its the mem settings the skew..reduce the skew.. get the GTL's right.. and u will have it..u can see my vdimm and that will give u a clue.. i am reducing the strength ...
to a moderate.. technically full strength 1000mhz 1:1 should be closer to 2.3vdimm

ures is more like forcing if by pure vcore powah...

flafinger u sure u want to sell off this batch??

this already pass OCCT... but i usually try one hour..

FPO Q746A332 scales very well

oh yeah uzairi dont try to copy asus x38

their NB stock vcore is 1.2v and that can do 400 fsb since asus X38 supports 1600fsb 45nm native.

but generally i think all mobo's should be around the same..

oops forgot to show trd8
repost pic
[attachmentid=395161]

This post has been edited by cstkl1: Feb 7 2008, 11:28 PM

okay will do it but do u know the reason y they made the fft run around 8-10 minutes.. because of the priority cycle..
u need to off all the programs..
its like shooting a gattling gun every 8 mins..
what ure asking is not stress dude.. what ure doing is every 1 min off and on the load and restart...

small fft is to just to test ure L1/L2 cache interaction with ure core.. hence y prime states run 8k min to max 4096 on 15 mins interval..
so wolfdale would be 8k to 6132 max.


what ure saying is every 1 min example on real life on and off a program that stresses ure core... not real world test..
my 2cents.


here is 1 hour pass.

memory stable as hell.. proc also..
perfect.
[attachmentid=395167]

This post has been edited by cstkl1: Feb 8 2008, 12:14 AM

so far i know mine scales up to 4.4ghz well
after that its a bit way off...

i think its ure mobo bios prob the way its controlling ure gtl's.

cause if i let this mobo control the gtl/fsb termination/NB.. will be force clocking it.. example was the earlier 1.296v..
i was forcing it to get my mems stable..based on some 65nm overclocking principles and habits..

so went back old school.. took days but now its just awesome..
testing now on 2nd strongest mem setting... bandwidth just jumped a big notch..


This post has been edited by cstkl1: Feb 8 2008, 12:46 AM

ah i see
but hmm isnt 1 min cycle afaik similar to restarting ure stress every 1 min ....
it does not stress the core more just completing the cycle faster..
i suppose it shows the error faster.. but real world.. no meaning.. get what i mean??.. ure running a stimulated test that has no meaning and will not happen..


there must be a reason for this..
cause as i said the prime95 stable test setting is custom smallfft 8k-4096k 15 minutes..

well i am running what u ask btw..

just learning to see is there a significance on this error to determine some poor setting..
so far i can see one..

it will stess the core to the lowest vcore..
so ppl wil droop and fluctuate at the end.. might fail..
cause this test will make it go the the lowest.. since its changing every 1 minute

good test uzairi to see any amkong case..
but with ppl who have load line in their bios setting this test is useless..


This post has been edited by cstkl1: Feb 8 2008, 01:24 AM

wait retesting ure method again

it seems it clashes with rams.. on 64k..wondering y..

need to read intel proc spec.. on memory mapping.

clash..

hmm the NB voltage..

weird ure test.. on strap 333 1000mhz CL4 okay.... even though it fails memtest.
on strap 400 1000mhz CL4 not okay.. even though all test says its okay

This post has been edited by cstkl1: Feb 8 2008, 03:30 PM

64k.. it loads the 1024mb of the mem..

32k is correct.. since l2 cache is 32k...

64 k is incomplete..
if u want to test system.. blend is better..

i wonder the extra 32k which part of the mem it takes .. the higher mem?? or lower..
hmm dunno about u but i get worried when i see half of my rams especially D9's being stressed like that..

hmm dude i tested it 3 times on different settings
confirmed.. ure way is just stressing the ram also.. no wonder it was disapproved.. as the test for small fft is for proc..

if u want to stress the proc further make the max fft 32k. for e8400.


This post has been edited by cstkl1: Feb 8 2008, 03:56 PM

errr flaw??

cause think about it
if its going to take 1gig of ram..
y not just increase the max fft to 4096..

hmmm 4C less on core
2C more on NB compared than small fft 8k. i think it just stresses the NB and RAMs..

its not proc only...

hmm fast way to kill ure rams.. on 1min..

---

Added on February 8, 2008, 4:52 pmbut good stuff.. found out need 0.02v more on my NB

This post has been edited by cstkl1: Feb 8 2008, 04:52 PM

nay..
cause ure purpose is to stress ure proc.. first not ure rams.. or NB..

so 32k max or just leave it at 8k..

my two cents on the topic..

err i dunno how much stress u want
err y not go or TAT..

if u pass that.. GOD will come down and bless u..kekeke although i dont think it supports penryn.

check ure 5v rail.

---

Added on February 9, 2008, 1:11 pmkeke i find it weird

unknown proc with no screenshot that broke the world record for 4ghz sp1m sold off to unknown person.. easily

my proc with better overclocking result hasnt..
weird..

lowyat admin should allow to have some music .. would play x-files..
unsolved mystery unless..err

This post has been edited by cstkl1: Feb 9 2008, 01:11 PM

-------edited--------

This post has been edited by cstkl1: Feb 9 2008, 06:04 PM

same here

weird

plugged in a e6600 and i got back min 8k max 64 k also..

hmm dont trust orthos much nowadays
prefer prime but thats 15 minutes min 8k max 64

something is awefully flawed here..
cause i was testing with two versions of orthos.. have u noticed that with e8400 u always get core 0 failure while core 1 happily chugs along

usually my e6600 and e6850 either both fails.. or either one.. but never always the same core.

---

Added on February 12, 2008, 4:36 pmdude i found out something and its weird
if u get the older orthos version which u can choose each core..
and run on each no probs when u stress each core by itself

but when u run the 64k test on both.. problem..

hmm tested with a E6600 no issue..
but on e8400 there is..
interesting..


This post has been edited by cstkl1: Feb 12 2008, 04:36 PM

that batch all it lyn and digital mall.
how i know cause thats the one i bought for my friend..

yeah tested that on my friends UT in singapore last weekend
good stuff..



This post has been edited by cstkl1: Feb 13 2008, 12:17 AM

no
priority 1 will do.

---

Added on February 13, 2008, 11:07 amtsk tsk tsk atleast some screenshot

dude get ure fsb termination voltage right before u pump ure NB

rule of thumb for fsb voltage < vcore.

coult be ure rams/nb.. go 1:1 and reduce strength..
run memtest 256/512mb at least to 1000%.. to stress ure NB.

no suitable voltage.. trial and error.. all mobos are different

reduce ure multi get the bus and mem correct since ure vcore is perfect with orthos small fft..

but i do warn u, this proc abit weird.. its very inconsistent at times even if u pass all the test.. no idea why..
tested it on two mobos and same result on 3 different procs, retail and oem.

thats what the discussion with uzairi was all about.
heck for those that have been used to overclocking 500 bus if i tell this they also found it weird 4ghz and 4.25 okay.. 4.5 orthos okay/not okay .. nb/mem not okay/okay..

i trouble shoot and found whenever u go above 480 bus.. this will start..
small fft goes beserk.. prime 95 also..

This post has been edited by cstkl1: Feb 13 2008, 11:08 AM

NB differs from chipset and strap so cannot help u there

fsb as i said < vcore...

the rule for this differs.. some will say force push nb, get the right fsb and go on to reduce NB
me and a few guys i know we did old school for this proc.. from 400fsb till 506 today we have already excelled it on all the voltage for every 5 mhz increament.
so i usually base on that since all of them passed 2 hours min on orthos and memtest 768mb ( choose the number for max a single program that u game.. crysis 512-1024mb should be okay) for 1 hour..


wei uzairi found one more bug...

1st day orthos stable 8 hours
2nd day on the 10 min.. i was unrar a 4gig file.. orthos rounding error..

---

Added on February 13, 2008, 11:28 amignore it..
could be fsb could be gtl also..

This post has been edited by cstkl1: Feb 13 2008, 11:28 AM

later looking to source a ES..

so then i can compare the bug with retail vs ES vs oem.

nah 467 still okay
its only after 480.

hmm to much vcore for sure..

lower it to load around 1.35v

and ure NB hmm p35 1333 mhz right.
try 1.3-1.6x ure lowest NB voltage
and FSB termination 1.3-1.6x ure lowest fsb termination voltage

tRD 10 on mem ( use memset for now) CL5 1:1..

This post has been edited by cstkl1: Feb 13 2008, 05:23 PM