Nice software workaround. Might give LinX a try then instead von IBT, even though AgendGOD can probably adjust IBT just as easily.
Printable View
Nice software workaround. Might give LinX a try then instead von IBT, even though AgendGOD can probably adjust IBT just as easily.
anyone have a link for F4J?
Ok, so far i am finding F4 official bios sucks. Failed prime after 30 mins, now back to F3, my 3.8ghz running prime 1hr and still going.
damn Gar, did you try F4J? I have not had a single issue with it.Quote:
Originally Posted by GAR
F4k is doing great for me at the moment. I got linpack stable at 3.7Ghz.
http://www.station-drivers.com/page/...a-ex58-ud5.htm
Can someone help me to bring my temperatures down?
I was doing 40C idle and 75 load, yesterday with F3 Bios.
I flashed in to F4R now and temperatures went up 10C, I dont know what the hell that Bios did though.
Please let me know, what I can tweak to bring the temperature down. Everything is Stock.
Are you running stock air? if so get water cooling.Quote:
Originally Posted by icecold316
The test are done with the latest F4 bios at aircon temp of 22d.
First I started to test @ 3.6 (180 x 20)
Vcore @1.2875
QPI @1.2
DRAM @1.64
Temp idle:
http://i204.photobucket.com/albums/b...3Dec282040.gif
Temp load:
http://i204.photobucket.com/albums/b...2Dec282009.gif
Temps are pretty good. Now straight up to 4Ghz.
Using the same setting as my previous UD5:
Vcore@1.38125
QPI@1.34
DRAM @1.64
Temp @ load
http://i204.photobucket.com/albums/b...1Dec292200.gif
Yes. Is stable @4ghz too:
http://i204.photobucket.com/albums/b...2Dec292231.gif
http://i204.photobucket.com/albums/b...2Dec292357.gif
For your reference I posted my link on my previous UD5 thread:
http://deluxeforums.hardwarezone.com....php?t=2174311.
Do note that the test of UD5 was done outside the case whereas the extreme was in the case:
http://i204.photobucket.com/albums/b...7/DSC04169.jpg
but does it pass Linpack?Quote:
Originally Posted by pa9797
http://www.fileden.com/files/2008/12...3/LinX_0542.7z
any reason to go f4m over f4j?
Quote:
Originally Posted by safan80
If it does what you need it to do on a daily basis, whether it be folding, rendering, benchmarking, gaming, surfing....putting a system through that much stress to prove stability is for a niche crowd saf; it is far from a standard and never will be.
Haven't tried yet, but from what I gathered here:Quote:
any reason to go f4m over f4j?
- faster init
- some fix related to virtualisation or something
- supposedly better (ram) performance
So I'll try F4M next, but F4J is pretty good already.
Guys,
anyone got a link to F4M bios for the EXTREME? Hicookie's link is down. Also is there some kind of overview page where I can see/download all the versions? I am getting kind of confused here (too many bioses lol).
http://www.xtremesystems.org/forums/...postcount=1031Quote:
Originally Posted by jcool
right there :)
Trying to achiev 20k in everest but dont know how i should come there, is it my memorys that suck with timings on higher speeds ? Trying F4k right now
http://img515.imageshack.us/img515/8512/20756220fy5.jpg
http://img515.imageshack.us/img515/2...pg/1/w1280.png
any reason you are using that clock generator?
what is your uncore at
Are you basing 'this stresses cpu more than that' etc purely on temps?Quote:
Originally Posted by Hornet331
I hope not :p:
Quote:
Originally Posted by Xello
No its based on my observation -> prime 5h stable (blend/smal) -> linpack crahed after 30min. :p:
Prime doesn't stresses the memory part enough for Ci7.
If linpack can run for several hours everything can run. :yepp:
Prime was good for C2D but it's not enough for Ci7.
I think from now on i will skip prime and just go with linpack + memtest. :yepp:
Try running DDR3-2000+ with at least CAS8-8-8 ;)Quote:
Originally Posted by ante_ante
If you don't have D9GTR/GTS or HCF0 you can forget about it altogether. D9 will need 2V+ to do it, if you are prepared to risk that is up to you (and depends on your cooling).
Also raise your uncore to around 4Ghz . CPU Mhz/QPI speed is irrelevant here, Uncore is key.
Quote:
Originally Posted by icecold316
Yes, I am using the stock cooler for now. But I dont get it, why a BIOS update raise the temperatures by 10C.Quote:
Originally Posted by safan80
I will be getting XigamTek Heatsink in couple of weeks, as they said the 1366 Adapter will be availabe in newegg in 2 weeks time.
That only means that linpack is a quicker way to test though, you no doubt would have had prime error / freeze / BSOD eventually.Quote:
Originally Posted by Hornet331
That's good if it's true though. I was just saying because the way prime works it could load a cpu less with small FFT yet still make it hotter in the past.
I always thought for some reason that even though prime was only using 10-20% memory, that it was cycling through the memory or something, like memtest does, to test out all of your ram :shrug: Not quite sure how it works tbh.
Passed 50 runs linX @ 4.2ghz with 1.37v today, took about 2hr 50 mins i think, so i'm going to run blend til tomorrow morning just out of curiosity. If no crashes i'll probably stick with linX too :up:
Flash back to previous bios, test temps, verify it's a bios issue?Quote:
Originally Posted by icecold316
Dont know what chip i got on my memorys. Maybe should try to check it.Quote:
Originally Posted by jcool
Running uncore 18 on that pic.
Hi all,
Compliments of the season to you all - now for the question.
Just recently put together a system with Gigabyte EX58 UD5+ Corsair dominator 1600 ram And I7 920 chip all runs fine But .....
When I boot up either cold or restart the post screen - ie the first Bios screen never shows up - I have tried the setting in bios for the splash screen on or off plus pressing the Tab key but it refuses to show.
The only way I can get it to show is if I go into bios and load optimized defaults and then it shows splash screen and I can press tab key to see post screen. Anybody else seen this or know the cause of it?
I have loaded the official F4 bios to fix the resume from sleep issue with str, but I had the above problem before this anyway - any ideas anybody?
Would be grateful for input on this one
Ps the problem even happens at stock speeds - only running at 3gig at the mo - with turbo mode ie only 3 gig on 21 multi if that makes sense.
Mark
Temperature came down to low 42 to 45, after changing the QPI VTT from 1.235 to 1.215Quote:
Originally Posted by Xello
When I loaded the bios first, easytune showed 1.43........ that is everything else on stock.
do you guys get higher or most stable overclocks from f3 to f4j on the extremes?
Like any stress testing screenies with difference?
Does this happen all the time or just when you cold boot?Quote:
Originally Posted by tipes
If it happens all the time, are you running SLI or Crossfire? For example, running Crossfire and having your monitor plugged into the wrong card will cause the problem you described: you can't see the POST screen and get no video signal until you boot into Windows. Connecting your monitor to the other video card will resolve this.
If you're NOT using multiple GPU's, then I'm not sure what the problem is and would need more info before trying to determine what it could be.
Fully agreeQuote:
Originally Posted by ReverendMaynard
HiQuote:
Originally Posted by yenniedn
Thanks for your post. I am only using a single 8800gts graphics card, and yes this problem happens all the time. I have also tried both connections to the graphics card with no joy. Hope I can find out what is causing this as it is slightly irritating. If you need more info - let me know.
Mark
So what version of EasyTune 6 really works with X64 Vista?
the system doesn't even have to be close to stable for web surfing, or benching. Linpack is made by intel for intel cpus and it will grow over time. I have games that stress the ram more than prime, or OCCT does. Your OC could do what you need it to do for a limited time, but if you start getting data corruption or hardware failures all that work is for nothing. If your OC is not rock solid you could end with data corruption in anything you esp. if you use the onboard controller to control your hard drives. Hey it's your RM data and not mine so it doesn't matter to me if your OC is just prime or OCCT stable.Quote:
Originally Posted by ReverendMaynard
the one not released yet ;) :rofl:Quote:
Originally Posted by BulldogPO
I ditched F3 because the F12 boot menu didn't work when OCed.. with F4J it works fine.Quote:
do you guys get higher or most stable overclocks from f3 to f4j on the extremes?
Like any stress testing screenies with difference?
Happy new year guys :)
My linX 3hr~ stable settings yesterday failed blend in 9hr 32mins. I guess linX isn't infallible, or you cannot use it as a quick test, certainly not 5 or 10 runs- you need to run it as long as blend.
All we were testing for prime, occt and Linpack in the socket 775 discovered that Linpack quickly discovered the errors in vcc and vdimm, but NO the errors in MCHv. I think things like that happen too in 1366. It is necessary to do tests with everything.
Perhaps one of the most interesting problems that introduces Linpack is the effect of the heat. My hard drive is still dead, but my system is resurrected. Unplugged the hard drive, and left the system at idle for several hours (accidentally), and to my surprise everything is back to being as before:D.
I have made many tests to understand exactly what happened, and I discovered that switching off only one of my three Noctua that blow to the board, i need automatically to increase a notch the IOHv to pass Linpack test. Linpack, because of the enormous stress of heat it produces, and not only because of the electrical stress, has shown me a direct and immediate impact between a couple of degrees plus or minus and the stability of my system as never before saw overclocking socket 775. However, it is very difficult for me to know whether the main reason for this discovery are my OC values, or is the x58 architecture, or is the Gigabyte Extreme ...
You reply too me like I don't know wtf I'm doing when oc'ing lol. I don't give a rats ass about corrupt data, I've been overclocking for 10 years, I've learned how to effectively back up my data. I'm just glad that I'm not an OCD stability guy. Life it too short for that bs.Quote:
Originally Posted by safan80
who said its a quicktest anyway. :p:Quote:
Originally Posted by Xello
QFTQuote:
Originally Posted by Xello
Linpack is awesome for speed at quickly pinpointing OC points that are a notch or so from stable, much quicker than prime, but if going to equate to prime 12 hrs, it has to run for many hours as well.
I have yet to see a 12 hour prime blend or small ffts oc on mine fail linpack, even overnight linpack runs, though others claim to. Also on my rigs, have not seen a 100-200 pass linpack run AT MAX MEM setting fail prime yet, though others have, and no doubt possible. A 200 run linpack on max mem takes about 6 hrs. On core i7 need to run two version to get near max mem though, one at near max, other on "infinite" run of 256 or 512. The newest version use two at half max...havent played with it enough to know how it compares to max mem.
I have never had a crash from bad OC after being 12 hrs prime stable, so I will continue to use that as my gold standard, though that may change as i use linx more. Though it is not a bad idea to run at least short runs of both, after being stable in one, as they test in different ways and one or the other may pinpoint a weakness better in your particular rig.
Quote:
Originally Posted by vega22
Prime stressed the MCH even less then linx, the only test program that i know that stressed the MCH to some point was memtest.
Thats why i still keep memtest in my test setup. I tested my setup for 7h linx stable and have no problem so far.
Sure there are people out there that run 12h+ prime linx etc. but what they get from that i dont know. :p:
My system runs 12-14h a day when im home but far away from the stress levles linx puts on the system.
For myself i consider a system stable after 6h+ of linx, everything else is a waste of time in my eyes, cause it adds nothing to stability at my usage behaviour. :yepp:
But the discussion about stability is as old as ocing itself.
I´m not the only one who has seen fail OCCT and Prime in 5 minutes, while Linpack was over 50 runs, everything because of the MCHv. But Memtest can be as effective as Prime or OCCT for MCHv. I have not tested that.
It´s impossible to reach an agreement between a game system and a workstation talking about stability. Everyone do what the peace of their minds need.
What vega22 said is perfectly true.Quote:
Originally Posted by Hornet331
Prime blend --> custom --> check box "run FFT inplace" is the quickest and most reliable way to determine a mobo problem. Linpack won't detect these unless they are severe, as for memtest.. well, if memtest freezes due to a failing MCH, your rig is unstable to the extent that Windows won't even start to boot 99% of the time. :rolleyes:
HI GUY ,F5A BUG BIOS ,CAN'T 2000CL7 F4J F4M PASS OK
http://pic.kaixin001.com/pic/photo/4...4_38474280.jpg
2000 CL7 PASS CL7-7-7-21 1T D9JNL
http://pic.kaixin001.com/pic/photo/4...4_38474999.jpg
the problem is, theres no MCH voltage any more. :p
It was true for the C2, and for C2 is till would use Prime as my main stability testing tool but not for Ci7.
MCH voltage is gone now and instaed it is linked to QPI/VTT (Vdimm for L3 cache).
I booted windows several times and linx crashed after several minutes, and all i had to do was increasing QPI/VTT, so its not ture anymore that you can't boot windows with to low qpi voltage...
Many seem to forget that everything that was ok C2D is not valid for Ci7.
Anyway as i said befor how to test stability and even the basic definition "what is stable" is not defind by anyone, its rather a thing of personal reception.
I have my own guidelines and standards what i consider stable and how i test it, and i never had problems since 8 years now, so for my needs my methods are the correct way. :p:
Anyone can do as they please i dont stop anyone form doing things how they want, i just that for me some things that got mentined here are a waste of time in my eyes.
The one that came in CD of Gigabyte works fine for me in Vista Bsns 64Bit SP1 version.Quote:
Originally Posted by BulldogPO
Only problem is BCLK in QuickBoost looks funny ... 560Mhz, 600 Mhz and 640Mhz(It is probably showing Bclk x 4)
Love to know how you got that to work, tried every version and it still crashes on Vista64 SP1&2
I got the newest version working with a fresh install. Mine was corrupted somehow. All I did was install Vista Ult x64 sp1, then install easytune first off and it was fine. I even installed sp2 and it still worked.Quote:
Originally Posted by BulldogPO
-=Mark=-
Hehe...IOH is the x58 northbridge voltage...Quote:
MCH voltage is gone now and instaed it is linked to QPI/VTT (Vdimm for L3 cache).
Yeah but the MCH (memory controller) is now on the CPU not the NB.Quote:
Originally Posted by vega22
It's also difficult for me to believe my temps are holding me back, which are topping 55-60c, when i see people here topping 70-80c regularly, and 90c+ without fatal issues.Quote:
Originally Posted by vega22
Compared to prime, pretty much everyone, on the posts i was reading, on this forum and others. If you google you'll find many posts stating between 5 and 20 runs as commonly used to verify complete stability. Like i said yesterday (or the day before maybe) the reason i switched to it was because prime started causing crashes after huge periods of time, and i was after a quicker way to test. Linpack was a much quicker way to test, by popular opinion anyway. If i'm going to have to run even that for 6-8hr+ to get reliable results, i'll stick to what i know (prime) :up:Quote:
Originally Posted by Hornet331
ok I think I might still be drunk from last night but dont most mother boards for sli/quad sli need the cards to be on the top pcie lane and the bottom lane? I am trying to change some things on mine :) but the sli connector is short and it looks like both cards will have to go into the top and the middle lane. From reading the manual it seems like that is the way to do it. :confused:
I think I might leave it for another day cause the beers is not helping :ROTF:
gigabyte x58 extreme motherboard
You wouldn't want a card in the bottom slot anyway, if you can help it, it runs at half the bandwidth of the top + middle slot (x8 compared to x16+x16)Quote:
Originally Posted by bench2die
It´s important to note that i have spoken of thermal issues only in relation to use of a very strong test as Linpack. My cpu and mobo temperatures are good, even by air. But, as i´m experiencing, the key is the union of heat stress plus electrical stress. I think that is not important 10 degrees at 3.8Ghz, 180 bclk, full load Prime, but maybe only 3 degrees at 4.0Ghz, 211 bclk (my settings), full load Linpack. Comes to my mind the word "electric fatigue" or, better, "PCB fatigue".Quote:
It's also difficult for me to believe my temps are holding me back, which are topping 55-60c, when i see people here topping 70-80c regularly, and 90c+ without fatal issues.
The key is that with Linpack we can achieve very quickly (generally) a system full Prime95 stable. But is not so fast to reach a system Linpack stable (although this is faster than achieving a stable prime95 system using only prime95). The problem is compounded if Linpack from a certain OC level begins to introduce abnormal errors due to its enormous and abnormal power.Quote:
Compared to prime, pretty much everyone, on the posts i was reading, on this forum and others. If you google you'll find many posts stating between 5 and 20 runs as commonly used to verify complete stability. Like i said yesterday (or the day before maybe) the reason i switched to it was because prime started causing crashes after huge periods of time, and i was after a quicker way to test. Linpack was a much quicker way to test, by popular opinion anyway. If i'm going to have to run even that for 6-8hr+ to get reliable results, i'll stick to what i know (prime) :up:
Pass, that's out of my field even coming to guess work :DQuote:
Originally Posted by vega22
Hmm, there doesn't seem to be a way to define exactly how long a linpack run = how long a prime95 run though. I've read for months and months that it beats out prime hands down and will do in a very short time what prime will do in 1-2 days, i'm just saying that in my opinion everyone that talks like this is hugely exaggerating. Unless my one test yesterday was a fluke :p:Quote:
Originally Posted by vega22
I also forgot for a while but my old QX9650 which was 26hr~ blend stable failed IBT (same thing as LinX really right?) in like 10 minutes. Gah why is everything so f***ing complicated.
Sorry for being offtopic here guys, but I just wanted to tell everyone that newegg has the new Giga boards.
UD3R for $199 http://www.newegg.com/Product/Produc...82E16813128375
UD4P for $259 http://www.newegg.com/Product/Produc...82E16813128374
UD3R finally broke the $200 mark for X58 mobos :)
oh "cheap" X58, even with Ultra durable 3. :)
THe DS4 has no ultra durable 3 (well at least offical :p: ).
Showing $209 still for me. NewEgg has had these boards for a couple weeks now. Couple other boards coming out soon from other manufacturers, definitely some interesting times. Egg also has the -DS4 for $245.Quote:
Originally Posted by Khalmyr
Heh, doesn't it seem like they have a bit too many boards out that are a bit too similar?
Yeah, there are a few subtle differences but I donut understand why GB needs to have six X58 boards on the market. It's important to note that the budget boards only support Crossfire. I wrote this up a few weeks ago on another forum:Quote:
Originally Posted by Xello
UD3R - 4 DIMMs, 16GB DDR3 2000, Crossfire Only, no eSATA, 2 PCI-E @ x16
UD4 - 6 DIMMs, 24GB DDR3 2000, Crossfire Only, with eSATA, 2 PCI-E @ x16
DS4 - 6 DIIMs, 24GB DDR3 2100+, Crossfire Only (3-way), eSATA, 12+2+2 phase power, 3 PCI-E (16/8/8)
All of the above have ALC888 audio, Realtek 8111D 1000Mbit LAN, and 1394. With the above boards, there are some subtle differences relating to number of SATA ports, legacy PCI slots, etc.
UD4P - 6 DIMMs, 24GB DDR3 2100+. SLI & Crossfire (3-way), 12+2+2 phase power, 3 PCI-E (16/16/8) with ALC889A Audio, 1 x 8111D LAN, and 1394
UD5 and the EXTREME are the high end boards with 12+2+2 phase power, SLI+Crossfire (3-way), ALC889A, Dual 8111D LAN, 1394, etc.
The only board that has an onboard TPM chip is the UD4P. All six boards have (fake)RAID 0, 1, 5, 10, JBOD
Is there any confirmed success using the pencil vdroop mod on either the Extreme or the UD5?
Has anyone done a before and after comparison?
Just curious if anyone else has had simular results in updating the Bios on the UD5 to the latest F4.
When i got my 920i/UD5 on release i could manage 185x21= 3885mhz @ 1.3250v ..... I wont go over this voltage, but no matter what i tried, how low i set the multiplier/uncore etc.. I couldn't get over 185 BLK
Just updated to this F4 Bios.. bloomin brilliant!
Not only can i finally get into Bios Instantly!! But im now using 200x20 = 4Ghz @ 1.3250, from a quick look 208 BLK seems to be the limit, and i cant seem to use speedstep once above 200x18.
Voltages i've messed with is:
Vdroop Enabled | 1.3250v | QPI/vtt 1.30 | Speedstep = OFF
Results:
200x20 = 4Ghz | Uncore 3400mhz
Akasa AK-967 heatsink | Idle = avg. 33 | Load = avg. 69
First Pentium since a PI 166mhz, very happy indeed with this, anyone else got a similar boost from F4 Bios?
(Edit: Just looked back, people mentioning f4k/j bios? i just used the update bios from web program so i've no idea).
@fiskov, if you read back a few pages to when F4 BIOS was released, all it did apparently was messed up most people's OC's and many went back to F4j or F4m.
You can't install 6 sticks of the Corsair Dominator DDR3 on the Gigabyte X58-EXTREME board can you? It doesn't look like the one in the blue slot closest to the CPU socket will fit because of the NB cooling. I've got the board already but I haven't taken it out of the package yet. This may help with my decision to go with the eVGA X58 SLI instead if it means I can't use all the memory slots on the board. You'd think Gigabyte would think of things like this and wouldn't try to put such crazy cooling on the NB when the NB doesn't even get hot in comparison to previous gen boards.
Guys, any news on Giga's utilities on Vista 64?
Whether or not you can do 6 sticks of RAM will depend on the HSF you use. My TRUE 120 will not allow for a 12GB setup if a fan is used on the side of the TRUE closest to the front of the case. I have my TRUE in a push-pull configuration and the only way for me to go 12GB is to lose the front fan (or get shorter ram).Quote:
Originally Posted by burningrave101
Well...Numbers:
19x211/Qpi x36/Uncore x16/Memory x8/HT off/Turbo Off
Bios F4m/CPU Clock Drive=700mV/Pci-e Clock Drive=700mV (more stables than 800/900)
Mems: 8-8-8-24/subtimings Auto
7-7-6/7-7-7/6-5-1/6-6-1 (settings imported from F4j -a little bit faster than F4m defaults-)
Absolute minimum voltages in Bios:
Cpu=1.38750v
Qpi/Vtt=1.40v
IOH Core=1.28v
Vdimm=1.66v
(everything else -> auto)
Less than impressive, but is not too bad:
http://static.zooomr.com/images/6640314_30101c30d1.jpg
Ultra Linpack stable (12h +), with these temps:
http://static.zooomr.com/images/6640281_2e5322187d.jpg
Only a few degrees more in mobo and ioh -> freeze
EDIT: my batch CPU is "A"
Hi,
I am hoping someone can help.I cannot get bclk 170:mad: and above to fully post (system powers on but keeps restarting until a failed overclock is displayed).I have tried many combinations and voltages to no avail.Below is what i have tested:-
bclk 170 x20
qpi/vtt 1.3 also tried a bit higher
cpu vcore 1.3 also tried a bit higher
dram 1.6
mem multipler x8
unicore x16
qpi x36
All other settings on auto for now
bclk 169 with turbo on is fine (same as above except lower volts ,vcore 1.28 and qpi/vtt 1.28.
Rig:-
3 x 2gb corsair cl8 1600mhz 1.65v
i7 920 batch 3838a820 rev c0/c1
true black with Scythe S-FLEX 1600RPM
gigabyte ex58-ud5 f4 bios (also tried f3 and the new f5a beta)
corsair 750tx
4870 512mb
linksys pci wireless card
1 x VelociRaptor 300gb
1 x wd 750gb
1 x samsung spinpoint 1tb
1 x samsung sata dvd writer
silverstone fortress ft01
This was a treat to replace my old e6600 rig (running @3.6ghz).I thought i could get at least 3.6ghz so i am a bit disappointed.Have i hit the limit of this cpu or is it the motherboard?
Many thanks and a
Happy New Year!
Tony - Try these settings, once you are stable, work at bringing down the voltages, particularly vcore, qpi/vtt and dram.
Based on the restart problem: f4 bios has been giving lots of folks fits - I would get ride of that first - most folks are running f4j or m.
Thanks for the reply, i will try those settings,Where can i get that bios version you have metioned?
I will hunt down the link for F4j, one other thing once you go beyond qpi and dram at 1.35v/1.65v you need to keep them within .5v of each other to prevent damage (so we are told). Also FYI intel max v's (I am going .1 over this on qpi and dram at times - taking up the intel engineering margin).Quote:
Originally Posted by tonytiger
Tony - I just sent you a PM.
Tony - - link to some UD5 bios are at - (some folks are having success with f4k also) -
http://www.station-drivers.com/page/...a-ex58-ud5.htm
Thanks for link,it appears i may have my vtt/qpi voltage set to high.I have reduced this to near default (1.18v) to start with ,still using f4.I have managed to boot into vista now.I have also tried bclk 171 but i had a hard lock when trying to post this message(hard disk led on).I must say s775 was a bit easy compared to this although i have been out of the loop for a while.Looks like i have steep learning curve ,but all part of the fun!Quote:
Originally Posted by SteveRo
Should i be able to reach 3.6ghz?
The official F4 bios is not going to help you at all Tony. As mentioned before try the betas, I think the most popular is F4r for the UD5 (anyone?). I know I wouldn't change my Extreme from F4J without a very good reason.
Quote:
Originally Posted by fornowagain
I did try f4f and f4l with the same settings but i still had the multiple restarts which resulted in the overclock failed message in the bios.As of now i have:
bclk 170 x20
vcore @1.2875
vtt/qpi @1.18
cpu pll 1.84
qpi pll 1.14
ram @9-9-9-25 1.66v
qpi/unicore and mem set to auto for the momment.
All cpu features of apart from HT.
Just ran a 1 hour test OCCT ( i know this wont prove stablity i will be using prime 64 and linx to fully test)
cpu idle 34,31,35,35
Load around 63,62,64,64
This will be my starting point to fully test stability.
Your is the UD5 right? Some of those betas were as bad as the official F4. I was pointing you towards the equivalent of the Extremes F4J, which I thought was the F4R? Honestly you're on to a looser until you get a stable bios, like night and day.Quote:
Originally Posted by tonytiger
Thanks i will try f4r to see if this board is more stable.
I am now using F5A bios, i have gone back to the stock cooler, as my Thermalright was not making full contact with the IHS, horrible contact. Anyway i have a Noctua P12 on the way, using stock cooler at 3.36GHZ with 1.120v. Nice and stable.
Have always worked for me, don't know why you guys are having problems :confused:Quote:
Originally Posted by VIctorj
Hi Gar,Quote:
Originally Posted by GAR
Have you tried lapping the Ultra 120 ??? How are you finding the F5A bios for stability/overclocking etc
Will have to check my 120 extreme to see how it is sitting - my idle temps at stock with turbo mode enabled are 36.31,32,30 room temp about 19c
prime temps at load 62,58,58,57 at 21 multi 2.8gig
Mark
My True Black seems ok ,but it does move if i twist it quite hard.I might try the washer trick at some point.I always liked Gigabyte Boards but this e58-ud5 is going to take time to tame,i loved my last board a p35-ds4.
again?Quote:
Originally Posted by GAR
i thought you leaped it already? :)
I lapped it, thought it was making good contact, BUT, after taking it out to reseat because of stability issues, i saw that contact was poor.Quote:
Originally Posted by Hornet331
is f5a still the latest bios for ud5 ?
Yes.Quote:
Originally Posted by freshy
Tipes and Gar - 965 at stock with c's and EIST on - at idle - I am at 30C (open case). Room temp is about 19C.Quote:
Originally Posted by tipes
(during oc - I send this guy to 100 (short duration) almost every day).
I am still having problems,I have tried bios f4k but i still have the same problems:-
I can post normally up to bclk 172 ,anything over and i cannot post normally.System restarts a few time before displaying the overclock error.
BCLK 171 and 172 are untstable:mad: anyway with hard locks in vista and even in the bios.Turbo on/off or ht on/off makes no difference also using more voltage on the below settings makes not difference.If i cannot boot above 172 i cannot test!
f4 offical bios. tried f4k,f4l and f5a
I am now running 21 x 170 3570Mhz
vcore 1.2875
qpi/vtt 1.18
qpi pll 1.84
cpu pll 1.14
ioh core 1.12
ich i/o 1.58
ich core 1.14
dram 1.66 manually set 8,8,8,24 2T
qpi,unicore and mem on auto (make no difference if i set x36,x16,x8)
also cpu clock drive set to 800mv pcie set to 900mv
all other settings on auto
All power saving options off.
I have run linx on the above settings for 2 hours.
Does anyone else have trouble over bclk 170?I must have a bad cpu or motherboard.I think i might try another board once the chipset matures , there will be some better board released i am sure.
tony - maybe try to isolate the problem?
Turn the mutli's for cpu, qpi, uncore and ram as low as you can - then start raising block - you shoud be able to get to 200 or better!
turn multi's to -
cpu - x16? - if it will go this low?
qpi - x36
ram - x6
ram timings on auto.
To make sure voltages are not your problem - set voltages:
cpu - 1.4
qpi - 1.34
dram - 1.64
make sure all the c's and eist are off.
maybe start with HT off
start at a bclock that has worked before and move up from there.
Once you know your mobo's max bclock you can move on to memory or cpu ocing.
I word on memory ocing - I am finding that memtest x2 successful pass (no errors) in insufficient to find max memory oc (speed and timings).
I wonder if this is because of the new IMC being on the chip.
What are folks using to determine max memory oc?
I've never used memtest for anything other than verifying stability @ stock (ie if your dimms don't pass memtest @ stock then it's RMA time)
In my experience you will always get errors in memtest on overclocked sticks, even if they're 100% stable in prime etc. Maybe this is not the case, but it definitely has been for me the past few years.
Quote:
Originally Posted by SteveRo
I have already tried cpu x17,x18 and x19 ,qpi x36,ram x6,unicore x16,17 and 18 ,these settings will not post correctly ,get the restarts then the overclock message above bclk 172 .I have also already tried those voltages they make no difference.I can't understand how other people have easily got bclk 180+.
I have noticed the auto setting are qpi=36,ram=8 and unicore=20.All cpu power management is always off.172 is the max i can get ,its not stable in vista anyway (freezes) also freezes in the bios.
I have just accepted i have a bad board/cpu.I might mess with the corsair mem sticks later (swap them over try dual and single channel etc).Also i might strip the whole lot down and run outside the case with 1 hard,video,ps2keyboard and mouse and 1 stick of memory.I guess i am unlucky this time:mad:I have spent about 20 hours on this and never had such a hard time overclocking ,i started with a vanilla pentium mmx all those years ago.I used have a mach2 - loved that and loved the massive overclocks.
Thanks for your help i do love this forum!
tony, there are some cpus that have low max bclk's, and you may just be unlucky....I did not read all your messages, but I assume you hit load optimal settings and save after each bios change to get the settings of the new bios. Also, I would try running with 1 stick of ram, if does not work, try different 1 stick of ram. Earlier I had 1 stick ram that was defective that was screwing up my OCing. Also future bioses may help. I am stable at bclk of 200x21 and even 205x21 with prime on F4j and F4m (extreme), but on bios F4, I cant get any higher bclk than 170's, so something screwy in F4 bios on mine.
Tony - what bios have you tried? - I see it, try to stay with one of the f4 beta's not the f4 final.
F4r for ud5 seems to be the one most folks like for that mobo.
I think I read that F4k should be good for UD5 - you already tried rats - :(
Maybe time to start taking memory out.
You only need the uncore at twice the memory, have you tried just x12 UCLK with x6 mem?Quote:
Originally Posted by tonytiger
I would like to try F4m for the ggbt extreme - is there a link for this anywhere? does anyone have it?
Never mind - I found it, had it on a flash dr.
...This board is for me a X-File.
Today I have been testing a new OC policy, trying to find how far I could go up the bclk without changing a safe voltage in the qpi/vtt: 1.34v. Beginning in 200, I've gone up, I tested, I have gone up, I tested ...And I have come to 211, and I've tried very seriously with Linpack. Stable!!
I have taken almost a week to find the lowest achieving super-stable voltage in qpi @ 211: 1.40v. Today, all is different: 1.34v.
I am very puzzled. What has changed so much in a few hours? Today it's cooler in my city, but passed 3 hours of Linpack the PCB should already be pretty tired. The system should be hung. But the board has passed 6 hours of testing.
Now I'm also trying to lower the Vcore. The system remains stable after half an hour of testing. Yesterday was restarted in 5 minutes.
I see the Auto values shown ET6: always the same.
I dont understand anything ...
Edit: System reboot. Vcore has not much changed his behavior
Is it just me ? or 1.65V is missing from the BIOS for DDR Voltage option.
I tried to slightly overclock I7, but found that using AUTO voltage seems to be a big mistake. When I looked into EasyTune after putting BCLK to 150, DDR Voltage was 1.792V<Auto in Bios> and CPU Voltage went upto 1.35V <Auto in Bios> and QPI/VTT was set to 1.215 in BIOS.
I went to Bios and set CPU Voltage to 1.185, Memory to 1.5V and QPI at 1.215V. Setting QPI lower that 1.215 means BSOD while loading windows.
I just use 1.68, which ends up showing 1.66 in the PC health page
Tony flash to F4f
use NB speed four times the RAM frequency speed
also your uncore is TOO LOW....bump it up to 1.5v just for initial testing
leave the rest of volts on auto except RAM (1.64v and vcore > whatever you required there)
boot direct to 190/200/210
i run past 1.5v uncore all the time for bencing........my RAM volts are 2.1v usually so dont be shy trying volts as long as uncore is not really low and your vdimm high
if you decide to run higher NB speeds this is usually where uncore and vdimm require a LOT more volts..........
i am only suggesting you start high to check if its the volts just in case
then you can drop them as you keep testing to find the sweet spots for your setup
I just finished testing F5a and it works fine. I even used my bios profile from F4r/F4k. Only thing I did notice about F4k that I didn't like is that even when the memory timings are manually set to 8-8-8-21 it wants to run the ram at 10-8-8-21.
edit: the F12 bootMenu works fine for me.
Already tried f4f and if i set my qpi/vtt higher than say 1.20 it will not post correctly (keeps restarting and powers off then starts @defaults with bad overclocking message) also noticed it displays 6F when these restarts happen.Quote:
Originally Posted by dinos22
Used the following to test voltages
BCLK 169 (i used 169 as it is stable)
Following your voltage advice
vcore 1.41 (for testing)
vtt qpi 1.5 (won't post correctly) BUT BOOTS normally if i use vtt/qpi 1.18
If i leave mem,qpi,ram multiplyers on auto they are x8,x36,x20.
I think i might try f5a again and 1 stick of dram.I think the issue here is i have a poor cpu:mad:Not noticed in this thread of any really bad i7 920's but overclocking is normally luck of the draw.I wonder if its worth tring another x58 board:rolleyes:
I might try f5a again,do you mind if i have your bios settings?Quote:
Originally Posted by safan80
I didn't write them down yet... I'll post them later.Quote:
Originally Posted by tonytiger