FSB downclock mod on the intel GL960 and GL40 --- useful info for PLL Modders

Discussion in 'Hardware Components and Aftermarket Upgrades' started by naton, Aug 21, 2011.

Thread Status:
Not open for further replies.
  1. RickiBerlin

    RickiBerlin Notebook Deity

    Reputations:
    609
    Messages:
    979
    Likes Received:
    18
    Trophy Points:
    31
    Exactly for this purpose I have a computer. I need this only to test and write posts. 1x1GHz would therefore be sufficient. I think we are all mad. :D

    The impetus for these tests was to you, and the two of you posted topics. (Theme 1 and Theme 2)
    So my thanks go to you. (And-nando4/kizwan/moral hazard and and and) :notworthy:

    I'm not sure what contribution you think. Perhaps from a different thread?
    In any case I was and am aware of this behavior. Thanks anyway for your explanation.

    Again, I'm not sure what you are referring to. No "HDX Dragon" owner ever tried a "T9900". But I see no problem with this CPU. Why should not this work? If "X9100" work, then work "T9900" and any other "1066 CPU". (Apart from Quad)
    One problem is, of course, the multiplier locked. But that's what "SetPll" and more. For me would be "T9900" not an option.

    Unfortunately, I have the "reference frequency adjustment" is not yet understood to 100%.
    The key lies in front of me (your little table), but it still did not click in my head. See your own pictures here in comparison:

    [​IMG]

    "133MHz to 166MHz" is the same "SocketMod" as "266MHz to 200MHz."
    Why I just do not understand the simple dynamics.

    Next:

    [​IMG]

    "166MHz to 200MHz" is the same "SocketMod" as "133MHz to 266MHz."
    The solution is simple, but I do not see it. Help me to understand it!

    But did you read that right. I am now writing in short sentences. Then there are fewer errors.
    I own PM965. At the beginning of my tests, I needed the "SocketMod 266MHz to 200MHz" for a working "X9100" on PM965. In the bios the correct frequency is displayed. (2300MHz) Everything was quite normal.

    After many more tests, I suddenly noticed a wrong frequency display in the bios. Under the same conditions (SM-266 to 200) now "1909MHz". This is a "SM266MHz to 166MHz" correspond.
    The OS, however, shows the correct value (2300MHz) to. For this reason, I also noticed the error until very late. Who's going so often in his bios.

    Further investigation of this failure rendered the following findings:

    Each CPU is working suddenly without "SocketMod" with 166MHz on my PM965.
    Whether "X9100" or "T8100" or "T4400". The bios always shows the corresponding value of 166MHz. The operating system was formed from the corresponding value of 200MHz. So it's not a problem but a mystery.
    A 166MHz processor is normally displayed in the bios. Tested with T5450.

    For a correct display in the bios I need now a "SocketMod 166MHz to 200MHz."
    Whether "X9100" or "T8100" or "T4400".

    My guess is a damaged socket. What damage would produce this behavior? Certainly only a "CPU Pin" no contact with the base. But which one?

    The fact is, "X9100" would not work normally without "SocketMod" on PM965. For me, it makes it now though.

    Another factor is, each "SocketMod" carries over into the OS. (CPU frequency)
    With me no longer. In the OS is always the fundamental frequency 200MHz.


    Changing the subject:
    Since I the principle of "SocketMod" still do not understand 100%, I can currently only use your pictures.
    I am missing the following "SocketMod" for PM965:
    First "266MHz to 133MHz"
    Second "200MHz to 133MHz"
    Third "166MHz to 133MHz"
    Please show me these modifications!

    I think it is too simple as I can understand it.

    How does this SocketMod? (200 to 166)?

    I know only "266 to 166". This corresponds to this mod:

    [​IMG]
     
  2. naton

    naton Notebook Virtuoso

    Reputations:
    806
    Messages:
    2,044
    Likes Received:
    5
    Trophy Points:
    56
    I spoke about the T9900 as I used to have one and I know that it won't post on the GL40 and GL960 without the mod from 266 to 200 FSB. So I'm wondering why the X9100 can boot in your laptop without any mod !!?

    It is quite simple. You have to keep in mind that your reference is the CPU and not the socket. So the starting point in the case case of the X9100 is the 266MHz FSB, and BSel-2-1-0 = L L L.

    In both case we connect BSEL1 to a Vcc hole to force BSEL1 to be high. It is straight forward.

    because in both case BSel0 is connected to Vcc and no change is made to BSEL1 and BSEL2.

    This is odd.

    As I explained in my last post the 166 to 200 mod doesn't do anything since BSEL0 is already set to low for the X9100, T8100 and T4400.

    It is strange that you can boot the x9100 with the 166 to 200 mod. It seems that for some reason your BSEL1 is locked in High instead of Low !!!

    BSel 2 1 0 | Frequency
    ---- L L L | 266
    ---- L H L | 200
    ---- L H H | 166
    ---- L L H | 133

    266MHz to 133MHz: BSEL0 + Vss
    200MHz to 133MHz: BSEL0 + Vcc and BSEL1 + Vss
    166MHz to 133MHz: BSEL1 + Vss

    Note:
    For the X9100 this two mods give the same result: 266MHz to 133MHz and 200MHz to 133MHz.

    The T8100 and T4400 will not boot with the 166MHz to 133MHz mod, because this is the same as doing 200 to 266 mod.
     
  3. AlphaHeX

    AlphaHeX Notebook Consultant

    Reputations:
    115
    Messages:
    258
    Likes Received:
    0
    Trophy Points:
    30
    I will look at it. The question is how to verify that certain BIOS component is APIC initialization piece of code?
     
  4. RickiBerlin

    RickiBerlin Notebook Deity

    Reputations:
    609
    Messages:
    979
    Likes Received:
    18
    Trophy Points:
    31
    Because my socket has an error. At the beginning of my tests everything behaved exactly as described by you.
    At that time worked for each of you described "SocketMod" correctly. Both the BIOS and in the "OS".

    This had changed in the meantime. Of course it is not easy to understand. One knows the behavior of each "SocketMod" but not the exact behavior when has a "CPU Pin" no contact with socket.

    It even looks like this as if "BSEL 1" and "BSLE 0" locked on high. Because each CPU works so without "SocketMod" with 166MHz. (equivalent to LHH)
    And for this reason do I need a "SocketMod" to "LHL" for each CPU in order to get to 200MHz. (for X9100 and T8100)
    Just a reminder, the error affects only the BIOS. The "OS" is always correct.

    Yes I know it works for me but because of the "socket error".

    By your endurance and re-explanation, I finally understood this principle.
    I thank you for it. :notworthy: With new information, I'll write here.

    For you as a conclusion: Your statement is 100% correct. Everything is just a fault of my system.

    Because of my supposed "Socket Error" This is certainly work for me. I will try this soon.

    Here you had accidentally chosen the wrong quote from me.

    Sorry I had to restructure the text. I hope you could understand everything.
    Once again my thanks to you!
     
  5. moral hazard

    moral hazard Notebook Nobel Laureate

    Reputations:
    2,779
    Messages:
    7,957
    Likes Received:
    86
    Trophy Points:
    216
    @RickiBerlin, I can't help but I can tell you that I had a similar problem.

    Before my MSI GT627 Died, I was playing with BSEL mods.
    After several tests, I had noticed that my SP9400 would boot with a 200mhz FSB (instead of 266mhz originally). This was without a BSEL mod.

    I have no explanation.
     
  6. RickiBerlin

    RickiBerlin Notebook Deity

    Reputations:
    609
    Messages:
    979
    Likes Received:
    18
    Trophy Points:
    31
    Apologies to other users. My theme is really something off Toppic. ;) There should be afraid of anybody.
    My tests were really hard and have nothing to do with the regular "SocketMod".
    (15000 on / off switching operations of Testmoterboards, 130xCPU change to the socket, some operations 30xSBios Flash)
    All this because of "QX9300" try on "PM965".

    Thanks for your feedback. :) Those are good prospects. (Moterboard death) :rolleyes:

    For me there is no problem until now. On the contrary. I do not even need "SocketMod" more and can use any CPU immediately. :D

    Interestingly, the only reason would be to know.

    The first thing I had suspected the bios. After 10 different flash methods that do not change but I gave up. (including hardware reset and cmos battery overnight distance)

    Whether it is the "socket" is? Maybe it's a completely different cause.

    My attempt at "QX9300" I always had "SocketMod 266 to 200". If I wanted to test a new bios, I had another CPU (800MHz CPU) use. (only for Flash process)
    For this, I removed the "SocketMod" Of course not every time.
    The 800MHz CPU worked without any problems even with existing 266zu200Mod.
    Whether this was causing this issue?
    It is not the time to understand. Perhaps there comes the knowledge.
     
  7. naton

    naton Notebook Virtuoso

    Reputations:
    806
    Messages:
    2,044
    Likes Received:
    5
    Trophy Points:
    56
    If a BSEL is stuck in low one can say that the pin is broken or is not touching the contact in the hole. In other words, I would have said that you socket is damaged.

    I can't find any logical explanation to why your BSEL(s) is stuck in High. This is really odd !!!

    Am I missing something?
     
  8. RickiBerlin

    RickiBerlin Notebook Deity

    Reputations:
    609
    Messages:
    979
    Likes Received:
    18
    Trophy Points:
    31
    Just a reminder:
    BIOS Frequency display my PM965 without "SocketMod" with each CPU = 4x166MHz.
    Whether "X9100 (266MHz CPU)" = 1917MHz [x11, 5] or "T4400 (200MHz CPU)" = 1833MHz [x11])

    In the operating system without additional tools, however, frequency of 4x200MHz. (X9100 = 2.3 GHz or T4400 = 2,2 GHz)

    Now it gets crazy:

    For test reasons the "QX9300" I had a "SpeedStepOFF SBios" on the system. With this BIOS, the multiplier is locked on x6. This is for information only.
    I'm not sure what role it plays in the mystery. Of course, this lock with "ThrottleStop" are repealed.

    Today I tested exactly "this SocketMod" here described by "naton".

    Result:
    "X9100 SLB48 in bios" 2300MHz (11.5 x200MHz) / "in the OS" = 1200MHz (6x200MHz)
    "T4400 SLGJL in bios" 1833MHz (11x166MHz) / "in the OS" = 1200MHz (6x200MHz)

    Following this and using the same "SocketMod" I made a "SBios Flash" on an unmodified bios. CPU during the flash was "X9100 SLB48".

    After successful flash and before the first "OS startup" I went into the bios to the "System BIOS Information"

    Bios frequency indicator for "SLB48 X9100" = 1917MHz (11.5 x166MHz)
    Now I started the OS.
    In the OS CPU frequency = 2300MHz (11.5 x 200)
    So far, everything ok. I finished the OS again and went into the bios.
    Now been shown in the BIOS suddenly 2300MHz. This remained the case after 30 other motherboard starts above)

    Now I removed the "SocketMod BSEL1+VSS"
    Result: No change in frequency for "X9100 SLB48"
    (Without "SocketMod" Now 2300MHz in the BIOS and OS) - equivalent to 11.5 x200MHz

    The "T4400" I also always testetet parallel under the same conditions.
    Result: For the T4400, there was no change. In the BIOS always 1833MHz (11x166MHz).
    Of course, the "OS" with this "normal bios" then always 2200MHz (11x200MHz).

    Now the new mystery:
    I now put "X9100 QHBQ" (ES version) "on the socket without Mod
    Result: In the BIOS 1917MHz (11.5 X166) / OS = 2300MHz (11.5 x 200)

    I could not believe it. I changed "QHBQ X9100 (ES version)" and "X9100 SLB48" repeatedly against each other.
    The result was the same. I started several times the "OS", no change.
    "Bios frequency display" no "SocketMod":
    "X9100 SLB48" - (OEM version) "= 2300MHz
    "X9100 QHBQ" - (ES version) "= 1917MHz

    It looks as if a Biosflash is responsible for the mystery. But what exactly happens?

    --------------------------------------------------------------------------------------------------------------------------

    EDIT:

    I tested further. Without renewed Biosflash.

    First "X9100 QHBQ" - (ES version) installed with BSEL0 + VSS (166zu200).
    "Bios frequency" = 2300MHz ("without SocketMod" it was 1917MHz)

    Then "X9100 SLB48" - (OEM version) without "SocketMod"
    "Bios frequency" = 2300MHz

    Then "X9100 SLB48" - (OEM version) with BSEL0 + VSS (166zu200) installed.
    "Bios frequency" = 2300MHz

    Then "X9100 SLB48" - (OEM version) without "SocketMod"
    "Bios frequency" = 1917MHz :confused: :eek: :confused:

    Now the socket is firmly back on 166MHz on your CPU. This happened only a "SocketMod of 166zu200".

    Who can solve this riddle?
     
  9. apostmax

    apostmax Newbie

    Reputations:
    0
    Messages:
    9
    Likes Received:
    0
    Trophy Points:
    5
    Now I am gonna get a bit out of topic, but Ricki, can you tell us exactly the steps needed to run a QX9300 on a PM965? even if it utilizies just one core I bet it outperforms X9100.. is any special bios/mod involved or it just runs after having done the BSEL mod? thanks
     
  10. RickiBerlin

    RickiBerlin Notebook Deity

    Reputations:
    609
    Messages:
    979
    Likes Received:
    18
    Trophy Points:
    31
    I've solved the puzzles. :)

    The only reason was faulty contacts on the socket.

    For mainly affected the contacts "BSEL1", "BSEL0" and 1x "VCC" were.
    See picture in the spoiler.

    [​IMG]
    All mysteries I have described can thus also explain exactly.

    Here is the explanation:

    If 'BSEL' has no contact, then corresponds to the "bios CPU frequency" of a "SocketMod of BSEL to VCC" (corresponds to high)

    This "CPU frequency" is not transferred to the chipset. For this reason, the CPU appears there with a normal "200mhz BSEL". Therefore, we the "socket contact error" not recognize the operating system.

    Examples of contact error: (all examples without any "SocketMod")---->

    First only "BSEL1" defective (equivalent to "SocketMod266zu200" but without effect in OS)

    "X9100 / 1066 CPU" = "BIOS 2300MHz" / "OS2300MHz"
    "T4400 / 800 CPU" = "BIOS2200MHz" / "OS2200MHz"

    Second only "BSEL0" defective (equivalent to "SocketMod266zu133" and "SocketMod200zu166" but without effect in OS)

    "X9100 / 1066 CPU" = 133MHz is not possible!
    "T4400 / 800 CPU" = "BIOS1833MHz" / "OS2200MHz"

    Third "BSEL1" and "BSEL0" defect (equivalent to "SocketMod266zu166" and thus indirectly "SocketMod200zu166" but without effect in OS)

    "X9100 / 1066 CPU" = "BIOS 1917MHz" / "OS2300MHz"
    "T4400 / 800 CPU" = "BIOS1833MHz" / "OS2200MHz".

    Precisely this assumption I had mentioned here:

    By the way I've found a "1066 CPU" can not underclock to 133MHz.

    @naton, here is you made ​​a slight mistake. Written in "RED" should "VCC" to be!

    Edit:

    As you see, you had to post this one flaw.
    It's the opposite. BSEL without contact corresponds to "high".
     
Loading...
Thread Status:
Not open for further replies.

Share This Page