Have a real 48k Speccy? yes ? I need your help.

edited April 2010 in Development
Im looking to run this quick test on as many (only) 48k Spectrums as possible

files moved

http://www.zxspectrum4.net/temp/timing_type/

I want to try and collect a snapshot data from the results so...

If you know the board issue, the CPU, ULA you have inside please report the results and the info to richard@zxspectrum4.net

Thanks could you email the results throught in the format

Example.
MODEL: 48K Spectrum
REPORTED TYPE: TYPE1
CPU: Zilog Z80A
ULA: 6C001E-7
BOARD ISSUE: 6A
SERIAL: 123-4567890
OTHER NOTES: extra components, memory type etc.


Note: if you dont know the internals it doesnt matter still report the result.

Thanks in advance Rich
Post edited by Richard_Butler on

Comments

  • edited March 2010
    This looks like spyware.
  • edited March 2010
    shhhhhhhh

    more results please :)
  • edited March 2010
    Got many results in yet? ;)
  • edited March 2010
    are you kinding I cant move from results.....

    ok I lied, not many, well... all of 2 ... !



    Come on people I dont belive no one has a real speccy anymore ?
    spare 5 mins to run the test!

    Spread the good word!

    Ah maybe ill be better posting on ebay "just before you sell it, please run this test"
  • edited March 2010
    Report sent.
  • jpjp
    edited March 2010
    Any chance of a plain old TZX? - my machine resets right after the turbo loader has finished...
  • edited March 2010
    are you kinding I cant move from results.....

    ok I lied, not many, well... all of 2 ... !



    Come on people I dont belive no one has a real speccy anymore ?
    spare 5 mins to run the test!

    Spread the good word!

    Ah maybe ill be better posting on ebay "just before you sell it, please run this test"

    I'm trying to load the file into my speccies, but failed, my new pc is unable output enough volume I guess. Strange that Grey +2 is loading (by wires connected to the board) just fine with the same sound card output..

    I'm still fighting for it, though a normal loader (without turbo block) would be fine.
  • edited March 2010
    jp wrote: »
    Any chance of a plain old TZX? - my machine resets right after the turbo loader has finished...

    Thanks to woodys update on tape2Wav the wav is now at standard speed, try this see how you get on, :smile:

    thx
  • edited March 2010
    I've forwarded this request to the folks at ecss and Speccy.org. I know you specified real 48Ks only, but if you need feedback from Timex users just let us know.
  • edited March 2010
    zxbruno wrote: »
    I've forwarded this request to the folks at ecss and Speccy.org. I know you specified real 48Ks only, but if you need feedback from Timex users just let us know.

    thanks Bruno, as we are tabulating the results may as well include a list of timex, by real spectrums I actually ment no emus even clones give some interesting results as long its details what the machine is!
  • edited March 2010
    FOR All:

    Quick Note what this is about, Ive not had time to a do doc on it yet so this will have to do ill post the real tests soon......



    I originally wrote a suite of tests to check the contention values I added to our emulator to match the table at the wiki scratchpad from Woody, Phillip and others good work people.

    The problem I thought was with somany numbers could easily get a number wrong and so the tests were built to check the timing though I quickly found the only emulator giving the correct "early" results was fuse and helped with some emulator "fixes" ;) you dont have to thank me!

    But no emulator appears to be producing the correct "late" chatteristics of the late machines I have here, So I wrote a quick check which differs from the "fuse-test" currently used to check for "early/late" timing to find some late machines in the wild a quick way to compare the differnce with the check is to run the checker on an emulator see what result you get specEmu, Fuse "type1=early" if the early timings for the tighest loop add up, If theres an option for late timing tick it and run again the type will still be 1 showing the emulation of "late" its not a correct

    Needless to say although I dont emulate anything other the 48k speccy I will do a diff test suite for the 128,+2,+2A,+3 so at least emulation autors have tools to check the calculations.

    Ok this has got out of hand and a bit of a side step of what i was doing but still with no real handle on the "Early/Late" it would be nice to have some more to go on hopefull someone will solve the puzzle.

    Does all this make a difference, probably not.
  • edited March 2010
    Arda wrote: »
    I'm trying to load the file into my speccies, but failed, my new pc is unable output enough volume I guess. Strange that Grey +2 is loading (by wires connected to the board) just fine with the same sound card output..

    I'm still fighting for it, though a normal loader (without turbo block) would be fine.



    UPDATE: added a ".tzx" version at standard speed for those using this format to load into their speccies

    thanks folks.
  • edited March 2010
    So I wrote a quick check which differs from the "fuse-test" currently used to check for "early/late" timing to find some late machines in the wild a quick way to compare the differnce with the check is to run the checker on an emulator see what result you get specEmu, Fuse "type1=early" if the early timings for the tighest loop add up, If theres an option for late timing tick it and run again the type will still be 1 showing the emulation of "late" its not a correct

    [strike]Interesting. And where can one find this checker? It would be interesting to run it on a few emulators I have... :) [/strike]
  • edited March 2010
    The checker is the purpose/program at the start of this thread.

    Important Note: Please keep in mind this program wasnt built to check "other peoples emulators" and create a witch hunt or annoy but for Authors to debug their own emulators if they care too, and find how many variants are in the wild (at least a ratio) the exact reason for some of the results are still unknown and the impact they have is also not known to cause any issues.

    The tests were infact inspired and helped along by other authors as an aid for development and debugging.


    Without the people and emulators ive already mentioned this wouldnt be here and I wouldnt be discussing it.

    they know who they are "the ones that had no life" did I say that, ill wait the abuse to start :)
  • edited March 2010
    Post count in this thread won't make any difference :p
  • edited March 2010
    ok I lied, not many, well... all of 2 ... !

    C'mon chaps! We're getting a bad name here!

    Surely there are more than just 2 of us who own a real speccy??? And surely there are more than just 2 of us who can spend 5 mins loading up the test program and report the results via mail??

    For those wondering what the big deal is... it turns out that there a few spectrums in the wild that behave a bit differently than expected of them with regards to an issue known as "late timings". The effect of late timings can generally be seen only in some demos (although some games could be affected too I suppose) where the rendering of some advanced effects is slight off.

    Previously, some solid investigative work by various emulator authors had helped pin-point how late timing machines work (and thereby emulate them) but recently Richard has discovered that some more work might be required in emulation to get this *exactly* required. His test program therefore reports machines which have early timings and late timings and this is something that can be used by emulator authors to determine what exactly is going on.

    So, pull up yer socks lads! You can do it! :)

    Disclaimer: I don't have a real speccy to test it on. Sorry! :D
  • I posted some test I've made on my 48Ks, and posted them on speccy.org:

    http://www.speccy.org/foro/viewtopic.php?f=11&p=20221
  • edited March 2010
    Thanks could you email the results throught in the format

    Example.
    MODEL: 48K Spectrum
    REPORTED TYPE: TYPE1
    CPU: Zilog Z80A
    ULA: 6C001E-7
    BOARD ISSUE: 6A
    SERIAL: 123-4567890
    OTHER NOTES: extra components, memory type etc.

    Thanks.



    Sorry not had time to make a submission form I shall try for when we conduct the real tests.


    Woody: yes more posts, shooting up the list now!
  • Richard, I've just sent you the results.

    In the speccy.org post you could even see screenshots of the results.
  • edited March 2010
    These are my results (already emailed):

    As you will see, The issue 1 reports "unknown type" as well as the Inves Spectrum+ clone. My two 6A boards report "Type 2" and the rest of them report "Type 1".
  • edited March 2010
    thanks for results,

    Please dont post the results in the thread else we'll get lost with numbers flooding the questions just email the results then we can make a clear table and leave the thread for questions/queries ;)

    thanks again.
  • edited April 2010
    The tests have now been started and also new thread here

    http://www.worldofspectrum.org/forums/showthread.php?p=436464#post436464

    ;)
Sign In or Register to comment.