accuracy of your reading.
Ok that's more like it. So my hunch about the calibration thing was right. So you're overvolting by 0.016v instead of 0.05-0.1, much better.
The overvolting is still high at around 1.45v?
Anyways here we go, higher levels of LLC show different vdroop/vrise depending on voltage. For instance for highest gain at 1.5v and above you want to use level 8 and set a high voltage, Level 9 shows constant no change from idle to load and level 10 shows a droop at voltages of 1.5v and above. Below 1.5v LLC level 10 shows significant rise under load, at some levels like level 7 and level 6 1.23v might show a drop while higher voltages show rise under load. Weird trends, probably how the PWM handles all these levels dynamically. Anyways right here we have great ability for voltage regulation, you can spot them throughout, for instance sometimes like for level 9 at 1.5v i get a rise from idle voltage and then the VRM settles down and i get a nice stead 0.001-0.002 rise over idle. This happens at level 7 and 1.45v, under load it went from 1.480v idle to 1.474v to 1.483 and then to 1.487 and then stuck at 1.487 for the rest of the test, it took about 30 seconds to change the voltages. My nice DMM has a ripple function and the ripple on some of these levels is almost non detectable, you need a scope. Obviously this being an analogue PWM it stabilizes voltage by leaving on the MOSFETs until it hits a voltage and then turns them off and then back on when voltage drops, so its pretty hard to read a steady voltage right off the bat, you need to wait a few seconds for the VRM to settle. But once its settled it is great at keeping at that voltage, of course this is a mix signaled PWM so current is under digital control and votlage is under analogue control, which means current is outputed based on algorithms and voltage is outputted based on error correction, amplification, and compensation. Enjoy the results guys, i changed my previous charts to match this one:
Loving those results.Quote:
Originally Posted by sin0822
Hopefully my CPU will be happy at 1.445v BIOS LLC5 @ 5 ghz.
Was funny when I did 1.450v BIOS llc4, and and the system instantly restarted on prime :P
NVM. restarted. Doing 1.440v BIOS LLC6 now.
Ok seems ok. Stopping it after 35 minutes.
Damn, now what level do us users that like the old Level 1 use? :D
Haha, that's too many choices :)
level 6 or 7 depending on your voltage :)Quote:
Originally Posted by Lsdmeasap
My voltage is dependant on the LLC :D
I'll play mess around with it, I was just playing above :) You're right, 6 looks about ideal for my normal voltage ranges with 7 being better for higher voltage choices.
I do actually like this implementation, it just looks daunting looking at all your testing results.
So have you heard if this is a test run of all these levels for the current BIOSes/boards only, or do they plan to leave them in from now, and and possibly add to other series like X58 too?
I have no clue, i think it shoudl be come permanent on every board, it will make them more attractive to OCers.
Nice sin0822 now it looks similar to my board as well at 1.5v+.
Too bad i can use LLC9 and LLC10 only for my CPU at 5.0GHz.
Prime is fine but LinX is no way to go more than 5 runs once it vdroops to 1.526v I will get BSOD 124.
Ok, is THIS ok now?
http://img841.imageshack.us/img841/7989/5000d.png
Not running ANY more prime at 5 ghz. Already degraded my chip by at least 0.01v in the last DAY. Was doing FINE at 1.445v and llc5, then today decided to restart (didn't even BSOD, just restarted, and I know automatic restarts are disabled, because I was getting hal102x.sys (X-fi) bsod's properly when I was having trouble with Dolby Digital Live, X-Trap and Ventrilo (not related to any OC) at BOTH 1.445 and 1.450v (the voltage in hwinfo32 seemed the same in both tests at load tho). So I did 1.440v and llc6 and went past 30 minutes and stopped when the third loop began. Temps got 3C hotter.
I'm not degrading my cpu anymore for this....
7 hours prime stable
http://i55.tinypic.com/v3odch.png
wouldn't even make three full passes and bsod 124 while running linx set to 4 threads:shrug:
What cooling are you using?
I may reapply the MX4 on my Ven-X a different way, but the last time I tried reseating, it made a whopping 1C difference...
water, I don't think temps are the problem, just more vcore:DQuote:
Originally Posted by Falkentyne
ouch you sure its degradation , cuase i can name ten things that make it seem that way , your chip gets hot Falkner , but that doesn't mean degrade. . and if you are correct . all chips have a break in period regardless where they will sway off a bit . . . your chip is fine my friend put back old bios you will seeQuote:
Originally Posted by Falkner
welcome zsamz , we just where about wondering when you where joining usQuote:
Originally Posted by zsamz_
are these boards good , hmmm let me tell youit feels like the DFI UT P35-T2R: Tweakers Rejoice again days , just don't use the newest bios first unless you wanna pull your hair out lol , see you soon old friend
yea I forgot the gigabyte days with him lol , feels like the crew is getting back together to slam another good board huh sergio :rofl: oh my , just let him know we will not tolerate gigabut jokes nomore lol he he heQuote:
Originally Posted by andressergio
Quote:
Originally Posted by stasio
Thx for the support Stasio.......however bclk
still reads 99.8 here......didn't load old profile... & flashed with Load Optimized Defaults??? Suggestions?? :shrug:
blck @ 100 work with xmp profile loaded-----Quote:
Originally Posted by Expat GriZ
same here mate blck not solved nor reset same as old bios...sorry stasio but i been checking for all weekend to say thisQuote:
Originally Posted by Expat GriZ
only ok is new levels but still deciding to flash B3 bios...
its like first bios you gotta change multiplier to 1003 Expat GriZQuote:
Originally Posted by Expat GriZ