Jump to content
HWBOT Community Forums

Caos Wolf

Members
 View Profile  See their activity

  • Posts

    41

  • Joined

  • Last visited

 Content Type 

Posts posted by Caos Wolf

    Skyline - Athlon 64 X2 3800+ (Manchester) @ 3360MHz - 25sec 531ms SuperPi - 1M

    in Result Discussions

    Posted

    Thansk GRIFF!

    I've a few days off work and already have plans ?
    1st one done today (even thoug it was probably the best of the bunch)
    Next time 4200X2 ?

    Athlon 64 X2 3800+ Manchester LDBHE 0601TPMW     3541 SS 1,7 3300 bench
    Athlon 64 x2 4200+ Manchester LDBFE 0610SPMW      
    Athlon 64 X2 4400+ Toledo CCBWE 0609DPMW      
    Athlon 64 X2 4600+ Manchester LDBHE 0540WPMW      

    TerraRaptor - Pentium 3 1.4Ghz S @ 2123MHz - 1min 1sec 198ms SuperPi - 1M

    in Result Discussions

    Posted

    The score is of 2001,in that time there was few people on ln2 on the cpu,this man had cpu and ram on ln2?he's a genius...

     

    But this genius...did not think sl6by is better for superpi?

     

    @ gradus and @ terra:you have a mission,destroy this record :D

     

    At that time Japanese 'clocker were a lot ahead if compared to the rest of the world.

     

    Also weren't you the one that said that the smokes of the cpu pot would also cool down the ram in the order of -30 degrees?:)

    TerraRaptor - Pentium 3 1.4Ghz S @ 2123MHz - 1min 1sec 198ms SuperPi - 1M

    in Result Discussions

    Posted

    This score is very strange,I don't understand if the memory is 1:1 or 4:3,may be 1:1 with 2 2 2 7 9.

     

    Uff...find it :)

    http://holicho.lib.net/bn_02.htm (pretty much at the middle of the page right under the 59s picture)

     

    4 : 4 : 1、2-2-2-5/7

     

     

    pc100@219 2 2 2 7 9...bh-5?lol

    The stock rating ins't usually a precise meter of the overclocking ability of a chip :)

    If these rams are really able to do 200 2225 @ air I don't see why shuldn't be able to do 220 @ LN2

    TerraRaptor - Pentium 3 1.4Ghz S @ 2123MHz - 1min 1sec 198ms SuperPi - 1M

    in Result Discussions

    Posted

     

    Good link,thank's.

     

    But,are you sure than ram was 1:1 and not 4:3?I don't understand japanese.

     

    Here you can find the translation in italian http://translate.google.it/translate?hl=it&sl=ja&tl=it&u=http%3A%2F%2Fholicho.lib.net%2Fbench%2Fbench_104.htm

     

    ...but doesen't specify the latency of the ram.

     

    Gradus (just a couple of post above) confirmed that latency e frequency were 22257 1:1 219 cooled by LN2.

     

    Also there is the italian translation for the database http://translate.google.it/translate?hl=it&sl=ja&tl=it&u=http%3A%2F%2Fwww.ace-sup.com%2Fmemory%2Fmtec.html

    TerraRaptor - Pentium 3 1.4Ghz S @ 2123MHz - 1min 1sec 198ms SuperPi - 1M

    in Result Discussions

    Posted · Edited by Caos Wolf

    Hello guys,

    I'm sorry to intrude again but I find something that you may find interesting.

     

    First:this is the complete setup that Holicho used for sub 1m

    2hqw8ly.jpg

     

    I think that he was 1:1 whit the mems at 22257 or (more probably) 22279

     

    I made a reaserch for those mems (mtec 0103u) and came up with this table (in japanese)

    http://www.ace-sup.com/memory/mtec.html

     

    If I've understood it correctly those mems should be able to do 200+ mhz 22257 on air.

     

    I think those tests were made with air cooling beacuse the result of the mems i have experiece with are -pretty much- the same of mine.

     

    ...so if you are looking for the holy grail of SDRAM those should be the real deal.

     

    Also that site is a great database for sdram frequency.

     

    Hope I was helpful :)

    Gigioracing - SDR SDRAM @ 227MHz - 227.34 MHz Memory Clock

    in Result Discussions

    Posted

    Tonicom PC166 @ 215 3-3-3 ln2 1m...hunix/hynday imo better

    PC 150 Kingmax @ 200 2-2-2 ln2 1m(dunno about 3-3-3, some 205-210)...not best too

     

    I make some posts about ram, when will i have more statistic.

     

    Great man, thanks for sharing.

    I'm looking forward to see your comparsion.:)

     

    Maybe one of this days I'll try my hyundai with AT-H chips.Never thought these thing would be good :D

    Skyline - Pentium 3 500Mhz Katmai @ 560MHz - 1334 marks PCMark 2004

    in Result Discussions

    Posted

    Priority change is allowed, don't worry about that.

     

    What i was preoccupied about was the version used.

     

    If I understand it correctly "newer" version of pcmark04 make the application run in normal priority even if set to realtime from taskmanager,so using a older version (with priority =realtime) give a substantial advantage.

     

    Totally legal. Good job and thanks for the explanation.

     

    My pleasure man :)

    Skyline - Pentium 3 500Mhz Katmai @ 560MHz - 1334 marks PCMark 2004

    in Result Discussions

    Posted

    To Mr.Scott:

    As I said yesterday I ran some more test to identify the reason to this high scores.

     

    I realize that (probably cause of my habit with spi) I've set pcmark to run in realtime and that gave me a boost of ~300 points.

     

    I wasn't sure (actually I'm still not sure) if this was legit so I made a quick search and came out with this:

     

    This is from futuremark.

     

    pofpcmark04 wrote:

    Our official policy is that PCMark04 should be run with normal priority.

     

    This is because normally one does not set the priority of any program (from the Task Manager or otherwise), but let's the program decide the priority. In other words, normally users do not even know about priority classes.

     

    When you change the priority of PCMark04 to Realtime in the Task Manager it says:

    "WARNING: Changing the priority class of this process may cause undesired results including system instability. Are you sure you want to change the priority class?"

     

    Notice the warning about undesired results including system stability!

     

    Technically this means that if you have rebooted your system and exited all the unnecessary programs from the background, the only processes running are PCMark04 and some system services. If you then set PCMark's priority class, system services may not be run when they are needed. This can have the above-mentioned undesired results including system stability. This also means that PCMark04 is not doing what it is meant to do.

     

    Since I am now ranting, I will continue some more.

     

    If you run your processor heavily overclocked, it is no wonder programs seem to behave badly. Actually, it might even be so that the programs are not to blame, but the processor may not be working as it should. In my honest opinion, this is why heavily OC'd HT processors show better PCMark scores when HT is disabled. Or has anyone had a better PCMark score with HT disabled than HT enabled when both test runs have been run without OC'ing or running PCMark04 with realtime priority class?

    (From hardwareheaven 2003)

     

    And this:

     

    Using realtime-priority has been intentionnally prevented in Build110 as it produces results which don't reflect the way applications normally run. Best Regards, Futuremark Support

     

    So I looked into the executable to see which version i was using and find out it's the 1.0.0.

     

    Then I gave a look to the hwbot rules for pcmark04 and:

     

    Allowed benchmark versions: All versions

     

    and

     

    Disallowed tweaks/cheats : Any software or human interaction altering the perceived speed of the benchmark program, tricking it to believe it ran faster

     

    Should run the exe in realtime be considered "tricking it to believe it ran faster" ?

     

    From rog.asus.com forum:HOWTO: Benchmark for HWbot

     

    12. PCMark04

    Set Windows for "Best Appearance" and make sure Classic view is not on.

    (Dual Threaded)

    set affinity for 2 cores (Disable any others if possible to gain higher overclock)

    set priority to real time.

    Run Benchmark

    repeat several times for best results.

    Save Screen Shot Showing Score, CPU and Resolution.

     

    I've searched also into the HWBForum but didn't find anything :(

     

    In conclusion:I'm still not sure if this submission shuld be considered legit.

    If only one person tells me that this result shuld not be considered regular I'll remove it ASAP (my life don't depends on that :D) and submit my best result with priority set to normal (somthing like 1000 points)

     

    I know i wrote a wall of text (in a super mario-ish kind of english :D) for a 15 years old cpu,an almost irrilevant benchmark and for one useless gold cup, but I always fell the need to be extra sure of what I'm doing.

     

    I prefer to be the incontestable last one than a suspectful first.:)

×
×
  • Create New...