Backporting newer browsers to Win9X with KernelEx

[Goodmaneuver]

DMA buffer affects all I/O operations which includes the video buffer. The settings are just a guide and are what I am using to give improved performance. System.ini will over-write the device manager reserve DMA buffer setting and visa versa so do not re-tick the 64k box. There was a diminishing improvement return on the video when going above about 50MB. You can use 44000 max file cache if you want but I have no trouble with 512000, 16384 will be too small other members have suggested a maximum of 132MB. On winME installation of ram has to be<2GB, that means 128k less will work. I have 1.5GB installed on this machine.  X-Setup Pro will show the buffer size which the system is using and what they say is not fully correct as I have explained. If members are interested in X-Setup Pro some settings are deadly so back-up beforehand. (CPU boost works)

[cov3rt]

19 hours ago, Goodmaneuver said:

DMA buffer affects all I/O operations which includes the video buffer. The settings are just a guide and are what I am using to give improved performance. System.ini will over-write the device manager reserve DMA buffer setting and visa versa so do not re-tick the 64k box. There was a diminishing improvement return on the video when going above about 50MB. You can use 44000 max file cache if you want but I have no trouble with 512000, 16384 will be too small other members have suggested a maximum of 132MB. On winME installation of ram has to be<2GB, that means 128k less will work. I have 1.5GB installed on this machine.  X-Setup Pro will show the buffer size which the system is using and what they say is not fully correct as I have explained. If members are interested in X-Setup Pro some settings are deadly so back-up beforehand. (CPU boost works)

interesting, i wish i had this information earlier, although i still do not understand a lot of things, for the device manager part, do i still need to initially set it to 64k or can i just edit the system.ini and it should take care of the DMA aspect from there, but i would still need to manually enable dma on the storage device?, ( although unofficial service pack 3.64 automatically enables DMA on storage devices too ). 

is there a particular reason why past 50 MB or so, there are diminishing returns? diminishing returns as in, you don't gain AS much benefit but still there is some benefit? can you also explain to me what the pagebuffers means? how do i know what is the best setting. maxphyspage and minfilecache and maxfilecache are straightforward, however i am not understanding these other modifications. also the site below lists 1-256 as a range for the dma value, is this just a random range they put, because 256 would only be 256 kilobytes, less than one MB only, or am i calculating it wrong? 

i also noticed a 32 bit setting, i forgot exactly where i got the info and / or what it does and where you put it, are you aware of this setting?

http://smallvoid.com/article/win9x-16-bit-dma.html

so i researched more and noticed a source saying the dma buffers for windows 95 should be less than 16 MB, does this apply to windows 98 too? my question altogether is are there any compatibility problems with having more than the 64KB setting from device manager and / or what would be the best setting in system.ini that wouldn't cause compatibility issues and would operate the "fastest", also factoring in that "32 bit setting". i don't want to use any "tweaks" that would be detrimental or cause compatibility issues. i use those specific vcache and maxphyspage settings in my builds because that's the most reliable way that seems to work i suppose. 

https://www.dell.com/community/Desktops-General-Read-Only/I-O-subsystem-error/td-p/319016

Edited January 18, 2019 by cov3rt

[Goodmaneuver]

@cov3rt,The tick in DM will still be ticked. If you re-tick the DM driver settings 64k box it writes over the system.ini DMABufferSize=2048 to DMABufferSize=64. You can run with 64k size buffers if you want, make PageBuffers=700 and the total DMA size will be similar to the setting I suggested. When viewing video at high monitor resolutions the up sizing of the DMA buffer is necessary to improve frame-rate on browsers, they may not have GPU memory employed for the resizing of the output. Yes diminishing gains thats all, remember it is only 50MB so no trouble allocating memory. PageBuffer is RAM memory working space and PageFile is HDD memory working space both controlled by the virtual memory manager. DMA buffer size affects all I/O, storage devices included. You can choose Double Buffering (double the size) for storage devices in MSDOS.sys DoubleBuffer=1, which I run as I have SCSI for more speed. Yes 256 is 256kB, X-Setup Pro says a range up to 64kB so discrepancy. Trial and error are my settings. The memory allocations that are affected are controlled by vmm32.vxd I think and vmm32.vxd will be different between OSs but 16 bit should be able to allocate up to 2GB of total memory. The answer is for the bottom short cut given is for a general non startup problem where there is a full buffer. The Win95 in question is different to my WinME if it using ios.vxd and vcache.vxd as I am not using them. What the member is asking is how to remove the program, all VXDs auto start that are in the system IOSUBSYS folder I think (check bootlog.txt), so removing the offending VXD would help and VXDs are not always compatible between the OSs.

You can change the ThreadingModel from Apartment to Both right throughout the registry, appears to have no ill effects and can speed up performance especially when there are many modules on single thread. Edit from top to bottom with 9x systems. Make sure a "match whole word only" approach is taken when using a text editor and just edit Classes let the system up date the Local Machine settings otherwise a registry corruption may occur.

My system.ini settings that may have a performance influence including vCache *WinME*. Conservative swap file usage is a disadvantage in my experience (creates problems).

Recently upped the settings to:

[386Enh]
ADIXctlLevel=1
ConservativeSwapFileUsage=0
Paging=on
PageBuffers=512
DMABufferSize=256
LocalLoadHigh=1

[MathCoprocessor]
FPUFlags=1

[vCache]
MaxFileCache=512000

Edited August 6, 2019 by Goodmaneuver
settings change

[roytam1]

just refreshed my KM74g archive with sha384 support.

 

[deomsh]

On 1/18/2019 at 3:14 PM, Goodmaneuver said:

My system.ini settings that may have a performance influence including vCache *WinME*. Conservative swap file usage is a disadvantage in my experience (creates problems).

Recently upped the settings to:

[386Enh]
ADIXctlLevel=1
ConservativeSwapFileUsage=0
Paging=on
PageBuffers=512
DMABufferSize=256
LocalLoadHigh=1

[MathCoprocessor]
FPUFlags=1

Missed this post.  

Please can you explain ADIXctlLevel, ConservativeSwapFileUsage and FPUFlags?

Edited January 9, 2020 by deomsh
Language

[Goodmaneuver]

ADIXctlLevel=1; I am not sure about this setting and its understanding.

I got this setting from Win2K. Application Data Interface transfer control logic level. Programs invoked by XCTL commands run one at a time as part of a single task but is multi tasking. When a user selects a menu item the program should issue an XCTL command to transfer control directly to that program. The first program which receives the control directly is at highest logical level which is Level 1. More investigation is required with multi tasking and different tasking methods. This setting may not make any noticeable difference.

ConservativeSwapFileUsage=0

If conservative swap file usage is employed then the maximum RAM memory usage is used before switching to the HDD swap file. This caused problems with certain programing and so I did not set it to 1 anymore.

[MathCoprocessor]
FPUFlags=1
 
This setting came with Borland C. This was originally set to 0 but I thought that 1 made a performance improvement. Most of the following has been quoted from https://xem.github.io/minix86/manual/intel-x86-and-64-manual-vol1/o_7281d5ea06a5b67a-197.html

x87 FPU Floating-Point Exception Mask Bits

The exception-flag mask bits (bits 0 through 5 of the x87 FPU control word) mask the 6 floating-point exception
flags in the x87 FPU status word. When one of these mask bits is set, its corresponding x87 FPU floating-point
exception is blocked from being generated.

Zero is Invalid Operation, 1 is Denormal Operand. These are the only settings that can be used as a BSOD I think will occur with higher settings, see the above link for what other settings are.

It is the Floating point value rounding and how it is achieved that is what is at stake here. See https://en.wikipedia.org/wiki/Denormal_number The following is a quote from this link.

"Some systems handle denormal values in hardware, in the same way as normal values. Others leave the handling of denormal values to system software, only handling normal values and zero in hardware. Handling denormal values in software always leads to a significant decrease in performance. When denormal values are entirely computed in hardware, implementation techniques exist to allow their processing at speeds comparable to normal numbers;[3] however, the speed of computation is significantly reduced on many modern processors; in extreme cases, instructions involving denormal operands may run as much as 100 times slower.[4][5]"
 

[siria]

roytam1 said:
> just refreshed my KM74g archive with sha384 support.

Thank you for updating! A few more of those pesky modern ciphers :-)

@Readers with old browsers, who only see empty rectangles instead of MSFN crosslinks:
The link goes to here
https://msfn.org/board/topic/180462-my-browser-builds-part-2/?do=findComment&comment=1175953

(KG74, KMG74, KM-Goanna74, K-Meleon-Goanna 74, officially for Win2000, or as fallback browser for Win98se with KernelEx2016)

Edited January 11, 2020 by siria

[sparty411]

I'm trying to use KM74g on Windows 98SE with the latest KernelEx updates, but I'm getting the error "K-Meleon failed to initialize. Please reinstall." I set KernelEx compatibility mode to Windows 2000, and have tried all of the other available Windows compatibility versions as well. Any ideas?

[schwups]

Make sure you have Core Updates 4.5.2016.18 or higher. Don't use update 17. Otherwise maybe a missing dll.

 

Edited January 22, 2020 by schwups

[sparty411]

Looks like I hadn't installed the latest updates correctly. All is well now! Thanks!

[Bruninho]

On 11/5/2018 at 12:33 AM, Heritage said:

I created a Greasemonkey script which should hopefully get YouTube working again on Firefox.  Just be aware that it's quite limited in what it can do (sometimes slow to load, fullscreen doesn't work, some videos still don't play, and there's no quality select).

Tested On the Following:

• Adobe Flash Version 10.3.183.90
• Firefox 9.0.1
• Greasemonkey 0.9.22.1

Installation Link

GitHub Page

Where can I find Greasemonkey 0.9.22.1?

[Wunderbar98]

Paste and go in browser, download takes a while to initiate.

http://archive.org/download/Firefox_Legacy_Collector_Dump/legacycollector.org.tar/srv/www/legacycollector/public_html/firefox-addons/748/greasemonkey-0.9.22.1-signed.xpi

[schwups]

KM74G: I no longer have runtime errors when I set msvcr80.dll in the KM Program Folder to Base (Kexbases,Kexbasen).

[Jody Thornton]

17 minutes ago, schwups said:

KM74G: I no longer have runtime errors when I set msvcr80.dll in the KM Program Folder to Base (Kexbases,Kexbasen).

So what are you able to browse on Windows 98 SE with that configuration?

 

[schwups]

I can open almost any site. This isn't the problem. Ok, YouTube doesn't work anymore with it. It's not about browsing, but the setting avoids the runtime error with the following crash for example after adding a bookmark. The bookmark plugins Netscape and IE Favorites become useful here.