Jump to content
HWBOT Community Forums

trodas

Members
 View Profile  See their activity

  • Posts

    1129

  • Joined

  • Last visited

 Content Type 

Posts posted by trodas

    trodas - Pentium 1 90Mhz @ 10.8MHz - 10.75 mhz CPU Frequency

    in Result Discussions

    Posted · Edited by trodas

    ludek - so I made todays some memory info screens for you on WinNT, installed on the new shiny AMD K6-III+ CPU: http://valid.x86.fr/ljdyc2

    (yes, that board can run FAST too :D It run like a wind with the 500MHz K3 with even L3 in use :) )

     

    Win_NT_ram_requirments.png Win_NT_ram_requirments2.png

     

    As you can see, a 9592k of memory usage with pagefile, 9580k memory usage w/o pagefile :D

     

    (there is still some optimaliztion that can be done, but it won't be much, I would say 0.5MB at the very best case, if I manage to change the properties of the last memory hungry service from the two ones left? HWbot competition on how overclockers can optimize their Win down to smallest memory usage will be fun :D )

     

    Are you sure you want to repeat that for the WinXP? :D :D :D ...because I'm going for SuperPi 32M, so you have time to think about it :P

    ludek - EP-3PTA @ 207.2MHz - 207.19 MHz Reference Clock

    in Result Discussions

    Posted

    XOC=1 so I don't know... On my today;s submissions says "validated".

     

    Interesting. Did you also used XOC=1 for todays validated subs?

     

    blowen caps has got 5% of manufacturer's declarated capacity

     

    Classic...

     

    but (shock!) is was POSTing without problems

     

    Must be quite stable sturdy board! That is actually a good sign.

     

    Epox 3PTA maybe fly as good as TUSL2

     

    But to match Gradus speeds, you need to go thru plenty of Qimonda rams in order to get the possibility of 3-3-3-7 at 240+MHz ones :) And I doubt that it will be easy to get the board into such clocks w/o extreme mods :) I would stick with the Asus TUSL2-C, as TerraRaptor is using to these absolutelly stunning scores with SDRAMs!

    You can source it easily on eBay: http://www.ebay.com/itm/ASUS-TUSL2-C-motherboard-Intel-Tualatin-1-4GHz-512MB-memory-/281808235276?hash=item419d146b0c

     

    Recapped using 2.5 V caps, will see if they blown when "extreme oc" will come :)

     

    Please don't overvolt them. 2.5V polymers have at most 2.8V that they can take - for brief period of time (at least that is what Nichicon ones can take, but I never tried this...). Use 4V polymers, if you are planing to go bad a*s on the poor mobo, please :D

    trodas - Pentium 1 90Mhz @ 10.8MHz - 10.75 mhz CPU Frequency

    in Result Discussions

    Posted

    ludek -

    I crapped the Batman's Reveange. Mobo was struggling to take me away from setup.

     

    Very sorry to hear that. The work and attempts you put into this... stunning! And sad... Maybe you could get back to it one day - get a proper socket, flash the bios in other mobo and plug it back there and give it a try? I use BIOS Saviour for that. But I'm kinda surprised that such old mobo use such modern bios chip... My TXP4-X use the old, big bios chip... your one looks much more modern.

    What else this mobo have in sleeve? :)

     

    ...witch makes me wonder, if I go the right direction.

     

    Please use contact-cleaner to MEMS and maybe vCORE mod. Take a look at my AMD PR75 score. I used 4.5 V and TEC cooling. It made system ultra stable.

     

    But... on the very same mobo I made this score: http://hwbot.org/submission/2978646_

    So if the rams/mobo works fine at 83x1.5 P90, then why it should not work at 60x1.5 K5 PR75? We know, that the mobo and the overclock of the K5 PR75 is stable, because if it is not, then it will not made this score: http://hwbot.org/submission/2986787_

     

    So I'm all for cleaning the contacts of my AMD K5 PR75 CPU, but it would be better if I can set the HDD right to use DMA in the first place. That would be much better start :)

     

    Of course a Vcore mod will be great. The mobo gives top 3.4V, witch I use. If I can get that to 3.6V, I have no doubt the AMD K5 PR75 will run at 99MHz (66x1.5). I already seen it almost finish the post at such settings... The mobo is mainly targeted for downclocking anyway, so there is not much need for excesive voltages :) Sure, 4.5V + TEC cooling easily "convince" any CPU for some overclock, yes :)

    But mine P90 run on ALL the FSB settings the mobo give - 7.14 - 50 - 55 - 60 - 66 - 75 - 83MHz :)

     

    So memory contacts are innocent there.

     

    there is still possibility to shave some kBy... and I believe that the last running service is not need anyway, so... there is room for optimizing :)

     

    damn you're good :( It that 16bit desktop?

     

    Yep, WinXP did not allow 256colors. 16 or 32bit only (on some cards 24bit as well or - worser - only).

     

    Using NVIDIA GF4/5 cards I seen less ram usage when 32bit colour set.

     

    Could be a bug in their drivers, as 16bit framebuffer obviously consume half of the memory for 32bit one. But I could try it...

     

    Which services you left working?

     

    I told you, only ONE service is left running :) IIRC it is Remote Procedure Call (RPC). And that one could be disabled anyway... but to operate the services I must first start the Plug&Pray service ;) Also I must manually end the userint process that is on when WinXP starts and later exit automatically...

     

    But 23.5MB of used ram is worth it, is not it? :)

     

    I always left 8 (Plug and play, 2 Hard disk services, event log, hid, task skedule and 2 or 3 "important" services) (or nine?)

     

    Yea, the Windows claim they are need, but they are not need in reality. That is like claiming that WFP is need. Of course it is not. And I'm much more proud on my work on the WinNT optimizing. I think that there is chance to get them under 10MB with 800 x 600 x 256colors screen :) WinNT are very tolerant. They - for example - use a little bit less ram and allow deleting the bmp screen logo, that are in background when you login. I really like system that can be tweaked to such extremes :)))

    Only one question remain - will wPrime finally work on it, once I get it to be SP3...?

     

    But wait. If you turn ON pagefile, the amount of memory is incorrect. Turn off the pagefile and then tell me the numbers

     

    Yes, page file is on :) But it will be IMHO even lower, when I turn it off... :))))

    stivut - Core i7 6700K @ 4900MHz - 1093 points Cinebench - R11.5

    in Result Discussions

    Posted

    That happens, lol. Congratulations for your WR, mate! :D ... Why you not deleted the score and repost it back under correct Cinebench? :) What you can do alone should not need to be done by admins...

     

    Or they will be doing nothing except fixing other people mistakes (and possibly introducing own to the mix...).

     

     

    ...the only way you cannot delete the score is, when you participated with the score in competitions, IIRC... Is this the case?

    Just how far back in Windows do benchmarks run on? (Like, Windows 3.x, or?)

    in Benchmark software

    Posted

    Windows 2000 run even the latest CPU-Z, so validation should not be a problem. For Win98se you need to dial back to CPU-Z 1.51. For WinNT 4.0 SP1 you need to dial back to CPU-Z 1.30.

    Also you can include programs like TSCWin and TestCPU, like I did: http://hwbot.org/submission/2978646_

     

    Pifast and Wprime should run

     

    PiFast does run on WinNT 4.0 SP1 w/o a glitch, but wPrime does not. It require the msvbvm60.dll library and that library, even in the oldest version, require the WinNT 4.0 SP3. Then it should work, but I did not yet get around to test that, but it is likely. W/O ActiveX (on for example highly optimized WinXP) you cannot install the CPU-Z libraries, but that it is.

     

    However, for a i386 / i486 systems I'm affraid I have bad news for wPrime. Current version 1.55 have a nasty bug, that prevent longer that 24h calculations to take place. It check for time and when the time "move back", then it write "Cheat detected" and stop the calculation. There is no known way for me to get around it. I just set the time to 0:01 and run the calculation. Unless it exceeds 24h (1024M on AMD K5 PR75 @ 90 is getting close with 17h 48min: http://hwbot.org/submission/2986787_ ), then all is fine. But it ignore the day, when it check if the time is not dialed back during the run, so unless your tests with wPrime is under 24h - OR - there is a fix for that, you have to strike wPrime from the test list, because I'm affraid that on i386 even 32M run could be over 24h...

     

    So, this is just something to be carefull off.

    Test/Tryout: Skylake 5G Tweaker's Challenge

    in Skylake/Kaby Lake/Coffee Lake (-X) (Z170/Z270/Z370/X299) OC

    Posted

    Massman -

    ES is no problem

     

    zeropluszero -

    It'd be nice if the submitter perhaps mentioned it in the comments, so we know that we're mere mortals and dont need to compare ourselves.

     

    First at all, the Engineering sample is always mentioned in the CPU-Z validation, for example:

    http://valid.canardpc.com/6cr7g9

    "Type Engineering Sample"

     

    ...so with mantadory CPU-Z validation you will always have this information. As for the "mere mortals dont need to apply" comment, well, that is despondence. Yes, compared to first batches, the ES are usually a bit better, but not that much. This is not like you suggesting a difference between man and God (when we dwell into the superstitions).

    And compared to later batches are the ES mostly par to par or even worser that normal retail CPU's.

     

    This is not a situation back in 2000 or so, when there are P4 ES CPU's with unlocked multipliers given as promo to certain people and you simply cannot buy a P4 with free multiplicator. That was, of course, absolute and decisive advantage (IIRC the multi can be cranked all the way up to x50...) for overclocking. Now these ES are mostly overrated.

    Almost all of the "Golden" chips are from normal retail stock, just classic binning process...

    trodas - Pentium 1 90Mhz @ 10.8MHz - 10.75 mhz CPU Frequency

    in Result Discussions

    Posted

    PS. as expected, the K5 PR75 @ 90MHz crashed on the end:

    Super_Pi_32_M_crash_K5_PR75.jpg

    ...witch makes me wonder, if I go the right direction. To put long story short, I stabilized P90 @ up to 125MHz (more this board did not manage) for SuperPi 32M with PCI latency bump from default 32 to 42.

    So on AMD K5 PR75 I reacted similary, bumping the latency up and up... This crash was at 96, so I set 102... and even floppy drive have problems now. That got me thinking, if I did not do it entierly wrongly, so I try default 32 with the CF card and we see :(

     

    ...

     

    So I used the pause to check on the memory usage: 23.5MB it is:

     

    TXP4_X_memory_usage.jpg

     

    What do you say, ludek? :) And there is still possibility to shave some kBy... and I believe that the last running service is not need anyway, so... there is room for optimizing :)

    trodas - Pentium 1 90Mhz @ 10.8MHz - 10.75 mhz CPU Frequency

    in Result Discussions

    Posted

    Thank you, guys! Yep, finally I get the validation with correct CPU clock :) It need to disable DMI in ini file (DMI=0), as Franck suggested, and then it display the correct clock. I did also a little trick - I start first CPU-Z with DMI=0, then changed DMI to 1 to show the bios mobo correctly in the mainboard window :) There are all the images I took: http://s923.photobucket.com/user/ax2cz/library/10MHz?sort=3&page=1

     

     

    ludek -

    Finally you did it !

     

    Thanks, mate! :)

     

    Why so anormous amount of mem?! 128 would be great even without pagefile

     

    Why remove the memory anyway? I'm happy to overcomed the 256MB limit of i430TX and I did not need to touch the rams at all now :) Unless I get better rams (hint: ram ranks makes difference), they stay there. I cannot get them to lower timings, so unless I get more SDRAMs with 2+ ranks, then these rams stay there :) See Aida64 report: http://depositfiles.com/files/o06t2paka ( http://www.mediafire.com/?8xub1mvfp868497 )

     

    Congrats my friend. How much mem is used by system when displaying desktop? I was able to reach 49 MB on XP with pagefile off

     

    I believe I get notably lower. My WinNT (optimized by me) build is under 11MB even with 1024x768 res (256colors), but the WinXP build will be around 30MB or so... I will check that for you, when SuperPi 32M again crash on the cursed K5 at the end... :)))

     

     

    Massman -

    Nice!

     

    Thanks! I believe I could get lower too... :)

     

    I'm surprised it only takes 15 minutes to boot Windows XP!

     

    Well, I optimized the Win to the bone, so... hopefully I could get it even faster with the CF card :) W/O the keyboard mod it took nearly 21min, see the vid bellow.

     

     

     

    Dead Things -

    Now you just need a PODP3V to get it even lower. Good work!

     

    Indeed. 7.14MHz is possibility. Just get a Pentium Overdrive, force it to x1 multiplier and now the 7.14MHz will happen :) Should not be a problem... Also I use patched bios now, that support AMD K6 3+. These CPUs have software multiplier control. Only for DOS, yes, but one can create a boot selector with SetMul, where I set x2 multi and get 14.3MHz for AMD CPU.

    The only catch is - I lack the K6 3+ :( But maybe next year... :)

     

     

    RomanLV -

    You replace quartz 32 to 12MHz, I understood correctly?

     

    Nope. What you see with the 12MHz crystal is mod to make input devices work, when the FSB is forced to REF/2 clocks, as the PLL chip support: http://www.vogons.org/viewtopic.php?f=46&t=42426&start=20

     

    ICS_9169_PCI_bus_lock_possible.jpg

    http://web.archive.org/web/20000919074447/http://www.icst.com/pdf/9169c-27.pdf

    ( http://uloz.to/xL5L3QiW/ics-9169-pdf )

     

    I simply introduced my own clock to the PS/2 controller Holtek HT6542b, as it supports:

    http://www.alldatasheet.com/datasheet-pdf/pdf/64463/HOLTEK/HT6542B.html

    ...because that way it is not tied up to the FSB clock = mouse and keyboard works fine at 7.14MHz :)

     

    How fast (_really_!!!) are the clock (watches)? I want a video with seconds windows near real watches If all is correct - you are the best!

     

    Sure the clock is correct and I can prove that to you very easily :) As you suggest - real clock + video. Could be done, if you did not see how I did it. The PLL chip ICS 9196 support the REF / 2 (FSB = reference clock (~14MHz) / 2) as you can see from the documentation.

    So, there you have it. And I seen 2MHz clock for CPU and there is a possibility on validation... but it will be tough :)

     

    Albeit no, you're the best in any case )

     

    Well, to convince you, there are problems, what S3 Trio3D have with 3.75MHz PCI bus:

     

    7_15_x_1_5_P90_superlow_clock.png

     

    prtscr_cpu_z.jpg

     

    So it is not that easy. There you have a video from it:

     

    So it was not all peachy and easy... :)))

     

     

    ludek -

    I think I give up... I tried lots of stuff. 486 is not validable. CPU-z works on Windows 2K - big thanks to RomanLV for idea!

     

    I told you so, Win2k is fine.

     

    I have funny little Socket4 mobo. It's "Intel Batman's Reveange" LOL. Please take a look at it: LINK http://abload.de/img/adrklq.jpg. Sadly the BIOS was corrupted and I tried everything!!! Here is a photo what I had to do because of clock is ticking and I had no PLCC socket

     

    Crazy. Nice try, tough. Can you match 7.14MHz...? Or 2MHz? :) That would be a challenge and I know I can do it :)

    SPI 32 M @ i486 - oldschool OC

    in General overclocking

    Posted

    Well, if that does conform you a bit, then I tell you, that todays for the 4 or 5th time SuperPi 32M run crashed for me on the AMD K5 PR75 of mine at the end, when saving the score... I gut WinNT to bones to get good results, but it always crash at the end and I find no reliable way to restore the UDMA2 HDD access, witch is also key to speed.

     

    It is either 20MB/sec with little CPU usage (mind that this is K5 PR75 @ 90MHz, so not exactly fast CPU by any standards: http://postimg.org/image/qx7kmwknx/ ) or around 2MB/sec with 85+% of CPU usage. That suxx.

     

    Quess what impact that have on the score.

     

    ...

     

    So I sort of give up and try a XP run. While WinNT gives me under 1h per loop, WinXP need 1h 10+min per loop... and it probably crash again...

×
×
  • Create New...