Printable View
Holy mother of Jebus it reached 50 LinX cycles over night on settings it wouldn't even boot Windoze before. o_0;
I know it must be due to incompatibilities they fixed so results for others may vary but this new BIOS makes me feel like that maybe I'll be able to join the 4.5GHz club in the end lol.
Will be sticking with 11/05. As after that DFI removed the four turbo multiplier to core loading settings that had been there before, for just the one setting. So people like me who brought the board with an unlocked I7 EE CPU to OC dynamically using the turbo settings, got boned. You need the four setting instead of one as you can then run different multipliers depending on how many cores are in use. So a higher multiplier is possible when just one core is in use.Quote:
Originally Posted by Forin
So now if we want any of the new unlocked CPU's to do turbo overclocking on the cheap, we will have to get a new board. Good job DFI.
If you are confused just see the explanation I gave on the issue on DFI's own forums here: http://csd.dficlub.org/forum/showthr...8605#post88605
which OCZ did you use?Quote:
Originally Posted by p2501
for me the bios that change many stuff was the latest official (i don't remember the date...), i gained about 4 second at super-pi 32M.
i love this board.
p.s. i don't know what problem did you have, but 4 or 5 months ago i can pass linx at 4.8 ghz with an i7 920 with 6 gb of ram. :shrug:
I'm running 12GB of OCZ3P1600LVGK.
I seriously doubt it was the memory because many others are running these sticks just fine. Wasn't there also posting going on in the DFI forums about boards that had instabilities when paired with a W3520 while others where fine?! Btw, I described the problems in the post you cited. :shrug:
All is well now. ICH/IOH and IOH I took down to stock volts (were at 1.19/1.5), vcore -3 steps, vtt -3 steps, memory -2 steps, both qpi clock drives to normal, boot bclock back to auto. All of which was required to be stable before.
Somewhere in around 60 watts saved at 100% load, maybe even more.
Edit: Nope the above changes didn't work out, they were good for 50something runs of LinX but F@H on CPU and GPU killed it overnight. Btw, why isn't strike out for text working here?! :shrug:
nothing special here, no WR or something like that but today finally I had 10 litres of LN2 to try my i7 950.
I'm happy because it's been a long time since I had a good bench session at LN2, in the last six month I lost my time with p4 (8.1 GHZ then the RE died....), phenom 1090T with a crappy MSI 890FX and another bunch of rubbish...
now seems that the luck is coming back, this 950 work pretty well.
config:
- i7 950
- DFI UT X58
- Kingston HyperX 2000 cas8 MGH-E-Hyper
- Asus GTX 280 - stock cooling
- P&C 750
- Guglio's CPU pot
I've this graphic card from 2 days, then i didn't have all the time to test driver and tweak...I've tested only the max benchable frequency at stock cooling.
there is a lot of room for improve the score.
BTW in this bench session I'm focusing on the CPU, the VGA need more testing and some mod before run it at LN2 and do some damage on HWBOT.
today here was damn hot...
http://forum.hwproject.net/imagehost...2069a91965.jpg
setup pic 1
http://forum.hwproject.net/imagehost...2069a67202.jpg
extreme air cooling :asd:
http://forum.hwproject.net/imagehost...2069ab5f70.jpg
with the first 4 litres of LN2 i can reach 354k am3 with CPU at 5150 MHZ...not bad at all for the first touch with this CPU.
http://forum.hwproject.net/imagehost...207fa95fa4.jpg
when all was going well i tried 24x multiplier...not a good idea, in fact the cpu was going crazy and stop to work for 3 or 4 hours :D
I need to refill the dewar and start the round 2 :D
I'm setting up water cooling for my system. I have leaks in radiator.
I can confirm new beta BIOS 15/06 is stable for me. No problems. I quick tested on WC 200x21 4.2 GHz,1.28v and 4Ghz undervolted (linx stable).
I think that with fans on a radiator and lower ambient it might be possible to check 4.5GHz. Will see. But it will need >1.3v.
I managed to not fry my board remounting heatsink and applying MX-3, and temps even dropped significantly. I guess I'm lucky or something..
Setup will be:
Enzotech Sapphire rev.a
Swiftech MCW-30
XSPC dual bay res. pump
Thermochill pa120.3
Feser One
http://img88.imageshack.us/img88/391...0708030.th.jpgcooling
http://img198.imageshack.us/img198/2...0709045.th.jpg
http://img121.imageshack.us/img121/6...0709046.th.jpg
http://img683.imageshack.us/img683/2...0710049.th.jpg
http://img188.imageshack.us/img188/6...0710051.th.jpg
http://img30.imageshack.us/img30/857...0710057.th.jpg
edit:
Can swiftech be mounted like in photos? screws are too short so I removed some plastic washers.
Hi, since you are using the new beta can you please confirm something for me. In the genie bios settings has it gone back to the settings in this pictureQuote:
Originally Posted by Forin
http://img.i4memory.info/reviewimage...8_bios_002.jpg
i.e the following settings have returned
x 1 core Max Turbo Ratio: xx
x 2 core Max Turbo Ratio: xx
x 3 core Max Turbo Ratio: xx
x 4 core Max Turbo Ratio: xx
Or is there still just this setting instead?
Max Turbo Ratio: xx
Thanks
Max Turbo Ratio xx is available only:
http://a.imageshack.us/img411/690/bios0615.jpg
Thanks for the response. I will drop DFI a line. I know I seem to be the only person in the world to use this method of overclocking, but it is annoying that it has been taken away.
I think people here would be jumping up and down if they removed the ability to change the CPU base clock from the BIOS.
For my rig, the D724 and the DB05 bioses are the best. They are very similar in stability and performance. As i need the virtual pc machine, i keep the DB05. D429 falls in memory performance, perhaps because of this problem reported by coolldh (would be nice if someone could confirm this):
and falls too in stability (tests fails quickly with the same settings in which DB05 is stable). The only great advantage is that D429 does not completely turn off the computer each time you restart with VT-d/VT enabled. DB05 turn off the machine few seconds before restart (Vt-d/VT enabled). Perhaps itīs time to test the D615...Quote:
I found a bug in 0429.
if you set DRAM=Bclk*10 Ucore=Bclk*20
DRAM run in Bclk*8,Ucore run in Bclk*10
http://img215.imageshack.us/img215/6...0715080.th.jpg
This is setup:
http://img130.imageshack.us/img130/9...0715083.th.jpg
(don't buy Sony Erickson vivaz, it takes very bad photos, image stabilizer is very bad)
I think that this 22x multi for one core will be a problem when testing 210+bclk, because it will put my CPU on ~4,7GHz, before I even start to test.
HOW CAN I LIMIT MULTI TO 21x for i7-920?
Look:
http://img821.imageshack.us/img821/1958/41513080.th.jpg
There are some older screen, when I used destiled water for leak testing.
These temps are stabilized, max.
I don't have time on sitting and watching linx right now.
http://img24.imageshack.us/img24/2634/13759860.th.jpg
http://img257.imageshack.us/img257/5071/98079452.th.jpg
ph0b0s, my method of over/downclocking is to set just BCLK and leave VOLTAGE setting on auto, adjust VTT voltage. LOAD it's 21x186 1.2v, IDLE it's 12x186 (2,22 GHZ) at 0.9v. Nice and cool.
(I know that on screen its idling at 22x multi, but its with high performance power profile)
You can disable CxE in BIOS, but it will negatively impact your memory performance. I had DFI UT X58, actually I still have it, but it's on the shelf right now.Quote:
Originally Posted by Forin
sorry man, but everest bandwidth benchmark isn't the best performance meter
i use exactly the same subtimings/timings for all the 3 run
core affinity on the 4 core
priority very high
RTL 61-62-62
EIST - enabled
c1* - c1e
8.54.203
http://forum.hwproject.net/imagehost...1ac3b15e47.jpg
EIST - enabled
c1* - disabled
8.53.906
http://forum.hwproject.net/imagehost...1ac3b4679e.jpg
EIST - disabled
c1* - disabled
8.53.859
http://forum.hwproject.net/imagehost...1ac3b6f646.jpg
with both EIST and c1* disabled the bandwidth is lower but the time of the 32M is better...i don't mind if everest show lower bandwith if the REAL performance of the system is better.
only c1* enabled give me poor performance, both EIST and c1* disabled is the way to go.
hey guys, a long time user of dfi board here (not that good a user) i'm looking at making the move from my x48 to an x58, just want to know how these boards overclock, are they stable, and will i be doing my head in learning to overclock on one? i dont have the time to spend overclocking as i used to so do you guys recommend i got with this board?
Regards Andrew:shrug:
edit : i will be looking at using a i7 930 and ram around 2000mhz 9-9-9-27
the only issue that i still have is that with Elpida Hyper ICs based memory over 2000 MHZ the performance become very poor and sometimes the system freeze.
i've tried plenty CPUs and RAMs but i've the same problem with all of it.
so only issue is when trying to run over 2000mhz? would it be better to run slightly slower and tighten the timmings???
Andrew
i've problem over 2000 MHZ ONLY with Elpida hyper RAMs, now the only memories with these chips are corsair dominator GT (not all the model), super talent 2000 cas8, cas7, 2200 cas8 and kingston 1866 LoVo.Quote:
Originally Posted by terroristone
with any other RAMs no problem at all even at crazy frequency.
if you plan to buy one of the memories of above would be better if you go for another mobo.
for dram 2000Mhz, uncore 4200Mhz.
u have to pump vtt to at least 1.4v.
but watch out for the PWM temp it will hit 105C in minutes & throttling will occur eventually
how much vtt you must feed the uncore is cpu dependantQuote:
Originally Posted by owikh84
on the other mobos is the same
for the PWM overheating problem you can do 2 things, put the NB under H2O would help to keep also SB and PWM temperature way much low or put a fan blowing on the PWM heatsink.
i don't see digital PWM like a problem :shrug:
If the PWM heatsink had bad contact with the mosfet blowing a big table fan towards the heatsink also won't help.Quote:
Originally Posted by Mafio
that's for sure, but i never heard about that problemQuote:
Originally Posted by owikh84
i know at least 3 people (including myself :d ) with this mobo and none have problem of bad contact between mosfet and heatsink :shrug:
http://img641.imageshack.us/img641/826/30943556.th.jpg
http://img686.imageshack.us/img686/5527/58026010.th.jpg
At 4.5Ghz my mobo is downclocking from 21x multi to 20x under load.
IIRC you should set the following:Quote:
Originally Posted by Forin
Genie BIOS Setting
-CPU Feature
--Set VR Current Limit Max: Enabled
--Thermal Management Control: Disabled
--EIST Function: Enabled
But be careful, as then noting stops you from burning your CPU. ;)
Thank you, it's permanent 21x now.
I think that this option is most important:
--Set VR Current Limit Max: Enabled
Quote:
Originally Posted by donmarkoni
After testing some i7-950s I'm back to basic for some efficiency and BCLK fun.
Of course the benchmark is my favourite, the only one, the unmatchable: super-pi 32M.
rig:
- i7-920 // SS 1.2 HP
- DFI UT X58-T3eH8 // stock cooling - bios D615
- KHX 2k cas8 MGH-E // stock heatsink + silverstone 80mm
- XFX 6200 LE 256 PCI-E
- P&C 1.2 KW
- 2003 Server TW heavy tweaked // custom CW
- Winamp // Digitally Imported vocal trance (the key for a good 32M run, without this BSOD for sure :rofl: )
http://forum.hwproject.net/imagehost...f2091d84a8.png
32M / 5042.4 MHZ / 7.00.875
http://forum.hwproject.net/imagehost...f1f639b14c.jpg
RTL 61-65-67
all the other subtimings are showed in CPU-Tweker.
Tomorrow I will try to take the first place in 6.xx category here: http://www.xtremesystems.org/forums/...d.php?t=228269
Sad to see that there are only a few people that still using this crazy mobo :(
p.s. sems that I did a bit of pwnage on HWBOT, look at the frequency of the people around me :D http://hwbot.org/community/submissio...min_0sec_875ms