Return-Path: Received: from [192.168.1.22]) by mail. (Weasel v1.20) for ; 24 Mar 2001 01:00:40 From: "Digest" To: "OS/2GenAu Digest" Date: Wed, 24 Mar 2001 01:00:40 +1100 (EDT) Reply-To: "OS2GenAu" Priority: Normal X-Mailer: CASMailer MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Subject: Digest No.9 Message-ID: <200103240000.000029G6atmail.> Date:- 24 March 2001 1) Ian Manners Re: [os2genau] More info on 64M RAM 2) Daryl Pilkington [os2genau] ubject: [OT] iPrimus ADSL 3) Craig Brookes Re: [os2genau] More info on 64M RAM 4) Ian Manners Re: [os2genau] More info on 64M RAM 1==================================== Hi everyone re the 64Mb "limit"/"setting" If you have a setting in your CMOS that talks about RAM Type OS/2, or NON-OS/2, DO not set this to OS/2, this is only for OS/2 v2 and lower which used a different RAM access method, so if you are using Warp 3/4/WSeB/eCS, you should always leave this set at NON-OS/2. Changing it will give you around 16Mb of RAM, doesnt matter if you have 256Mb, OS/2 will still only see about 16Mb (varies depending on M/B). Robert is also right about the limit, some motherboard manufactures included the MicroSoft windows memory achitecture only, so for a period, there were a large percentage of M/B's were OS/2 couldnt see more than 64Mb RAM, Daniela's utility also includes a DOS based utility that tells you if you need to apply the patch, or not. These M/B's broke the standard, so strickly speaking its not a fix, but a Patch :-) Also, Scott Garfinkle has included support for these PCB's in the testcase Kernals for FP13, and FP14, the actual fixpack kernals do not include this extra code, as IBM states that it is not their problem that some companies are making "bodgy" motherboards :-) I suspect this will find its way into the FP's anyway, as it can cause drastic swap file usage, and is easily forgotten. The other way to check if your system is accessing RAM above 64Mb, is to run SysInformation/2. Cheers Ian B Manners 2==================================== G'day from the land of iPrimus ADSL. I currently have a nice Telstra linesman checking my ADSL PSTN line, because of a *clearly* audible electrical noise that sounds like a bad connection. I wanted to setup a fax server on the line & initial testing fell in a heap. We don't use the PSTN line for telephony anymore, (we have Orange One, CDMA phones), but my housemate Kate distinctly remembers noise on the PSTN whilst we were still using it for telephony & before the ADSL install. I plugged in a telephone into the PSTN line & rang Primus faults, whilst on hold, clearly audible noise was present. What this means is with this noise, the PSTN line "passed" the ADSL line quality test, mandatory for an install of ADSL. Just shows you problem determination needs to go back to basics. Reality is the the strangest thing of all... -- Regards, Daryl Pilkington //// The PC Therapist, Business Computing Integration O OS/2 Warp, Redhat Linux, DB2 IBM Certified Systems Expert email: darylpatpc-therapist dot com dot au ICQ: 91914134 Tel: +61-2-8902-1300 Mob: +61-425-251-300 Mob SMS: 0425251300.0000atorangenet dot com dot au (120 characters max, send no carriage returns) 3==================================== Hi Ian. I have been following this thread with interest as I have a situation that I hope you or someone can comment on. > period, there were a large percentage of M/B's were OS/2 couldnt > see more than 64Mb RAM, Daniela's utility also includes a DOS > based utility that tells you if you need to apply the patch, or not. I have been using an IBM PC300 6587-79T 133Mhz Pentium for the last 12 or so months which has 144Meg RAM installed in a supported configuration of 1 x 16M DIMM and 4 x 32M SIMMS. The MEMTEST.EXE included with Daniela's patchldr utility outputs the following: ------------------------------------------------------------------- Int12 returns: 639 KBytes Int15 (88) returns: 65536 KBytes above 1M Int15 (8A) returns: not supported Int15 (C7) returns memory table: local memory: 0 KBytes above 1M system memory: 0 KBytes above 1M cacheable mem: 0 KBytes above 1M address space: 0 KBytes above 1M before non-system memory Int15 (DA88) returns: not supported Int15 (E801) returns: not supported Int15 (E820) returns memory list: 639 KBytes at 0 K 146432 KBytes at 1024 K Int15 (E881) returns: not supported ------------------------------------------------------------------- As you can see the Int15 E801 is not supported so I would expect OS/2 to only be able to see 64M of RAM, however from when I installed OS/2 at FP10 or FP11, then FP13 and now FP14, OS/2 always seemed to find the whole 144M. I have also played with the 'inbetween FP' kernal upgrades. Theseus reports: RAM available to OS/2 = 08FA0000 bytes (143.625M). It appears that all of it is being used as 'paging space' by OS/2. (This is the 'proper' usage of the memory.) The BIOS is of an IBM variety so I suspect it may allow OS/2 to detect memory using the PS/2 routines? What do you think? BTW, there is no setting in the BIOS for 'OS/2 memory > 64M'. Certainly OS/2 seems to use all the RAM, the system barely pages, and the Sysbar/2 Pipe Monitor shows available RAM being reduced as I load applications. Once RAM is all used then the system starts to page as expected. The SWAPPER starts at 128M so there's plenty of free slots out there. I assume I have no problem here and the patchldr provides no benefit for this PC. I am still mystified as to how OS/2 detects the 144M. I have played with having the patchldr installed and OS/2 still detects the 144M. From the doco I assumed installing the patch would remove the PS/2 detection routines from the loader so then how did OS/2 find more than 64M if it had previously been using the PS/2 routines to find the 144M? Well, if you're still here then I can finally come to the point of my message... I wanted to eliminate this whole memory detection/BIOS/patchldr issue as a cause for the problem I am experiencing. I originally started to play with the patchldr because of a funny problem I was having with OS/2. This problem has been apparent through all the above mentioned fixpacks, both with and without the patchldr installed, and with different SIMMS at 144M, and with 128M and 96M configurations also. I always boot OS/2 from a powered off state. I can boot OS/2, load up applications, start doing things, there is still lots of free RAM available... anywhere from 50-70M for example, and I can use this system for a couple of months with no problems. Usually I only reboot it for software maintenance, etc, as I hate to boot this machine when it's working fine after a 'good' boot. The applications include things like IBM PCOMS 4.2, Lotus Notes client 4.5.7g, Netscape 4.61 refresh 000818, IBM library reader (numerous refreshes and updates for notes and netscape over the past year). The problem is half the time when I boot OS/2 the system seems to come up fine, lots of free RAM available, but as I attempt to open applications (I have tried the same order of opening applications, etc) the initial ones may start but then subsequent applications fail indicating memory shortages, lack of 'stack space', etc. The typical popup is the one that asks you to verify the path to the program, or open the program object properties. The EPM editor may already have a file or two open already, but then clicking on another file will fail with a popup from EPM indicating lack of memory. Most applications will now not load. Cut and Paste from anywhere starts to fail due to lack of memory. One or two 'little' programs like PMCalc may load... I assume it has a small initial requirement for memory. Currently executing applications will start to indicate memory shortages... notes will fail when opening an email... or the icons will disappear and fonts will revert to plain text... presumably due to lack of memory. Netscape will start to corrupt it's browser window... fonts become HUGE and graphics may disappear. All applications suffer these problems even though Sysbar/2 indicates that 50M RAM may still be available!! Sometimes if I shutdown one or two applications then I can start another application, but the system is still pretty much broken and a reboot is required. It seems like OS/2 either didn't detect the proper memory at boot, or it has lost track of it's memory or stack space, or an application has misbehaved. The funny thing is that if Theseus can be loaded it still shows OS/2 as seeing the whole 144M, so detection of memory seems fine, but then sometimes the 'problem' happens to the first application, like PCOMS, that I attempt to load after bootup (which includes things like SysBar/2 being loaded). I have also spent some time running without the SysBar/2, PU Monitor, and WarpCenter loaded so see if these boys were the cause, but no difference was noticed... I have done a boot, and specifically not loaded one of my usual applications, as a method of eliminating a possible candidate for bad behaviour with memory, but the problem has still showed up so all my usual applications seem well behaved. (also, the problem happening early after bootup seems to clear them as well) You can see why I have taken months and months with this problem, plus when I get a 'good' boot then I like to keep it powered up for as long as possible, so diagnosis is put on hold until the next boot. Oh, almost forgot, half way through this period I also swapped the complete internals of the machine with a spare, with the only common part being the IDE hard drive with my data on it. I then put the latest flash BIOS on the motherboard. I then formatted the OS/2 partition and did a clean install with FP10 or FP11 (I forget), latest Java, etc, and all the applications... So I don't think it is a hardware problem, unless it is something in the motherboard or BIOS of this type of PC that OS/2 doesn't like. Has anyone else ever experience this sort of memory problem with their OS/2? I would expect occasional memory problems that are application related, but not this breaking of OS/2. This is my major problem with this PC, and would appreciate any comments. I have other OS/2 PCs of non-IBM manifacture, with ASUS and SOYO motherboards, and their BIOSs support Int15 E801 so no problems there. These other PCs do not play up as badly as this one, although occasionally there is a funny crash or memory or 'stack space' problem. These seem to be more application related, as I would expect from programs like Netscape, and in most cases a restart of the application is all that is required. A reboot is hardly ever needed unless something like Netscape really blows up. I assume other people also experience Netscape's fragileness? What about the occasional wierd 'stack space' shortages from any application? I assume I don't need to ask about occasional for no apparent reason lockups of the WPS... That's just the WPS right? Having a nice fresh install, or 10 year old OS2.INI and OS2SYS.INI files (well looked after with WPSTools) which are loaded with years of customised settings, doesn't seem to make a difference with these WPS lockups. Thanks for your time, Craig Brookes 4==================================== Hi Craig Wouldnt even worry about RAM detection patches with IBM machines :-) What is the PC300's actual model number ? >will revert to plain text... presumably due to lack of memory. Netscape >will start to corrupt it's browser window... fonts become HUGE and graphics >may disappear. All applications suffer these problems even though Sysbar/2 What graphics driver are you using ? The S3 genuine part or Gengrad, or SDD ? Though your problem sounds more like a memory leakage of some sort, strange faults though given how they happen. Also try it only with the 4 x 32Mb SIMM's for a while, there can still be a mismatch between the DIMM and SIMM's which will not show in normal diags. (Must admit, I havent seen an IBM PC with this combo) >Oh, almost forgot, half way through this period I also swapped the complete >internals of the machine with a spare, with the only common part being the >IDE hard drive with my data on it. I then put the latest flash BIOS on the >motherboard. I then formatted the OS/2 partition and did a clean install >with FP10 or FP11 (I forget), latest Java, etc, and all the applications... >So I don't think it is a hardware problem, unless it is something in the >motherboard or BIOS of this type of PC that OS/2 doesn't like. Power Supply, its amazing how many startup faults that can be traced to the power supply :-) Did you swap the SIMM's and DIMM ? OS/2 is more choosy and less tolerant of RAM faults, also make sure that you are using the correctly matched physical RAM, ie, gold contact socket, make sure the RAM has gold contacts, the same with silver, as the Silver contact RAM modules can have nickel decay on the gold socket contacts, leading to wierd memory problems, though your problem sounds different as if you can get a good boot everything seems to be ok. >Has anyone else ever experience this sort of memory problem with their >OS/2? I would expect occasional memory problems that are application >related, but not this breaking of OS/2. Bit hard to go done the sw or hw route from were I'm sitting but I havent come across any faults like this one in all my years with IBM hardware and OS/2, Windows - yes.... >This is my major problem with this PC, and would appreciate any comments. >I have other OS/2 PCs of non-IBM manifacture, with ASUS and SOYO >motherboards, and their BIOSs support Int15 E801 so no problems there. >These other PCs do not play up as badly as this one, although occasionally >there is a funny crash or memory or 'stack space' problem. These seem to >be more application related, as I would expect from programs like Netscape, >and in most cases a restart of the application is all that is required. A >reboot is hardly ever needed unless something like Netscape really blows >up. I assume other people also experience Netscape's fragileness? What >about the occasional wierd 'stack space' shortages from any application? The latest netscape with FP15 and with the NS46DRAG.DLL file removed seem to be rock solid now, though I found Netscape improved dramatically from FP13 onwards. >I assume I don't need to ask about occasional for no apparent reason >lockups of the WPS... That's just the WPS right? Having a nice fresh >install, or 10 year old OS2.INI and OS2SYS.INI files (well looked after >with WPSTools) which are loaded with years of customised settings, doesn't >seem to make a difference with these WPS lockups. maybe your applications, I've got the same OS/2 on my computer here that I originally loaded on an IBM DX4-100 ValuePoint, just xcopied over to new harddisks, and updates applied, I use WPSTOOLS, and UniMaint as well, its normal for me to only reboot after I've hit a particular JavaScript email in PMMAIL, and thats only so I can restart PMMAIL, I can go months without rebooting otherwise but normally reboot with the application of FixPacks, or powerfailures. Other tricks to try, boot from some OS/2 disks, and delete your swap file, swapper.dat, if you have 144Mb RAM, also try Allow OS/2 to create a new swap file. Also, run a Disk utilitie to make sure you dont have a hidden bad sector that your swap file can/is using, as even though you have plenty of RAM, the swap file will still be used to store DLL's until needed. MAXWAIT=2 (dont use = 1) MEMMAN=NoSwap, protect (Turns off the swap file) THREADS=512 (if you dont have this) Could you send me your c:\config.sys and c:\startup.cmd files as well ? (privately) include your c:\mptn\bin\setup.cmd, and the syslevel.txt file created by running this command. syslevel > syslevel.txt Cheers Ian B Manners End==================================== List Admin Note: The format of the digest will be changing in the next few weeks or so, to allow easier reading, the codings just a bit more difficult but thank you for bearing with me. -------------------------------------------------------------------------------------- To Subscribe/Unsubscribe please go to http://www.os2site.com/list/ ================================================