True or False? USB2 + Win 98 + Intel 915 = Impossible

[awkduck]

On 3/2/2026 at 4:36 AM, SweetLow said:

tells enough

We gotta take it all, with many grains of salt.

I hope I don't come back, a week from now, and read my writing here. I'd have stop myself from edition all the things I say, that are "kinda" right, "close" to right, and "maybe, I shouldn't post when I haven't slept well".

You guys bare with me pretty well.

Edited March 5 by awkduck

[waltah]

First I want to apologize.  When I said:

"Unless we can interest some driver writer out there it may not have a solution within Win 98"  

I had come to the belief that there is no simple answer to Win 98  taking 5-10 times as long for a USB 2 copy to disk as XP.  That fixing this would require at least days, maybe weeks of work by someone who knows how -- not what most people would call a worthwhile project.  'Interest' is one thing, 'major project' is quite another.  

Again, sorry.  

Since then -- aside from 'life' which makes non-negotiable demands from time to time -- I've pretty well finished up the Win 98 on a Dell 4700 project.  USB 2 (and 1 of course) work fine, at ~1.1 speed and with the restriction that you cannot shutdown or start up with a drive inserted.  I've replaced the ATI Rage 128 (PCI) video card with a Radeon X300 (PCi-e) card, thus freeing that PCI slot.   I installed a Belkin 802.11g WiFi card which worked fine except about once a day it would hang with "has encountered a problem and needs to close" -- except everything was hung but the power button meaning waiting for a SCANDISK on restart.  Replacing that card with a noname (Ralink 2561 chip) fixed that problem.   The Retrozilla 2.3 browser will actually connect and display some websites -- Breitbart and The Gateway Pundit are examples.   

And I installed the Riptide game card in the remaining PCI slot -- joystick, audio, and a dial up port on one card with one driver package.  

All of it seems to work fine.  Then yesterday I pesuaded my wife to ask her paid-for AI about USB 2 under Win 98 SE and yes AWK, "if a you spends some money ( on the better ones )" ...  In two pages I got a complete and seemingly authoritative discussion of the whole issue.   

According to Wife's AI slow speed is built in to USB 2 under Win 98.  Win 98SE had (brand new) USB 2 support and it didn't work at all well.  In fact the problems were so bad -- including data corruption, even loss of the whole drive -- that some companies gave up.  Quoting:

"Motherobard makers got a lot of support calls.

"The Practical Workaround Vendors Used

"Many chipset vendors (notably Intel and VIA) implemented drivers that did one of the following:

"1. Run EHCI in conservative compatibility mode

"2. Disable EHCI entirely

"3 Leave devices attached to the USB 1 companion controller [instead of handling them to the EHCI after enumeration]"

Fascinating stuff.  Basically the claim is that some (many?  most?)  drivers for machines with USB 2 hardware looked at the OS and if it was Win 98 they said "Nope -- we promised you USB 2 but what you get is USB 1.1."   

It's certainly possible because when I plug in an HDD with XP my 150 megabyte copy takes around 6 seconds while using the drive holding Win 98 the same copy takes 45 seconds.  The only difference seems to be the OS.  Maybe the BIOS really is raising a middle finger to my USB 2 drive just because it hangs around in a Win 98SE neighborhood.   

Furthermore it doesn't seem to be just poor coding:  I've tried packages that replaced most modules beginning 'USB' with later ones and differences were slight.   

Of the three 'workarounds' (above) only #1 seems possible because if I disable EHCI in Device Manager USB 2 gets even slower, indicating that EHCI was in use.

Beyond that I don't know ...

[awkduck]

I'd have to say, this still isn't the kind of AI I was referring to.

This is the magnitude of service I mean. What you are getting is a meaningful scrape, of a very sophisticated database.

However, it is very poor at navigating "context". Something can be stated "as near fact" but is actually based on the back and forth conversation between two people, neither really understanding the situation. The further you go back, in to support dialog history, the worse it gets. We were very computer illiterate, en mass. The support personal, for Win9x, were really awful. Support, in those days, was unrealized telemarketing. While still pervasive, today, general awareness has improved.

If I had the time, I'd hunt down, buy a duplicate machine, and trouble shoot this for you; then upload/post an easy to install fix.

The issue is almost certainly your BIOS. This is a known issue, on many machines. It is one of the of the largest plagues, to those wishing to run WinXP off of a USB drive; even Embeded POSReady 2009 has issues with it (installing "too" and booting "from" USB).

The issue "booting up and shutting down" with the drive inserted, nearly declares "without much doubt" that this is your issue.

With all due respect, I hope it doesn't seem as though I am "trolling" you. I think you've expressed "well enough" you interest for developing this situation further. I'm not trying to egg/edge you on. This post is around 75% clarification, for those really wanting to enter masochism, and iron out an issue like this.

As for the other thing.... Try not to feel bad, or be offended. You could almost call it a "cultural" barrier. Not just culture of nation, but culture of creed. Short and to the point, and no sugar coating. If you spend time trying to efficiently communicate to a machine (at the lower levels) you start communicating that way to people. Some of us are that way, before we learn to talk to machines; It just makes us right for the job.

If I get a chance, I'll try to hunt down information on your BIOS interface. There should almost certainly be a way, to disable booting from USB; which would likely clear the issue you are having.

Congratulations, on the other progress you have had.

My first experience "with USB2.0" was on Win98SE, using a NEC PCI card. I had a 40gb IDE Harddrive caddy, and it worked great; from a DELL PII 450mhz machine, up to my PIII 1.4ghz machine. At the time, I transferred files"to and from" that drive faster, than my friends (having P4 1.8Ghz+/ WinXP/intergrated USB2). I also wiped the floor with their machines "at FPS" on Unreal Tournament; this before moving from a PCI to AGP GPU. They had hyper-threading, I only had a little bit of occasional overclocking. That was a great machine, and I'd still use it today. Their main issue "was" XP and integrated devices (no excuse for their AGP cards). Their machines could have been configured much better; but some of those early P4 CPUs where problematic. I blamed it on poor thermal grease and thermal design, but they got much better. Wasn't shocked to see things "more or less" revert to the previous like designs (P6 microarchitecture -> Core architecture). P6 is still cool (no pun intended).

[waltah]

2 hours ago, awkduck said:

IThere should almost certainly be a way, to disable booting from USB; which would likely clear the issue you are having.

There is -- 'Boot from USB' (or some such wording) is an option.  It's currently ON (default, I think) and I've meant to flip it off and see what happens but have not done so.  I'll check it tomorrow -- I mean, later today.  .  

A test I don't think I've mentioned:  I set up to do a USB to disk copy under DOS -- no Windows.  This was by booting from a USB stick -- apparently the only way to get USB under DOS.  Used XCOPY rather than COPY because the DOS COPY doesn't deal with sub directories.  This was 2-3 times faster than using Windows.  Since the comparison isn't exact one can't draw a firm conclusion but the result is consistent with 'Windows 98 is the problem.' 

Though I suppose the real meaning is that the USB code in the BIOS is faster than that in Windows ... not really surprising.  

Of course all AI is a 'scrape of a vast database' coupled to a language generator.  The most accurate results occur when the database covers the entire universe -- there are no possibilities 'elsewhere.'  This is the case for playing chess and finding coding errors, both of which AI does superbly.  On problems involving human intent, open end design matters (did Intel want fast USB 2 on machines running Win 98?) results are going to be messier and the paid-for-it answer that amounts to some manufacturers chose 'No USB 2 speed' as the lesser evil when running under Win 98 is plausible as a scrape of history although halucination cannot be ruled out.

I used Win 95 and 98 when they came out; I had one extended encounter with Microsoft support and was astonished to discover that their level two people knew less about the area (don't remember the details) than I did.  

[awkduck]

4 hours ago, waltah said:

This was 2-3 times faster than using Windows.

Yes, this would make sense, as you would not be using conflicting drivers; BIOS conflicting with Win98 (your current situation). In my previous posts, I mention "running" Win98 from a BIOS driven port (Windows driver disabled, for those ports, and booting using the BIOS USB driver for the primary harddisk support). I was able to reach the theoretical "maximum" speeds.

NOTE:There are USB drivers for DOS. You have to really look for them (Panasonic, and others).

5 hours ago, waltah said:

USB code in the BIOS

Often, yes. I've seen NetBSD USB drivers "near" match, even after the overhead of a running kernel. Otherwise, you will find there tends to be a difference is speed (any OS) though sometimes only a little.

5 hours ago, waltah said:

manufacturers chose 'No USB 2 speed'

It often was "not" a lessor evil. It was to push people forward, with the added excuse of saving developer time and product support. The era was rife with machines carrying the "Made for XP" sticker. There is a larger framework here, that can be explained away with consumer speak lingo. A good chunk of truth involves the murky reality, of things like the Microsoft Halloween documents, and deals made with various vendors, manufacturers, retailers, etc.

This is just the reality, of the "upgrade" mentality. Upgrade mentality "facts" have often been proven wrong, or even scandalous. Though there is potential for a scandal mentality, on the other side of the fence.

5 hours ago, waltah said:

two people knew less about the area (don't remember the details) than I did

Especially, if there was/is a new product on the horizon.

[Guest Drew Hoffman]

AI is fond of saying things are impossible when it lacks proper context and then confabulating details to support this.

I'm not sure this will help with solving your issues but I have to ask - How fast is copying the same 150mb file from one folder to another on your internal hard drive? If that is also capped at 3 MB/second your hard drive may be using PIO mode instead of UDMA. Installing the chipset drivers usually turns DMA on but not always. To check you would look in the properties of your hard disk controller in Device Manager and see if the DMA checkbox is checked. You may also check the Performance tab of System Properties and see if any drives are listed as using MS-DOS Compatibility mode which will greatly reduce performance.  

[waltah]

I've now tried turning off USB booting.  In the A10 version of the BIOS that I installed a week or so back (but not in the A06 that came on the machine) there are three options for the USB Controller:  'ON,' 'OFF,' and (new) 'No Boot.'   In 'No Boot' mode the OS can see USB devices but the BIOS ignores them for boot purposes.   This is a bit confusing because you can also turn off booting from USB in the boot sequence menu.  But evidently even if they are off there, they're visible to the BIOS during boot.  

Anyway if you select 'No Boot' for the USB Controller the machine boots up and shuts down just fine with a USB device plugged in.  Everything else is the same -- still does I/O too slowly -- but eliminating the excessive punishment for an unneeded device being present is good.  Thanks!

Edited March 13 by waltah

[awkduck]

9 hours ago, waltah said:

still does I/O too slowly

So, without giving hard numbers, there was no "noticeable" difference?

If that is the case "without causing you technical gymnastics" the easiest test I can think of, is a dual boot/warm boot.

Boot first into XP, then warm boot (reboot) into Win98. Obviously, I understand that this may be a test too far. This would conclude the "basics" that I can dredge up, right now.

A warm boot would "probably" not reintroduce anything, that XP disabled. I haven't tested this extensively; it could very well reintroduce things.

I'm glad the boot/poweroff issue is better. I could have guessed there might still be a problem. My i915 systems needed more "help" too. As mentioned before, disabling legacy USB (meaning also USB Boot) improved some USB issues. I had to do more, to completely clear the problems.

Oh, be sure to still test booting the machine "without" the drive inserted, for any signs of change. You may not notice anything difference, with just storage IO. But you never know, your i915 isn't the exact beast(s) I have.

Edited March 14 by awkduck

[waltah]

On 3/13/2026 at 3:00 PM, Drew Hoffman said:

How fast is copying the same 150mb file from one folder to another on your internal hard drive? If that is also capped at 3 MB/second your hard drive may be using PIO mode instead of UDMA. 

The thing I hate about taking more data is that often sometimes it proves I've been out in left field. 

Copying that 150MB file under Win98 from USB to the HDD takes ~45 seconds.  Doing the exact same using XP is 6-8 seconds.  But when I copy the same file under Win 98 from disk to another folder on the same HDD it takes 40 seconds.  Copying it to another HDD on the same machine (source disk on port 0, target port 1), 70 seconds.   

It seems that what I thought was slow USB 2 is partly or mostly a slow HDD.  

All of the disks involved are SATA and --- since Win 98 doesn't do SATA -- I've been using the BIOS SATA emulation which is known to be slower than the real thing.  

The 4700 has a PATA port or two.  Probably what I'll do next is try the copies to a PATA HDD.  

[awkduck]

3 hours ago, waltah said:

But when I copy the same file under Win 98 from disk to another folder on the same HDD it takes 40 seconds.  Copying it to another HDD on the same machine (source disk on port 0, target port 1), 70 seconds.

Interesting, that copying to another drive was slower. You'd think reading and writing, on the same drive, would be slower.

3 hours ago, waltah said:

It seems that what I thought was slow USB 2 is partly or mostly a slow HDD.

At least, you'd have an easier solution.

I've never used it, but you might investigate rloew's AHCI driver.

3 hours ago, waltah said:

Probably what I'll do next is try the copies to a PATA HDD.

If you do, please update.

[SweetLow]

8 hours ago, awkduck said:

If you do, please update.

... few days after asking the right, cleared from other factors result of those reading speed 

Edited March 15 by SweetLow

[waltah]

19 hours ago, awkduck said:

23 hours ago, waltah said:

Probably what I'll do next is try the copies to a PATA HDD.

If you do, please update.

Hooboy, have I got an update for you!   I stuck a PATA drive from a working win98 machine on that port on this machine -- looked good, I could see the files there (Windows ...).  The port was stolen from the CD drive I've used to install Win98 -- worked at normal speed for that. 

I did drag and drop for my 150 MByte file from the usual PNY thumb drive to the added drive.  It did copy -- in about 20 minutes.   It looked like the copy went fast in very short bursts every few minutes.  

During the copy the mouse/cursor worked okay but when I tried to open process explorer that took several minutes and it gave no ongoing data -- the numbers were frozen until the copy finished.  

The copy seemed okay -- I could open the files on the target volume and the folder deleted normally.  

19 hours ago, awkduck said:

rloew's AHCI driver.

Is that the 'real' SATA driver for Win 98?  That's a logical next step.  And if someone wants to explain why USB 2 flash drive to disk copy time goes up more than 20-fold when using a real PATA drive, I'm all ears.   

I have a feeling I'm doing something wrong but can't think what it is ... most errors in plugging drives, etc. just lead to 'don't work.'  

[awkduck]

1 hour ago, waltah said:

And if someone wants to explain why USB 2 flash drive to disk copy time goes up more than 20-fold when using a real PATA drive, I'm all ears.

Make sure DMA is enabled, for the drive.

You may need to reinstall/reset drivers and the registry, for this machine; or just use this drive for a fresh install.

Some BIOS have settings for what DMA types are available, to the controllers, among other settings.

1 hour ago, waltah said:

Is that the 'real' SATA driver for Win 98?

Yes, but I haven't used it (or even read the included manual). My first question would be about booting. Some drivers won't do you any good, until you are booted into Windows.

[jumper]

In File System Properties, you can increase the read-ahead buffering on HDDs and toggle Write Caching on removable drives.

I just measured 877000 Bps with System Monitor on a 30MB copy from PNY USB3 32GB in 4-port mini-hub to IBM 6GB PATA HDD over USB1.1(UHCD) on Win98fe. There is also a PNY USB2 64GB, another 16GB, and a mouse in the hub.  I have a USB2 PCMCIA card, but don't use it--I'm happy enough with what I have. The hub acts as a docking port and backups (with write caching enabled) and searches are fast enough.

Another copy from the PNY USB2 64GB had the same peak speed, but fluctuated slower. HDD has DMA enabled and all are Int 13 units: "If this feature is enabled for removable drives, the media must be in the drive when you start your computer." (But the 16GB works just fine when hot-plugged.)

All three thumb drives use Lexar JumpDrive2 drivers, but the 16GB is labeled: "Non-Lexar USB ThumbDrive". I think this one is rloew's mod.

Defragging the HDD can also help. Also test searching for text on the USB drive; this will read without any writes to the HDD (a bit faster than copying here).

 

Edited March 16 by jumper
"searching for text"

[Guest Drew Hoffman]

Are your SATA ports set to "Enhanced / AHCI" or "IDE" mode in the BIOS? Also does the SATA controller share an IRQ with the USB 2 controller in Windows? This is a known issue causing very poor performance and freezes. Here's the message from Win98 QuickInstall with info on how to correct this issue. TLDR: change the SATA controller mode to "IDE Primary". Only 2 of the 4 SATA ports will work though