Just got the board few days back, been busy to try out the board but the performance doesn't seem any surprise or out of expectation, or may be worst than p35 chipset, I hope it can be solve via bios, lets have a look at real "item"

the "Crazy Cool" backplate


CPU area


Dual x16 PCI-E


PCB Made in Taiwan
p/s: note tat ITE chipset, our software temperature sensor get reading from here


6 Layers PCB which help in dissipating heat



Ultra Durable2..does it work?lol


I stick a thermal probe at the base of heatpipe cooler


the heatpipe cooler really improve a lot from previous one, much better I would say,but still..x38 very hot

This post has been edited by coolice: Oct 10 2007, 12:40 AM

Voltage regulator


Clockgen


Fixed up with watercooling mounting and ready to test



First of all, look at the bios first


M.I.T



FSB(100~700)


PCIE Frequency



This post has been edited by coolice: Oct 10 2007, 12:23 AM

Robust Graphics Booster


C.I.A.2


Performance Enhance

set to extreme can gain some little bandwidth

Strap tied with ram divider




CAS Latency Time(3~7)


RAS# to CAS# Delay(1~15)




This post has been edited by coolice: Oct 10 2007, 12:23 AM

RAS# Precharge(1~15)



tRAS(1~31)



tRRD(1~15)



Write to READ Delay(1~31)



Write to Precahrge Delay(1~31)




This post has been edited by coolice: Oct 10 2007, 12:24 AM

Read to Precharge Delay(1~15)



tRD(1~31)



Static tRead Phase Adjust(1~31)



CMD Rate

it doesn't work in F3 bios

CPU/PCIEX Clock Driving Control


CPU Clock Skew Control




This post has been edited by coolice: Oct 10 2007, 12:24 AM

MCH Clock Skew Control



Voltage Freak?
VDimm



PCI-E Voltage tied together with vPLL


vFSB


vMCH



MCH Ref Volt Control


DDR Ref Volt Control


This post has been edited by coolice: Oct 10 2007, 12:25 AM

MCH Termination Volt Control


VCore




CPU Feature

switch them off if u r overclock

result will be posted shortly

thanks for viewing and sorry abt my english and the pic abit cacat after compressed by imageshack

This post has been edited by coolice: Oct 10 2007, 12:27 AM

Nice review, bro...

yeah nice reviews and testing.... dont see u online now guess you are doin the in depth testing again now

keep up the good work hehe

Our taiko is busy testing loh..haha and busy playin gwith ma psu.. oii faster test adn gime back lah i also want to play .. hahaha j.kj.k

nice review.. and keep it coming......

coolice koko busy playing new toy.

compared to P35 not so good ah this X38?

so TS, is this x38 worth to get over p35?

does performance gain better alot than p35?

performance wise..it still lose to p35..but i believe its due to immature bios

still compounding the result,so tired now,been slept for 5-6hrs perday since last week ..will post up tml

hmm wonder how darkscythe new toy will be against this

since vrzone winner currently on stock is the maximus..

waiting...................result lol

y don u test and post instead?

ya i wish have the mobo to test out.... since u have it why dont u do it? lol

boss u test and show us which one win lah....

system spec
C2D E6850 watercooled
Team Xtreem / Crucial 10th anny 1GB x2
nVidia 7300gt
silverstone op650 / op1000



P/S: this is F5f bios
Performance Enhance
Standard

Turbo

Xtreme

Turbo and Xtreme mode give a some performance boost but I unable to get it stable with Xtreme when running ddr1300 5-5-5-12

485x7
1:1 266 strap

1:1 333 strap

1:1 400 strap

Comparing between straps, I cant do high fsb if 266 strap was selected but with 333 and 400 strap, I can do up to 560fsb and tat is also my proc highest fsb

This is running at 550x7 with 400 strap and latter with 333 strap



5:6 divider 333strap

4:5 divider 266strap


This post has been edited by coolice: Oct 11 2007, 12:59 PM

New divider 3:4 400strap

3:4 200strap

200 strap give better bandwidth and latency



Clockspeed does affect bandwidth huh?

2:3 divider 266strap


5:8 divider 333 strap


Again clockspeed does affect bandwidth ?

3:5 divider 200strap

1:2 divider 266strap and 200strap


Not much different

Highest fsb 560x7



Try with 10th anny this time



Ok..i heard tat quite a lot of them saying f3 the best bios..so I give it a try
Again my highest fsb,take it compare with F5f bios above,1k MB/s reading bandwidth and 10ns latency gain


1:1 divider

5:6 divider

4:5 divider

2:3 divider

5:8 divider

3:5 divider

1:2 divider 200 strap

1:2 divider 266strap

In F3 bios,the 3:4 divider is missing and also only 1:2 divider come with 2 diff strap selection

New bios out,F5g...hehe..lets see 450x8
F3

F5g



Trying highest fsb can go with diff strap
333 strap,limited by ram,but np to do 560x7 at 333strap 1:1 divider

266strap,max fsb 455

200strap max fsb 447



This post has been edited by coolice: Oct 11 2007, 12:16 PM

400strap..limited by ram again..hehe


Some other result




*When I do comparing between dividers,ram speed always ddr1100 4-4-4-4

wwwwah boss u very hardworking leh.. no wonder few day takda see u online... good leader... taiko

this review sucks. lol (spam alert)

This post has been edited by Sanko: Oct 11 2007, 01:03 PM

hahaha lama neer see u post.. first post after u dissapear , spam.. and in the bigests ways.. lolz.. nice one

coolice koko been busy recently

anyway,i will stick to P35 if like this.

X38=Evolved 975x
P35=Evolved P965

i think X38 oc more stable than P35 but P35 can go further than X38.

x38 have great potential... Currently, it is quite obvious that the board need alot work on its bios and so on.

This post has been edited by irangan: Oct 11 2007, 09:26 PM

I challenge your Everest Memory Read bandwidth and Memory Latency !
800MHz, 4-4-4-12, 1T~




This post has been edited by ahpaul82: Oct 11 2007, 09:42 PM

hehe... nice ahpaul, keep it going. beat coolice till he no face. he day by day geting sombong. healthy challenge is always good!

waaaaa....get board not even half day also come out so thorough review...must be ponteng assignment

no deny the bandwidth on this board considered kinda low..but wait for me..



where got half day only

Before let you take further action,
i did a slightly changing in Memory setting,

here is my challenge again !

800MHz, 3-3-3-4, 1T








ok, is time for me to cabut !
coz kacau-ing your thread...

hahaha

This post has been edited by ahpaul82: Oct 11 2007, 10:57 PM

wei dude that result kindda low

i used to get 11k+ nearl 12 actually on my team xtream PC2-6400 CL3's 1T 3-3-3-8
looking at ure rams .. i would say it wont pass superpi 32m.


any 680 mobo can beat that

hey coolice noticed something

ure everest scores against the 680 is lower
but the sandra scores is on par or higher

suspect is because of the proc latency scores..
as u know the nforce mobo's are very bad at the proc cache latency scores which is y they always lose to the intel counterparts on proc benching

interesting...

This post has been edited by cstkl1: Oct 12 2007, 12:06 PM

im using NF650i board, not NF680i.

and don't compare old Everest with new Everest memory bandwidth,
coz old Everest alwaz can get higher score than the new Everest.
Do you know that ?

my main concern is the Memory bandwidth, not the SuperPI time.

sorry coolice,
im out of topic already

err ure using a beta version and want to compare???
and yes i knew and i have a version 4.0 Everest Ultimate Retail Ori...

where did i say anything about u using a 680 mobo.
do even know y ure 650 mobo everest bandwidth is higher than an intel chipset???

prove it in sandra bandwidth
WHY??.. Sandra vs Everest
Everest bandwidth is not real world bandwidth as it doesn't account for cache latency of the proc
hence separate scores.Real world bandwidth would be closer to the buffered/unbuffered sandra scores
Try it with similar specs and then try some gaming benchmark.Sandra scores also will show fsb/straps/vdimm weakness and actual performance in gaming so till then everest mem benchies is pretty useless as it doesnt even take account of the efficiency's of your mem controllers.

Did u know that??

Superpi
Superpi intel chipsets vs nvidia nforce chipsets... intel chipsets have better scores again.. the latency of the cache mems.... gee ever wondered y ure intel chipset buddies has a faster windows loading times??try this go to sandra.. try the mem latency test... that is going to be a shocking result for u especially at stock..

What coolice is doing is the correct way to compare chipsets.


This post has been edited by cstkl1: Oct 12 2007, 03:22 PM

@ahpaul82
bench again and this is my highest i need more vMCH




@cstkl1
ya..i did notice tat sandra and everest bandwidth differ alot this time which i never know b4,i tot it is just different calculation
btw..i got confuse,is there 2 latency?memory latency and cache latency?and the one shown in everest in wat latency

everest shows mem latency
sandra shows proc cache latency plus mem latency or something like that... or some mem controller latency
generally its like showing the true output rather individual components... i also very vague about this.

but as far as i know real world performance is shown in sandra...
i myself trying to figure out the mystery of strap/fsb/mem timings/chipsets.. need more date and more mature bios.

hmm i think if u saw the mem latency for the nforce mobo for sandra as stock u pengsan.. kekek damn bad..
i think last time i did ask this question and somebody here explain to me y
and then i googled a found a bit of an answer to this...

thats y i always thought this was the reason the nforce mobo always loses out on cpu benching vs intel chipsets.
the mem controllers. latency etc.

generally i think nforce mobos work best on amd procs.
the 650/680 mobos were a hybrid variation or so called upgrades of the 590's
not a really true new architecture like the x38 compared to the 975

heard that the 7series is also an upgrade revision of the 680's which i am not suprised that it will be coming out for amd phenom procs first.

wei bro if u paired that mobo with that sandra score with a gts/gtx/ultra
ure going to break the aquamark 220 mark.. very sure..kekeke

try it...as long u have an overclockable graphic card ( PS inrease the mem speed on the card for more fps overall first.. find the max and then find a corresponding max core/shader at that mem speed)

gila score leh 9.8k

hmm i took out my 680 mobo already and the other is in my gf house
if not would have shown u stock performance

if ahpaul82 dont mind
dude show ure sandra mem latency score at stock performance and also everest...

also coolice the performance of latency will improve with lower latency on mem especially when run at in sync with fsb

This post has been edited by cstkl1: Oct 12 2007, 04:55 PM

icic..no idea leh abt nforce chipset..never had a chance to play with it..hehe
oh..aq3..never try this benchmark b4..but 220k only if there is a card for me
to be honest..9.8k is kinda suicide run..but i think can get it stable with 9.4-9.5k

This post has been edited by coolice: Oct 12 2007, 10:00 PM

560x7 1:1 4-4-4-4 passed spi8m + 9.5k sandra

WoW!!!..quite a mad while reading u all posting here...get fainted...

I only able to do this...
didn't try at stock speed

Before changing Memset setting :

[Sandra Memory Bandwidth]


[Sandra Memory Lantency]


[8M SuperPI]


This post has been edited by ahpaul82: Oct 13 2007, 01:18 AM

After changing some Memset setting :

[Sandra Memory Bandwidth]


[Sandra Memory Lantency]


[8M SuperPI]


This post has been edited by ahpaul82: Oct 13 2007, 01:19 AM

hmm decided to start my own thread on the maximus

This post has been edited by cstkl1: Oct 17 2007, 01:47 AM

got a xigmatek to play with last night,so did some serious/stressful benchmark again(sorry,cpu test again )
1.43v set in bios,cpuz shown 1.408v,easytune5 pro report 1.42v

560fsb bench 3d06

9.9k sandra bandwidth for today

really nice proc u have there.

bid 1.5k for it??

since the vcore so low how come dont want to run higher mhz??

This post has been edited by cstkl1: Oct 20 2007, 02:32 AM

this proc will not for sell unless i found replacement
erm..the reason is after 4ghz,it doesnt scale well with vcore,last time when i did my p35-ds3p testing,i find the highest 3d06 benchable clock speed is 4.14ghz @ 1.57~1.59vcore..hehe
secondly was i worry tat xigmatek cannot handle the heat later

This post has been edited by coolice: Oct 20 2007, 02:37 AM

time for some ln2 then boss hahah

i see
good proc though..

hmm i am having probs with the maximus
it wont let me go pass 4ghz no matter what

nice going there, coolice. Lets have some LN2 bench ya


@Cstkl1

Its not that your maximus is not letting you go beyond 4Ghz....I guess you know the reason why bro.

hehe..will update the score later

show me ur dewar first

eh if condensation is happening at 1.6v again

cannot go above 4ghz.

dude u know my benches
i will only do stable benches
my 4ghz at 1.6 already stable like gila
3dmark/orthos/superpi all pass with flying colours
i need to up the voltages somemore since the coolit like getting condensation on idle... which is not good.

dewar??somebody here has a liquid nitrogen flask??

extremeocer..

last warning if not going to be flaming u like mad

dont promote ure sales thing here
post relevent posting on x38 and not cooling
this is not a cooling section

This post has been edited by cstkl1: Oct 20 2007, 11:54 AM

Promoting my sales???.....i am just stating the facts here bro. You say you cant go beyond 4ghz...and i am telling you why. Its that simple. You want to flame me....go ahead. You want to report me...go ahead. But before you take any actions....just a gentle reminder that everybody know what kind of attitude you have bro. Isn't OC related to cooling? Sometimes i really wonder on your OC skills.

Just accept the fact that cooling is limiting your OC....not the mobo or CPU, at least for now. I shall stop here....since this is not my thread.

This post has been edited by extremeocer: Oct 20 2007, 12:06 PM

wow....
what an insight when u have no clue what my temps are like
wow....
pro ocer here ppl
again this thread and the other is suppose to be a x38 thread.
go to xtremesystems forum and tell the same thing at all the x38 thread. y dont u?? helping malaysians is it??

and sorry i aint a rat like you who complain to mod everytime infact u did it already just a few days. In fact i dont see the point of complaining when i can hold my own on facts.

everybody here is testing the limits of x38 on air/water. if u want to start a extreme cooling thread on x38 by all means. or get a x38 and produce some results here.. till then all talk and no show makes u look really stupid

sorry my attitude is simple.
talk relevant things simple....not kopitiam style with loads of excuse.

infact y not post all ure scores.. since i cant even find them
and show that thell all can pass superpi32m/ ecerest cpu test/3dmark06/3dmark05/aquamark
also some sandra scores.
dont be so shy. show ure not all talk.. and no show.

anyways this is definitely my last reply to u..
i SERIOusly doubt ure OC skills with no post.

---

Added on October 20, 2007, 12:16 pmcoolice
did u see rhlc posting on ddr2 thread
hmm 500fsb q6600
nice stuff.

---

Added on October 20, 2007, 12:34 pmcoolice
i counter checked ure 3dmarkcpu scores
its missing by a 100 points.
check ure rams and vcore again.

This post has been edited by cstkl1: Oct 20 2007, 12:34 PM

just saw it..yeah..500fsb on quad under air/water is nice
ya ke u r using same setting as mine,right?
==..got sth to do with vcore meh?
btw,i try aquamark3 just now..843/1053 2900xt,the gpu score is about 35k-36k,izzit 2900xt not good in aq3 bench?

try 3d05 this morning,current highest is this
sory,forgot to open cpuz for cpu tab,proc was running at 4ghz anyway

yup using the same 500x8 1000 cl4

see your sandra scores for 500x8.
feeling this is difference in mobo.

2900xt not sure for aquamark.. heck i couldnt get it to work last time.

nice scores for 3dmark 2005

hmmm whats the cpu score for aquamark??

its ure proc.. for 4ghz 500x8.. ure sandra at 1000 cl4 post please
see whether it can hit 9000 consistently
cause your 3dmark06 cpu score should be around 3614 thats what i got for 4ghz...

currently my mobo is confusing me...
3.6ghz only need 1.38v

so wth is going on with going above 4
oh yeah hmm tested the limits
520fsb stable for benching only but failed orthos
500fsb stable...

envy ure proc...
geee darksycthe x6800 also rare can do 530..keke on air/water

did suprfly tell u or not on the new 505 bios... no vdroop at all
and bios set vcore and reported on probe only 0.02v difference
the loading calibration now working

This post has been edited by cstkl1: Oct 20 2007, 03:48 PM

erm..ok..i test the sandra later..hehe
500x8 aquamark cpu score is 22k
yup..he did,0505 solved the vdroop problem..but divider still cant get it work..zzz

dude post yours highest sandra superpi32m pass..

just curious if u dont mind.

which one?9.9k?tat is suicide run
stable run is 9.5k
btw,can u try 29xt aq3 for me also?it seem 29xt not well in benching aq3

This post has been edited by coolice: Oct 21 2007, 02:31 AM

i did before last time

its slower than my gts
at that time i thought the directppl driver problem
as it crashed ramdomly
but there was one guy who did some 2xx aquamark scores in xs i think

dude lets do some 515x7 comparison run your rams at insync

3dmark01/3dmark05/3dmark06/sp1m/sp32m/sandra/aquamark( just post.. interest in the cpu scores actually)


then can see maximus vs dq6.. which one is a winner

results a bit inconclusive at the moment

also ure gigabyte seems to have a better optimized bios for rams and OC
so just wanna see the results

C2D E6850 cooled by Xigmatek 1283
Crucial Ballistix Tracer PC-8500
Sapphire 2900 Pro 512mb
OP650

initial aq3 result from 2900 pro
500x8 1:1 4-4-4-4


finally break 200k ,highest score i got is 206k but din printscreen it
560x7 1:1 4-4-4-4


3d06 with 2900 pro



Change back to 2900XT 1gb
cpu 512x8 1:1 4-4-4-4,vga reached the limit,870/2178

same as above but cpu was running 500x8 1:1 4-4-4-4


personal highest score for 3d06

cstkl1,this is for u as request

result with 2900XT 1gig
514x7 1:1 4-4-4-4

aq3



3d06


3d05


3d01


super pi 1m


super pi 32m


max stable + highest sandra score


forgot to printscreen down the 500x8 bandwidth for u..but iinm..the sandra bandwidth is about 8.5-8.7k,not 9k

This post has been edited by coolice: Oct 22 2007, 11:41 AM

this is scores with 2900 pro

3d01 - 63923


3d03 - 43485


3d05 - 22485


3d06 - 12693

no wonder u asked me jsut now.. u already did it..

give me a few hours..
didnt have time yesterday

doing something again at 4ghz.

erm i guess ati driver is optimized for vista.. looks like ur score is not good enuff..

btw coolice, take care of my 29xt ya



This post has been edited by irenic: Oct 22 2007, 08:54 PM

wei dude y u didnt do 515x7
anyways posting the results in a few minutes

lets put here
Aquamark
coolice
2900 xt - 193295
cstkl1
8800gts - 208143

Sandra
coolice
X38 DQ6 - 9013 - 1030mhz 4-4-4-4
cstkl1
Asus Maximus SE - 8975 - 1030mhz 4-4-4-8


3dmark 06
coolice
2900 pro - 12693
2900 xt -12774
cstkl1
8800gts - 12700

3dmark05
coolice
2900 pro - 22486
2900 xt - 22482
cstkl1
8800gts - 21468

3dmark01
coolice
2900 pro - 63923
2900 xt - 64631
cstkl1
8800gts - 66710

will post the rest on this thread tommorow sleppy
sp1m/32m

This post has been edited by cstkl1: Oct 24 2007, 01:00 AM

owh..i tot u want to compare 3.6ghz clock speed,515x7 is over so i use 514x7

nah its ok
just repost the super pi 1m
32m nevermind

the rest is minimal on 7mhz difference.

wokey,,will do tml..hdd just corrupt just now..zzz

yeah i had to rma my ocz's titanium alpha vx2..kekeke

luckily one pair of team's just came back from rma today..

thats y i didnt post
the ram like failed memtest and was giving a bit inconsistent scores.

this time going to be carefull with the teams
should be getting some ocz's 9200 reapers in the new few weeks.

bro..i post the result on sun or mon..got back to hometown until sun for sth
btw..deciding whether want to let go this e6850 or not

no hurry
me also now not sure what ram to use

but this is going to be my 24/7 setup
520x7 at 1.376v

seems to be perfect
an windows load fast
games also damn fast
..keke

tried difference at 475x8 at 1.42 but hmm the rams are not running as fast

with the team ram at 3640 can score around 9.2k which is fast enough for me

hmm team cl3's d9gmh really handpicked yo...

the subtimings on auto already so tight.

hehe..my new team ram replacement should arrive yesterday..check tracking system shown attempt delivery at 4.42pm

well if u want to downgrade

u know who to look for

will trade with mine +RM500 well still negotiable..

so atleast u get some dough and not bad proc.. not as good as ures but can do 520..

hmm need it for my rig.. cause i really want to run 4ghz 24/7

anyways u have until end of this month to decide
cause if the yorkfields scale as well as they have reported on air then sorry lor..


This post has been edited by cstkl1: Oct 26 2007, 02:19 AM

hehe..the sales is up..but don accept trade in..paiseh

its ok already got buyer for my proc. but at a way lower price since i want ures like asap.

Those are cheap. Orthos and OCCT are the most evil so far if you can pass those you can pass anything

well,im not 24/7 stability freak,as long as it would able to run benchmark for me,tat will be fine

thats if u intend to find stable 24/7 setup for ur system

well FYI he only used SANDRA to get memory bandwidth....not to test stability.... i dont recall that he mentioned that he used it to test stability and its 27/7 proof lol