Announcement

Collapse
No announcement yet.

Video card holding puter back from 133MHZ FSB?

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Video card holding puter back from 133MHZ FSB?

    I have a PII 400MHZ that will run stable as hell at 4x120=480... (on an ASUS P2B-f MB, 128MB PC133 Crucial RAM, old Matrox Mill II, CL AWE32) but for some reason the thing will not run stable at 496. It will, however, do most everything fine until I run the quake demo2 about 2 times. I believe the temperature is about 45 degrees C, from my ASUS CPU temp monitor. My question is: if my comp will run fine until I get into quake, is this an indication that the graphics card cannot take the higher bus speed? If I go to a 133MHZ FSB, at a 44MHZ PCI bus speed the system will not even POST. In fact, when I turn the power button on the power supply comes on, but there is NO video (this supports my theory). Contradicting my theory is the fact that when I run it at a 133FSB, but at a 33MHZ bus the machine does the exact same thing. Both times at a 133MHZ FSB my computer will not turn off by hitting the power button, I have to shut off the power strip to get it to turn off. I have a VEK12 on the way, so better cooling is in the works if thats what you think the problem is.

    UPDATE****************

    hehe Here's an update: I just got my VEK12 from cooltech, and my processor runs at max 38 degrees C, a good 7 degrees lower that before with the retail setup. This didn't solve my problem, though. At 496MHZ Quake just exits the program and I'm back into window clean as a whistle after two timedemo demo2's. I can run windows fine. In fact, I'm typing this at 496 right now. I did try one more thing: when running prime95 on my 480MHZ system the system runs rock solid, no errors. Now when I run it on the 496MHZ it halts the torture test after one iteration claiming there has been a round off error. Does prime95 perform similar calculations that would be done for quake, or what? I know prime95 doesn't use the graphics card much...
    Well, my RAM is rated at PC133 through Crucial Tech. I think it SHOULD handle the 124MHZ speed. I tried changing my L2 latency with WCPUL2. I got it to change from 7 to 8 and the computer lasted a lot longer - running through at least 5 timedemos before exiting. When I turn the L2 cache to OFF the computer goes through about 12 timedemos, then quake exits. I cannot set the L2 any higher without the system rebooting. I'm not quite sure what the problem is here... Are there designated values the L2 will run at? Thanks (BTW Im back at 480 4 now)

    Any Thoughts????


    -neo


    ------------------
    Asus P2B-F, 400MHZ Pentium II (O/C to 480!!), 128MB Micron PC133, Maxtor 4GB, SB AWE 32, Sony Trinitron 17", Old Matrox Video Card



  • #2
    Hi neo,

    Have a look at your clock dividers (AGP & PCI)

    those should be set to 1/2 (AGP) and 1/4 (PCI) when using 133MHz FSB.

    ------------------
    Cheers,
    Maggi

    Not too slow running Asus, Intel & Matrox based system ...
    Actually boosted by the Millenium G400 32MB SGRAM DualHead 8-)

    Comment


    • #3
      In short: yes, Maggie. The mobo jumper settings (yuk!) that control the FSB speed also control the PCI bus at the same time. There are no sepatate multiplier settings on the P2B-F for the PCI bus. For instance: one settting allows a 133FSB/43.3PCI Bus, while another setting denotes a 133FSB/33PCI Bus. Neither setting will allow my computer to POST. By the way, my current video card is, in fact, a PCI card... a Millennium II
      Any more suggestions (please)?

      -neo

      ------------------
      Asus P2B-F, 480MHZ Pentium II (4x120), Global Win VEK12 hs/fan, 128MB Micron PC133, Maxtor 4GB, SB AWE 32, Creative DVD 2x, Mitsumi CDR 2x/8x, Sony Trinitron 17", Old Matrox Video Card, and a redhead with a pair of 36Cs (O/Ced to 38Cs)




      Comment


      • #4
        sure ...

        try disabling the CPU caches (L1 & L2) from the bios ...

        first the L2, then the L1 and finally both of them ...

        BUT: it might take ages for booting without caches !!!

        On my former Cel300A I was able to boot at 133MHz having both caches disabled, but it took ~5 minutes.
        Also my CPU mark dropped from ~38 to 0.5 !!!

        If you can boot with any or both of the caches, you found your bottleneck.
        Then you could consider fiddling with the settings for CAS, RAS and all those latency timings.

        ------------------
        Cheers,
        Maggi

        Not too slow running Asus, Intel & Matrox based system ...
        Heavily boosted by the Millenium G400 32MB SGRAM DualHead


        Comment


        • #5
          Ok, I'll try. It would be unbelievable to me if I could push my system to a 133MHZ FSB, but I would really be happy at a *stable* 124MHZ. Any suggestions for increasing stability? Same I guess. I've tried setting my RAM down to to 7ns timing, but that doesn't help. What values over an L2 latency of 8 will the cache take? I can't even get it to take higher than 8 w/o it rebooting.

          I know what you mean about the decreased performance. When I disabled my L2 cache the CPUmark went from 37.5 to 21.0. BTW, how did you get a mark as high as 38 on that 300a? What was your multiplier clock set at?

          -neo

          ------------------
          Asus P2B-F, 480MHZ Pentium II (4x120), Global Win VEK12 hs/fan, 128MB Micron PC133, Maxtor 4GB, SB AWE 32, Creative DVD 2x, Mitsumi CDR 2x/8x, Sony Trinitron 17", Old Matrox Video Card, and a redhead with a pair of 36Cs (O/Ced to 38Cs)




          Comment


          • #6
            hehehe ...

            FSB 103MHz > Celery @ 464 MHz (4.5 x 103)

            Leave your AGP divider at 1/2, because underclocking the bus a tiny little bit is better than oc'ing it more than 30 %.

            I didn't dig that depply into the latency settings .. I just made a couple of tests and then decided to let my rig run at 100/103MHz FSB .... the next step would have been 112 MHz (asus rev 1.0) and I made it boot with altered L2 cache latency, but it was pretty unstable.

            ------------------
            Cheers,
            Maggi

            Not too slow running Asus, Intel & Matrox based system ...
            Heavily boosted by the Millenium G400 32MB SGRAM DualHead


            Comment


            • #7
              Hello ?

              Neo ?

              Where are you ?

              Still fiddling with those settings ?

              ------------------
              Cheers,
              Maggi

              Not too slow running Asus, Intel & Matrox based system ...
              Heavily boosted by the Millenium G400 32MB SGRAM DualHead


              Comment


              • #8
                Calm down mags - It's only been a couple of hours!

                ------------------
                Cheers,
                Steve

                Using dodgy beta stuff in not that bad a PC with all sorts of bits n bobs in it helping to find ET calling home...

                ICQ: 29468849

                Comment


                • #9
                  Hey Stevie ... I was just kidding

                  I remember very well my excitement when discovering further options to increase my performance.

                  ...

                  Btw, one thing I haven't mentioned yet is to increase the VCore from 2.0V to 2.2V ...

                  ------------------
                  Cheers,
                  Maggi

                  Not too slow running Asus, Intel & Matrox based system ...
                  Heavily boosted by the Millenium G400 32MB SGRAM DualHead


                  Comment


                  • #10
                    Maggi knows a way! I haven't got the links handy... there's another thread going - look at the end of <A HREF="http://forums.gagames.com/forums/Forum58/HTML/000215.html" target="_blank">this thread</a> and you'll see what you need to do.

                    ------------------
                    Cheers,
                    Steve

                    Using dodgy beta stuff in not that bad a PC with all sorts of bits n bobs in it helping to find ET calling home...

                    ICQ: 29468849

                    Comment


                    • #11
                      I JUST CAN'T BELIEVE IT. Great News!!! I downloaded the softFSB and noticed it's recent support for nwer ASUS P2B-F MBs. Anyway, first I tried the original ASUS P2B-F setting, and next I tried the P2B-F(new). The first setting froze my desktop///restarted. The 2nd attempt: SUCESS! Running stable at 124MHZ FSB, 31MHZ PCI bus!!! So far of course... 20 quake timedemos later, still going strong!! I even bumped it up to 1280x1024 on my Mill II, stable! The best part?? My temp only went up 1 degree C. Running at 32 degrees C at idle, and 37 degrees C at 100% CPU! Celebration time baby! Next I will benchmark this puppy, then perhaps 133MHZ? heeh -->always wanting more. I think I'll run prime95 all night just to see!

                      Ecstatically yours, neo

                      ------------------
                      Asus P2B-F, 496MHZ Pentium II (4x124!!), Global Win VEK12 hs/fan, 128MB Micron PC133, Maxtor 4GB, SB AWE 32, Creative DVD 2x, Mitsumi CDR 2x/8x, Sony Trinitron 17", Old Matrox Video Card, and a redhead with a pair of 36Cs (O/Ced to 38Cs)





                      Comment


                      • #12
                        Well, I'm actually at work right now so I can't mess with it until later tonight (5 or 6 hours). I suppose I have to disable them at 480 MHZ, then go up from there. What can I do to my computer to allow stable performance at a 124MHZ bus? If my computer does not run stable with both the L2 and the L1 caches disabled, where does this leave me? I am really almost there at 496, just can't do anything really CPU intensive (hehe). It boots, POSTs, windows loads fine, everything... until I run prime95 or quake. I might try pulling cards also (as much of a pain this will be...)

                        One off the subject question: I am looking to get a new larger HD. I really like all I've heard about the new 18GB Western Digital 7200RPM HD. My question is: will I still be able to see all the speed the WD can do if I have turned off UDMA? Since the average sustained transfer rate is only like 15MB/s at most one would thing that 33MB/s Max burst rate is a bit of an overkill anyway.

                        BTW, my MB doesn't support voltage changes I don't think, unless you know of a way....?


                        -neo

                        ------------------
                        Asus P2B-F, 480MHZ Pentium II (4x120), Global Win VEK12 hs/fan, 128MB Micron PC133, Maxtor 4GB, SB AWE 32, Creative DVD 2x, Mitsumi CDR 2x/8x, Sony Trinitron 17", Old Matrox Video Card, and a redhead with a pair of 36Cs (O/Ced to 38Cs)








                        [This message has been edited by neo (edited 06-23-99).]

                        Comment


                        • #13
                          Hmmm...I was under the impression that no MoBo could divide the AGP in 1/2, only 2/3. This is limitation within the BX chipset itself and the 1/2 AGP divider won't be avaliable untill the Camino chipset is released.
                          I also would like to know how the new G400 can cope with an o/c'ed AGP bus. I know my old G200 could handle it great, i could even run it at 112mhz FSB at a 1/1 AGP ratio!
                          But alas, my current CL TNT won't do AGP speeds over 83mhz:-(.

                          Comment


                          • #14
                            I'd have to guess that the G400 would be as good if not better than G200 at overclockability. I've started a few threads that asked this question and got this answer. You are correct about the AGP multiplier, at least on my mobo. I can set the AGP to 1:1 or to 2:3. 2:3 is the obvious choice for a 124MHZ bus, so that is what I'm going to do. Right now my Mill II is running on a underclocked PCI slot at 31MHZ. I think that since Matrox builds quality card with great tolerability they can be run pretty far from spec just fine.

                            Update on my situation: The last time I played with the L2 cache I apparently left the latency set to 6(faster than usual). The damn computer ran fin at 124MHZ FSB at L2=6, but I bumped it back up to 7 just for GP. Ran the prime95 torture test all night (7 hours) with NO, count them, NO errors. Not sure if softFSB does something different than changing jumpers, but the comp is now runnning faster, stabler than before. So I added softFSB to my startup group and it runs great! Did briefly try 133, froze in both PCI situations (33 and 44 i think). Of course, that was at a fast L2 latency, so there is still hope yet. This is exciting. I ran CPUmark99 and got a whopping score of 38.8! According to the intel website that clocks faster than a PIII 500!!?? hehe

                            Someone needs to help me with those questions about the WD hard drive... if you please?

                            ------------------
                            Asus P2B-F, 496MHZ Pentium II (4x124!!), Global Win VEK12 hs/fan, 128MB Micron PC133, Maxtor 4GB, SB AWE 32, Creative DVD 2x, Mitsumi CDR 2x/8x, Sony Trinitron 17", Old Matrox Video Card, and a redhead with a pair of 36Cs (O/Ced to 38Cs)





                            Comment


                            • #15
                              Ok, here is the last update to my saga, which has ended in joy. The problem with my setup was, IMHO, the voltage was set just too low for a stable 496. I decided on a whim to up the voltage using the pin taping technique. First I went to vcore=2.2v, 10% off the normal 2.0v. 496 ran surprisingly stable, no hang ups, freezes, or errors. I decided to try and take it to the next "level" by setting the FSB at 133. Before this point, mind you, the computer would freeze with the softFSB set to 133. I did it and blam! no freeze! After running a few programs(quake/prime95) I noticed FSB=133 is just as unstable as FSB=124 *WAS*. With this in mind, I attempted to tape a different pin and get 2.4 volts out of my vcore. Whoa, when I restarted the voltage monitor said the voltage was 2.8volts!!!! I nearly shat my pants. Decided this was erroneous and went into windows to see... it was correct! I was running my poor processor to the bone at 40% over voltage spec, so after a few fatal exception errors I shutdown and saw I had taped the wrong pins (EEK). I was using electritions tape, and this ws a BAD idea, it left black gunk on my leads(damn). I cleaned them off with isopropyl alcohol, which did a respectable job. The next day(today) I received some teflon tape in the mail (ordered it just yesterday --->wow!) and tried the whole pin taping again. This time success! Boots up, set the FSB to 133MHZ and watched the temp rise and rise as I ran prime95 (with no errors I might add). It finally topped off at 50C degrees! Scary thought, huh? So I pulled of my VEK12, washed it clean and then noticed an elevated graphite square it has on the contacting surface of the heatsink. "How stupid is this", I thought. It minimizes the heat tranfer area from the hotplate to just that small square. Anyway, I scratched that puppy off of there!! Used acetone (nail polish remover) to get the glue gunk off of there, and now it is slick and flat. Reapplied thermal grease and yes! The max temp is now 40C degrees, MUCH more acceptable I must say. So here I am, running my PII 400 at 532, cool and fast. The other good thing is the fact that my ASUS MB supports a PCI bus of 33MHZ at FSB=133, so everthing runs like a charm. I hope the AGP slot can take the 88.67MHZ bus with my new card...

                              thats my story!
                              thanks to all those who gave me advice!

                              -neo

                              ------------------
                              Asus P2B-F, 532MHZ Pentium II (4x133 at 2.4v), Global Win VEK12 hs/fan, 128MB Micron PC133, Maxtor 4GB, SB AWE 32, Creative DVD 2x, Mitsumi CDR 2x/8x, Sony Trinitron 17", Old Matrox Video Card, and a redhead with a pair of 36Cs (O/Ced to 38Cs)






                              Comment

                              Working...
                              X