Just to be sure, I tested a new in retail box WD4000BB, same specs, but still got the same results. So, it doesn't appear that it's the drive, cables, or motherboard slots. Not sure where the bottleneck is...

I suggest testing the drive with Victoria or/and MHDD for slow sectors. Their presence may lower HDD performance significantly.

jst36y wrote on 2023-12-02, 10:55:

Just to be sure, I tested a new in retail box WD4000BB, same specs, but still got the same results. So, it doesn't appear that it's the drive, cables, or motherboard slots. Not sure where the bottleneck is...

The drives are likely fine. Here is an excerpt from a review of a slightly newer (2002) and higher end drive. Consumer dives from that era were not even able to saturate a 66MB/second interface (except when reading/writing their onboard DRAM cache/buffer). UDMA100 and faster was mostly of use in marketing at that point in time.

WinBench 99 pegs the JB’s outer-zone transfer rates at 48.8 MB/sec and the inner zone at 29.2 MB/sec. As a result, it shares, along with the 1200BB, the highest transfer rates yet turned in by an ATA drive to date.

Source: https://www.storagereview.com/review/western- … caviar-wd1200jb

It wasnt until UDMA6 that using that interface became less painful than root canal work.

My PCChips M930 is pretty similar (SiS 645), and here it is with a SATA to IDE converter, a low end SSD and Windows 98. So something like this would likely be the maximum speed.

Just in case, be sure to install the SiS IDE driver if you haven't. https://www.philscomputerlab.com/sis-chipset-drivers.html
In all probability you have, but it never hurts to mention it.

I mean, that doesn't seem that slow to me for the era. That chipset was not a huge performer. But as someone else suggested, the drive might have bad sectors which are slowing things down. I'd do a Spinrite Level 4 on it to be sure.

To rule out Win98 or drivers as the cause run IDECHECK in Dos 7.1, only load himem.sys and not emm386. The latest version is IDECHECK 1.55

You would run IDECHECK /DMA 0

0 is drive 1 aka c:\ drive. 1 would be the next drive etc.

https://github.com/Baron-von-Riedesel/IDEchec … 917e88fe9be12d2

For compairson I get 30468KB/S (31MB/S) on a 440BX which only goes up to UDMA-33. That is on a pentium II 233 in Dos7.1
I get the same maxium speed under Win98se benchmarks using it's default DMA drivers and tick option.

darry wrote on 2023-12-02, 13:39:

The drives are likely fine. Here is an excerpt from a review of a slightly newer (2002) and higher end drive. Consumer dives fro […]
Show full quote

jst36y wrote on 2023-12-02, 10:55:

Just to be sure, I tested a new in retail box WD4000BB, same specs, but still got the same results. So, it doesn't appear that it's the drive, cables, or motherboard slots. Not sure where the bottleneck is...

The drives are likely fine. Here is an excerpt from a review of a slightly newer (2002) and higher end drive. Consumer dives from that era were not even able to saturate a 66MB/second interface (except when reading/writing their onboard DRAM cache/buffer). UDMA100 and faster was mostly of use in marketing at that point in time.

WinBench 99 pegs the JB’s outer-zone transfer rates at 48.8 MB/sec and the inner zone at 29.2 MB/sec. As a result, it shares, along with the 1200BB, the highest transfer rates yet turned in by an ATA drive to date.

Source: https://www.storagereview.com/review/western- … caviar-wd1200jb

It seems you may be right when I look at old benchmarks. I'll test with some other programs to see.

Repo Man11 wrote on 2023-12-02, 21:24:

Just in case, be sure to install the SiS IDE driver if you haven't. https://www.philscomputerlab.com/sis-chipset-drivers.html
In all probability you have, but it never hurts to mention it.

Thanks for the input. Yeah, it's odd, I don't recall installing the IDE drivers, just the SiS AGP. Let me ask you, is it normal for it to show up in device manager as GENERIC IDE DISK TYPE 47?

jst36y wrote on 2023-12-02, 23:12:

Repo Man11 wrote on 2023-12-02, 21:24:

Just in case, be sure to install the SiS IDE driver if you haven't. https://www.philscomputerlab.com/sis-chipset-drivers.html
In all probability you have, but it never hurts to mention it.

Thanks for the input. Yeah, it's odd, I don't recall installing the IDE drivers, just the SiS AGP. Let me ask you, is it normal for it to show up in device manager as GENERIC IDE DISK TYPE 47?

I seem to recall at least one of my Win98 computers with that hard drive designation, but not in this case.

Yeah, I think it's just the speed it should be based on some benchmarks like this one here: http://ixbtlabs.com/articles/hddreview0108/index.html. I tested in another program, hdtach, and it seems to get around the same average sequential read of 37 MB/s. I did install the SiS IDE driver, but from what I could tell it doesn't do anything on Windows 98 except change the hard drive description in Device Manager (and perhaps enable DMA if it wasn't) as a diff of the previous and new Windows folder showed no changed files, just 2 new files for some hdinfo utility. I have a little external IDE to USB 3.0 device that I hooked up to my new machine and even there I was only able to get around the low 40s. I guess I'll consider this groovy mystery solved. I thank everyone for chiming in.

jst36y wrote on 2023-12-03, 04:10:

Yeah, I think it's just the speed it should be based on some benchmarks like this one here: http://ixbtlabs.com/articles/hddreview0108/index.html. I tested in another program, hdtach, and it seems to get around the same average sequential read of 37 MB/s. I did install the SiS IDE driver, but from what I could tell it doesn't do anything on Windows 98 except change the hard drive description in Device Manager (and perhaps enable DMA if it wasn't) as a diff of the previous and new Windows folder showed no changed files, just 2 new files for some hdinfo utility. I have a little external IDE to USB 3.0 device that I hooked up to my new machine and even there I was only able to get around the low 40s. I guess I'll consider this groovy mystery solved. I thank everyone for chiming in.

Spinning rust was really slow at the time (and arguably still is). Glad this got answered/resolved satisfaction.