dosbox-pure icon indicating copy to clipboard operation
dosbox-pure copied to clipboard

Windows 98 startup sequence slower with DOSBox Pure 1.0 Preview 4

Open Ric-82 opened this issue 2 months ago • 5 comments

Hi,

I noticed that the startup of the Windows 98 graphical interface is noticeably slower with version 1.0 Preview 4 of the core (about 7-8 seconds longer compared to the core version provided on Steam, which I believe is still Preview 1). In addition, the first startup of Windows 98 after updating the core required the installation of additional drivers from the Windows 98 disk.

Is this expected behavior, or is it only happening on my side?

Ric-82 avatar Nov 12 '25 13:11 Ric-82

First thing you should check is the driver settings on the NE2000 Compatible network adapter. Its resources should be set to Interrupt Request = 10 and Input/Output Range = 0300 - 0301F, otherwise startup can hang for a bit while it's trying to communicate with it.

NE2000 Compatible network adapter resources configuration

schellingb avatar Nov 12 '25 14:11 schellingb

Thank you very much for the quick reply.

In my case I’m already using a stripped‑down version of Windows 98 (98lite), so the network adapter and related components are not present at all.

With the previous core version, the startup with 98lite was actually very fast, but now something seems to slow it down. I can only assume it might be related to the driver that Windows asked me to install at the first boot after updating the core, but I have no idea what exactly it could be.

Ric-82 avatar Nov 12 '25 15:11 Ric-82

Perhaps I have identified the component that Windows 98 asks me to update when I start it for the first time with core version Preview 4. In fact, I don’t think it is a driver but rather a catalog. The installation procedure points to mscdrom.cat, which should be located inside catalog3.cab on the Windows 98 installation CD.

It seems to be related to the CD‑ROM driver (please correct me if I’m wrong).

Is this behavior consistent with the latest core updates? Could there be a correlation between this and the slowdown during Windows startup (which persists even after installing these files from the installation CD)? Or perhaps between this CD drive update and the issue I described here, where the core seems to fail in handling VHDs?

Ric-82 avatar Nov 12 '25 17:11 Ric-82

I think a mistake in commit 8b97a5d caused a conflict between the first emulated CD-ROM IDE device and the C: hard drive, causing Windows 9x to be confused during startup. I fixed the conflict and it seems to have resulted in quicker startup. Can you try a build with the fix from the bottom of this page?

schellingb avatar Nov 12 '25 18:11 schellingb

Ok, I tested the build with the fix and I can confirm that not only am I no longer asked to update drivers at the first startup, but the startup sequence itself has also returned to being as fast as before.

Ric-82 avatar Nov 12 '25 18:11 Ric-82