Curios, you shouldn't need that for Windows 3.x. It brings its own driver (unless you need it for dos games on that image and not actually Windows 3.x).

Dominus wrote:

Curios, you shouldn't need that for Windows 3.x. It brings its own driver (unless you need it for dos games on that image and not actually Windows 3.x).

I guess you have not tried Windows 3.11 using BOOT under DosBox before.
No. Neither Windows 3.11 native driver nor other DOS mouse drivers can find the DOSBOX mouse when BOOT method used – there is no mouse simply. Only mouse.com from Microsoft VPC works.

*cough*, look at my guide for installing Windows 3.x in Dosbox. I messed with running Windows 3.x from a boot image for years.

Just checked again, and no, you don't need a dos driver for windows 3.x to use the mouse. And I've used a different mouse driver with a boot image before - just search the forum and you will find alternatives.
Your problem with Windows 3.x might be:

- When you choose the Tseng drivers when you install Windows, the mouse pointer might not be visible. To fix this go back to the DOSBox prompt, change dir to the Windows directory ("cd windows") , run setup again and then choose the VGA or SVGA drivers that come with Windows 3.1x. Start Windows again (your mouse pointer should be visible now) and then run setup from inside Windows to choose the Tseng drivers.

<- from my guide and it might not be limited to Tseng drivers.

Dominus wrote:

*cough*, look at my guide for installing Windows 3.x in Dosbox. I messed with running Windows 3.x from a boot image for years. […]
Show full quote

*cough*, look at my guide for installing Windows 3.x in Dosbox. I messed with running Windows 3.x from a boot image for years.

Just checked again, and no, you don't need a dos driver for windows 3.x to use the mouse. And I've used a different mouse driver with a boot image before - just search the forum and you will find alternatives.
Your problem with Windows 3.x might be:

- When you choose the Tseng drivers when you install Windows, the mouse pointer might not be visible. To fix this go back to the DOSBox prompt, change dir to the Windows directory ("cd windows") , run setup again and then choose the VGA or SVGA drivers that come with Windows 3.1x. Start Windows again (your mouse pointer should be visible now) and then run setup from inside Windows to choose the Tseng drivers.

<- from my guide and it might not be limited to Tseng drivers.

Hi. I too have run Windows 3.11 for years directly under DosBox successfully. What I am talking about is using BOOT instead of running Windows 3.11 directly. When BOOT used, there is no mouse cursor (S3) or the cursor does not move (paradise) depending on which video card chosen. Only DOS drivers from Microsoft work; and Daum build by ykhwong has no this mouse issue.

READ correctly what I'm writing.

I messed with running Windows 3.x from a boot image for years.

I have NO problem with an image that has Windows 3.x on it and is booted from in Dosbox. Windows 3.x works correctly without having to use the DOS mouse driver you mention.
This is with Dosbox 0.74 and SVN of Dosbox.
I've got several images and ALL work, even my bare minimum image. If you are having problems then you either are NOT using plain Dosbox OR you are loading something that causes problems.

AND YES, on a booted image you need a mouse driver to make it work in other apps and games. BUT NOT for Windows 3.x.

I *JUST* now tested it.

Dominus wrote:

READ correctly what I'm writing. […]
Show full quote

READ correctly what I'm writing.

I messed with running Windows 3.x from a boot image for years.

I have NO problem with an image that has Windows 3.x on it and is booted from in Dosbox. Windows 3.x works correctly without having to use the DOS mouse driver you mention.
This is with Dosbox 0.74 and SVN of Dosbox.
I've got several images and ALL work, even my bare minimum image. If you are having problems then you either are NOT using plain Dosbox OR you are loading something that causes problems.

AND YES, on a booted image you need a mouse driver to make it work in other apps and games. BUT NOT for Windows 3.x.

I *JUST* now tested it.

So it seems an odd issue. I have tried the official one, the one with long file name patch and Daum build. Only Daum build with so many mouse patches (according to its configuration file) needed no DOS driver. Also I am using bare bone DOS with only himem/emm386 loaded. Here are my settings:
Host: Windows 8.1 X64
DOS: Windows 98 [Version 4.10.2222] and its himem/emm386.
Windows: Windows 3.11
Maybe it is this combination causes the problem.

Ah, it could very well be a problem of that Windows 98 DOS. Search around for that, it may be a known issue on resl devices as well.

cyberwalker wrote:

I find Windows 3.11 runs much faster through BOOT.

I find the exact opposite, which makes sense because: A) emulating the CPU instructions of real DOS is slower than DOSBox's internal DOS emulation, and B) disk images are generally slower than local drive mounting.

Are you running Windows 3.11 in Standard Mode or 386 Enhanced Mode? If Standard Mode then you should set core=dynamic and cycles=max in DOSBox settings before starting Win3 for best speed. Note that running EMM386 when booting real DOS causes DOSBox to automatically switch to dynamic core and max cycles with default settings, which could account for the perceived speed increase when booting, but again it only matters for Standard Mode (386 Enhanced Mode also causes the automatic switch).

Dominus wrote:

Ah, it could very well be a problem of that Windows 98 DOS. Search around for that, it may be a known issue on resl devices as well.

Actually Windows 3.11 + Windows 98 DOS works well on physic machine and DosBox (no BOOT). Maybe Windows 98 DOS needs some service the current DOSBOX mouse does not provide, so an additional DOS mouse driver from Microsoft is required to make up for it. I think porting some mouse code from Duam build back to the official source may fix this issue.

ripsaw8080 wrote:

cyberwalker wrote:

I find Windows 3.11 runs much faster through BOOT.

I find the exact opposite, which makes sense because: A) emulating the CPU instructions of real DOS is slower than DOSBox's internal DOS emulation, and B) disk images are generally slower than local drive mounting.

Are you running Windows 3.11 in Standard Mode or 386 Enhanced Mode? If Standard Mode then you should set core=dynamic and cycles=max in DOSBox settings before starting Win3 for best speed. Note that running EMM386 when booting real DOS causes DOSBox to automatically switch to dynamic core and max cycles with default settings, which could account for the perceived speed increase when booting, but again it only matters for Standard Mode (386 Enhanced Mode also causes the automatic switch).

Yes, surely “DOS” is slower, maybe not Windows 3.11. Since no Standard Mode for Windows 3.11, the core should be dynamic. I run speed.exe and the readings were the same literally. While DosBox’s local mounting is kind of innovative, at native host speed, the fastest among current emulators/virtual machines, still smarterdrv.exe balances out some of the disk/file access advantage.

I find through BOOT Windows 3.11 runs more than smoothly. I use IDLE.COM from VPC to lower CPU time, now DOSBOX runs Windows 3.11 (BOOT) with very low CPU usage, almost like VPC. I can even compile SQLite (DPMI) with Borland C++ 4.52 at “lightning speed” (comparing with PCEM, VirtualBox and QMEU). I think overall in DOS emulating/virtualizing DosBox is second only to VPC even with no virtualizing.

To make DosBox/BOOT run Windows 3.11 perfectly on Windows 98 DOS, to me there is just one small issue to be fixed: smarterdrv.exe corrupts hard disk image. I have to turn off Write Cache while using BOOT. The IDE code I guess may need to be revised.

I highly doubt that running smartdrv on an image is more effective than the normal mounting in Dosbox.

Using a boot image is only useful for the odd game that won't run correctly otherwise.

Dosidle works in non booted dosbox as well, btw

cyberwalker wrote:

Yes, surely “DOS” is slower, maybe not Windows 3.11.

Windows 3 uses DOS to access disk drives, in case you didn't know. Adding DOS code overhead and low-level disk access with an image file can ONLY drag down the emulation -- you do get better compatibility, but not more speed.

cyberwalker wrote:

Since no Standard Mode for Windows 3.11, the core should be dynamic. I run speed.exe and the readings were the same literally. While DosBox’s local mounting is kind of innovative, at native host speed, the fastest among current emulators/virtual machines, still smarterdrv.exe balances out some of the disk/file access advantage.

But you previously said Windows 3 is "much faster" when booting DOS, and I have tried to explain the objective facts on why it is not, or at least it shouldn't be with correct configuration.

FYI, the IDLE.COM included in the DOS Additions for VPC 2004 simply inserts a HLT instruction in the INT 28 chain. There are utilities like DOSIDLE.COM that perform a similar function (and more) that work in DOSBox, and I think Daum even includes that program on the internal Z: drive.

You seem set on using DOSBox like any other virtual machine, which is really not playing to its strengths, but suit yourself.

ripsaw8080 wrote:

Windows 3 uses DOS to access disk drives, in case you didn't know. Adding DOS code overhead and low-level disk access with an im […]
Show full quote

cyberwalker wrote:

Yes, surely “DOS” is slower, maybe not Windows 3.11.

Windows 3 uses DOS to access disk drives, in case you didn't know. Adding DOS code overhead and low-level disk access with an image file can ONLY drag down the emulation -- you do get better compatibility, but not more speed.

cyberwalker wrote:

Since no Standard Mode for Windows 3.11, the core should be dynamic. I run speed.exe and the readings were the same literally. While DosBox’s local mounting is kind of innovative, at native host speed, the fastest among current emulators/virtual machines, still smarterdrv.exe balances out some of the disk/file access advantage.

But you previously said Windows 3 is "much faster" when booting DOS, and I have tried to explain the objective facts on why it is not, or at least it shouldn't be with correct configuration.

FYI, the IDLE.COM included in the DOS Additions for VPC 2004 simply inserts a HLT instruction in the INT 28 chain. There are utilities like DOSIDLE.COM that perform a similar function (and more) that work in DOSBox, and I think Daum even includes that program on the internal Z: drive.

You seem set on using DOSBox like any other virtual machine, which is really not playing to its strengths, but suit yourself.

Hi I know Windows uses DOS service. By ‘faster’, I mean overall Windows 3.11 is much smoothly when BOOT is used; slower disk access is not much an issue with Smartdrv enabled. With no BOOT, Windows 3.11 is sluggish and jumping, at least on my settings. I understand your theory, but it is not the actual result.

Prove it

cyberwalker wrote:

Hi I know Windows uses DOS service. By ‘faster’, I mean overall Windows 3.11 is much smoothly when BOOT is used; slower disk access is not much an issue with Smartdrv enabled. With no BOOT, Windows 3.11 is sluggish and jumping, at least on my settings. I understand your theory, but it is not the actual result.

I have observed the behavior you describe with Win3 in Daum SVN builds, and I mentioned it to Ykhwong at the time, but official source has never had the problem AFAIK. There would have been many reports of such a problem if it was commonplace. If you're going to stick with Daum, perhaps the DOSIDLE.COM program on the Z: drive can help, but I don't know for a fact that it will.

Dominus wrote:

Ah, it could very well be a problem of that Windows 98 DOS. Search around for that, it may be a known issue on resl devices as well.

Good news. In DosBox SVN of July 2K15 the Windows 3.11 mouse issue has been fixed; previously I was using the one of April.

Dominus wrote:

I highly doubt that running smartdrv on an image is more effective than the normal mounting in Dosbox.

Using a boot image is only useful for the odd game that won't run correctly otherwise.

Dosidle works in non booted dosbox as well, btw

Not that it is more effective. Just Windows 3.11+ BOOT runs more smoothly with SmartDrv Read-Ahead enabled, than Windows 3.11 in DosBox with no SmartDrv.

SmartDrv also works for pure DosBox DOS. When SmartDrv used, Windows 3.11 runs smoothly too. It is like magic.

Dominus wrote:

Prove it

What to prove?

Dominus wrote:

I highly doubt that running smartdrv on an image is more effective than the normal mounting in Dosbox.

Using a boot image is only useful for the odd game that won't run correctly otherwise.

Dosidle works in non booted dosbox as well, btw

Dosidle.exe is less effective than idle.com from VPC. Again Microsoft did it best. Really disappointed VPC had been disconnected.