[Team.AU]Core i7 PCMARK05 W.R. + SuperPi LN2 action > GB X58-UD5 + Corsair Dominators

[dinos22]

Hey fellas

I had a couple of bench sessions in the last few days to test A LOT of bios files and different things to get the most out of the setup. I've had some trouble with LN2 but finally i think i'm squeezing it more....I've ended up on a bios that works best as per specs

Test Configuration:

* Gigabyte EX58-UD5 F4N bios
(vdroop mod)
* Intel Core i7 965es 3831A457
* 3GB 1866MHz CAS8 DDR3 Corsair Dominators (TR3X3G1866C8D)
* PCIe nVidia 8400GS
* 1000W HX1000 Corsair PSU Monsta
* SS Phase by C-BuZz
* F1EE by Kingpin
* 160GB IDE Hard Drive
* PCMARK05 tested with 3xiRAM Drives
* Windows XP Pro 32bit SP2 Lite
* Windows Vista Ultimate 32bit SP1 Lite


I have tested many bios files now and for benchmarking i am finding that F4N is the best so far. This is the first one that allowed me to unleash the speed of this CPU (somewhat). I still feel there is more in it

It also has a higher working voltage for uncore and subsequently allowed me to benchmark up to 4.8Ghz NB SICK

3DMARK XFIRE TESTING SETUP


Sorry no pics with Corsair TEC RAM Cooler but i used it the other day...works well
Should have left it to gain some RAM clocks but no matter



PCMARK05 TESTING SETUP



Who said Pentium 3 is dead. It makes for a killer iRAM rig



FIRST SET OF RESULTS ARE SS Cooled
PCMARK05 > 25812 fastest Core i7 system at the moment
used 3xiRAMs
i still a bit to go to catch the fastest PCMARK05 system overall
i had some issues with startup and hdd usage so that score will change..................it should havebeen in the 27.5K area somewhere



Third fastest 3DMARK Vantage Xfire 4870X2 run worldwide > 28559 points . I was 20 points behind Nazar with higher GPU clocks
This board has very stable power which allowed me to bench such high MHz on vantage
Evap temp was dropping to -35 to -36C during CPU tests



I did a single 19K run a few days ago....but it was my first run with SS and i used too much vcore




3DMARK06 Xfire run 31875points
canyons need some more work





3DMARK05 Single 4870X2 > 40420 points



3DMARK05 Xfire 4870X2 > 42959 points



3DMARK03 Xfire3 4870X2 > 140990 points



Aquamark3 single 4870X2 > 333910 points



3DMARK01 single 4870X2 > 92351 points



Here is a 32M SuperPi run i did on Single stage with F4N bios

have a look at the NB speed
This motherboard rocks heheheh




Moved onto LN2 now

LN2 with Core i7 is no picnic as many have discovered including myself. This board wasnt helping either but i think i finally wrestled some settings i could use in future to improve my runs

First of all F4N is definitely the bios to be on for LN2. I've used up a few tanks of LN2 and tried absolutely everything

This CPU has a peculiar cold bug. When i setup my settings in bios i have to bring the pot temp down to around -70C save and exit and the board will boot up fine.........if i used -75C it will freeze on first post, -80C it wil throw 55 code or 88 code.

When i reset cmos i can boot any any temp. i've tried as low as -110 or -120C i cant recall now

Second bug was to manage -77C within a degree or two for these superpi runs. It didnt like -79+ or below -71C at high frequencies . You could do whatever you want below 5GHz but as soon as you pass that you are in for a different ballgame

I think that at least with Gigabyte and some chips it might help some of you other guys trying to get your head around these CPUs... or at least give you something else to try and say how Dino is full of hahahahah





.................................................. ...........................

Anyways i was first testing different settings and used CAS6 in this run around 850 mark from memory

Reached a max of around 5226Mhz 1M

Disregard efficiency ....i can bring it down to 407xx mark on this board but its not worth benching as E8600 can beat Core i7 with ease



Had a couple of failed runs and realised it was temp related so i just managed the temps withing 1C (extremely glad i have F1EE pot from Vince for this )



second run was at 5153MHz but while trying different things i also relaxed timings.
At this frequency i should easily be in 6 minute territory with a 6GB kit i still dont have heheheheh nudge nudge and with proper tweaking

[Expat GriZ]

Congrats on WR !!! Very nice testing, should help everybody out on the Gigabyte boards. Love all the tape, an Ozzies best friend!! Keep hammering it!!

[Vivi]

Leet!!
i just wanted to congradulate you on your 32m 6th place but i see u made a thread so i can say it here!

Welldone you chop! excellent 32m, i wish i can do that haha. and your 05 single is also very good, i didnt even get that with 6.2 E8600, then again i didnt tweak and my 05 sucks

Congratz on 24th!

[dinos22]

Congrats on WR !!! Very nice testing, should help everybody out on the Gigabyte boards. Love all the tape, an Ozzies best friend!! Keep hammering it!!

"There is nothing that s(h)ticky tape cannot fix mate!!!" - Guido Hatzis



thanks Vivi

[chuchnit]

Where exactly can you pick up an iram? BTW great benching dinos. When is Shamino going to send you one of those X58 confidentials as a favor?


Edit: Google is my friend, Google is my friend..........

[eva2000]

woah man you're on fire.. 4800 NB frequency! dam what voltages were used for above benches ?

Love the i-ram setup LOL

4.564s to go for sub 7min!

congrats Dino!

[Dsjjang]

Congrats, dino~
all nice results and impressive UCLK~

[dinos22]

thanks guys

woah man you're on fire.. 4800 NB frequency! dam what voltages were used for above benches ?

Love the i-ram setup LOL

4.564s to go for sub 7min!

congrats Dino!

hahaha voltage is a bit scary
some of you will need to sit down

for 4.7GHz-4.74 i could get away with 1.63v uncore

for 4.8Ghz i was using 1.69v just to be safe
i've used 1.71 to test as well

F4N is the only bios that can give components this sort of voltage

i run RAM at 2.1-2.16v as well as a result of high NB speeds

this RAM will do CAS8 1GHz at 1.68v but if you want high NB speed you have to increase volts to RAM as well heh

i still havent tested the limits of NB speeds and uncore voltage
maybe we will see 4.9GHz too soon who knows heheh

[eva2000]

1.69v uncore just to be safe

Wonder what just 2x1GB dual channel would do for required voltages and clocks ?

[dinos22]

meh dual channel is slower

i like tripple channel

also 6GB kits are MUCH faster..........4-5 seconds than 3GB

[Viss]

Very nice numbers Dinos mate Like your iRAM rig

[T_M]

Awesome benching Dino

[Perica_Barii]

good results

[MARCO74]

Congrats Dinos fantastic run and amazing chip

[Oliver]

great work man, thats what i call to keep pushing it huh

go get that sub 7m will ya

[Achill3uS]

i always wondering how you can fill up one thread/bench session with such a lots of results nice work dino

[eva2000]

meh dual channel is slower

i like tripple channel

also 6GB kits are MUCH faster..........4-5 seconds than 3GB

maybe for pi but 3dmark ?

[dinos22]

Hey Tim

Looking forward to seeing you do your thing again soon mate . We need ya

eva2000 >> well for tests that like good RAM peformance like Aquamark3 or 05 you are going to need performance RAM and every little bit counts

btw you will notice that one particular benchmark likes dual channel RAM better in my screenshots

[giorgos th.]

very nice dude....
so F4n IS the bios to work with.....not just my opinion...
running for LN now....
so you found it`s better to work at lower temps heh?-100+C was no go?

[ZoLKoRn]

Great job m8!

keep pushing it for sub 7min spi32

[dinos22]

hey Zolkorn thanks man
i'm not as extreme as you guys but i try my best to do a Thai-OC-dance during my bench every time

very nice dude....
so F4n IS the bios to work with.....not just my opinion...
running for LN now....
so you found it`s better to work at lower temps heh?-100+C was no go?

-77C was optimal temp for me

i do have to test more but that wont happen for a little while

i will concentrate on E8600 again for the next month

[giorgos th.]

mmmmm maybe i should try with a bit of DICE first.....
maybe that explains the odd behaviour between SS - cscd - LN...

[dinos22]

mmmmm maybe i should try with a bit of DICE first.....
maybe that explains the odd behaviour between SS - cscd - LN...

LN2 i would start with first
because you have best control over the rig

i always wondering how you can fill up one thread/bench session with such a lots of results nice work dino

me tooooo hahahahah

[Zeus]

Sweet results there Dinos, i really dig the idea with the PIII and Iram but how can you stop running 32M when you're only a hair away from breaking sub7 minutes?

[jesperme]

Nice setup and results!