USB 2.0 NEC chipset at 2.0 speed

[Texas_RangerAT]

Just to introduce myself briefly, I'd consider myself an advanced user. I started playing around with Macs when I was around ten, mainly 6.0.x and 7.x system versions at that time. Soon after I progressed to messing around with hardware on a limited scale (e.g. replacing hard drives etc.) and that's roughly where I stand now - knowing little about componwnt level electronics and the inner workings of the OS and programming but fairly confident messing with extensions etc., getting a classic OS to work just as I want.

I occasionally use a beige G3 266 desktop running 9.2.2 with a bunch of software, partly dating back to the late 80s (mainly games). I do notice that if it crashes it's gone - Cmd-Option-Esc won't do anything. I've mainly had this happen with old games and with Claris Works 2 (originally designed for 6.0.7 and 7.0 I think). So partly my ancient software might be the issue but I'd still love an improved OS that lets me kill crashed apps without a forced reboot that destroys any unsaved data in other apps. I've been using X as my main OS since 2005 and haven't had a fatal crash since 2009 I think. All the ones I had before were related to a buggy WiFi driver (d-Link USB dongle), unplugging the dongle occasionally caused a kernel panic. Not like I quit using ancient apps under X either - while I still had an iBook G4 I frequently ran Classic with the same 80s/90s software I have on the G3 and up until 10.6 I used PPC software like Photoshop CS.

Regarding software patches for large drive support I tend to agree with what's been said, I'd avoid it at all cost. Long ago when I was running 7.6 on a Performa 450 with 120 MB hard drive I stumbled across a software called "TimesTwo" that doubled your hard drive space using on-the-fly compression. Needless to say it crashed after a while and I lost all data I hadn't backed up (not that much but sad enough).

Finally there's the issue of USB 2.0. I'm surprised I haven't seen it mentioned anywhere here but according to several users on German forums there is one USB chipset that provides full 2.0 speed under OS 9 using Apple's original drivers and that's the NEC D720101GJ. I haven't been able to verify this, my only card with this chipset only runs at 1.1 speed in a beige G3 running 8.6.1.

[Mat]

Finally there's the issue of USB 2.0. I'm surprised I haven't seen it mentioned anywhere here but according to several users on German forums there is one USB chipset that provides full 2.0 speed under OS 9 using Apple's original drivers and that's the NEC D720101GJ. I haven't been able to verify this, my only card with this chipset only runs at 1.1 speed in a beige G3 running 8.6.1.

Can you post a link to the original informations/discussions? That sounds extremely interresting.

[nanopico]

Finally there's the issue of USB 2.0. I'm surprised I haven't seen it mentioned anywhere here but according to several users on German forums there is one USB chipset that provides full 2.0 speed under OS 9 using Apple's original drivers and that's the NEC D720101GJ. I haven't been able to verify this, my only card with this chipset only runs at 1.1 speed in a beige G3 running 8.6.1.

Can you post a link to the original informations/discussions? That sounds extremely interresting.

I wonder if they confuse the fact that it might show in system profiler that it is a USB 2.0 card, but it doesn't ever initialize at that speed/protocol.  I might be wrong though.

[Philgood]

I think he refers to this article...

http://www.macuser.de/threads/powermacintosh-g3-nimmt-usb-karte-nicht.720072/

We should confirm it. I have 1-2 PCI cards with usb.
At least one card is with a NEC chip.

[Texas_RangerAT]

I think he refers to this article...

http://www.macuser.de/threads/powermacintosh-g3-nimmt-usb-karte-nicht.720072/

We should confirm it. I have 1-2 PCI cards with usb.
At least one card is with a NEC chip.

Exactly, thanks for searching! Imacmini03 states that s/he actually managed to copy large files at full speed so that does seem to have worked. In my case it didn't (not only with the SanDisk drive mentioned over there but neither with an external hard drive I tried later). I still haven't tried the card reader mentioned there (I have a LogiLink too) and I need to test the card under 10.4 to see if it works at full speed there. The current test system is a beige G3 running 8.6.1 installed in 2002 and originally working with a no-name USB 1.1 card.

One caveat: NEC made quite a few different USB chipsets, even some with an identical model number save for the two letters at the end. I might have another of those D720101 cards but I'm not 100% sure (I gave away a lot of my old hardware recently and the card might have gone with one of those Macs). In my experience most USB 2.0 cards won't be recognised by OS 9 at all or will even prevent 9 from booting altogether.

[Mat]

If thats true, we need the exact card model form Imacmini03. Texas Ranger, die you ask for that model, or did you simply try any card with this chip? I am asking, as the NEC D720101GJ seems to be a common chip, and well known for good 9 compatibility as 1.1 chip, from what I read now quickly.
Perhaps the special PCI card Imacmini03 uses had some special hardware, ... ? Or any missproduction that may do the trick? It may be possible, and Imacmini03 sounds real in this thread without joking, but somehow I doubt that it is possible. But on the other hand, we had this Bluetooth driver in 9 that the os9forever guy found a long time after he started to develop such a driver

[Texas_RangerAT]

No, I didn't ask for the specific card model but I'll do that and tell you as soon as I get an answer!

[IIO]

In my experience most USB 2.0 cards won't be recognised by OS 9 at all or will even prevent 9 from booting altogether.

thats my personal experience, too, but i have heard of others who successfully ran an USB 2.0 PCI card in their G4, providing 2.0 speed for OSX - and at least making no trouble when booting into OS9.

[Texas_RangerAT]

In my experience most USB 2.0 cards won't be recognised by OS 9 at all or will even prevent 9 from booting altogether.

thats my personal experience, too, but i have heard of others who successfully ran an USB 2.0 PCI card in their G4, providing 2.0 speed for OSX - and at least making no trouble when booting into OS9.

Yes, some do allow 9 to boot but aren't recognised by the driver at all. That's the case in my dad's Quicksilver. The card works fine in 10.4 but doesn't even show in the 9.2.2 System Profiler. Another card I tested in my beige G3 simply crashed 9 during startup.

Today I purchased a cheap no-name card with the NEC chip on eBay (new in box) just to see what I can get out of it. Unlike the card I have now it's explicitly marketed as a 2.0 card AND claims it's compatible with MacOS 8.6 and up! In the pictures of the box I couldn't find ANY manufacturer info so that'll have to wait until I receive the parcel. That could take a while as the seller only promises shipping within 21 business days.

[GaryN]

I have a 4-port USB2 card in my MDD. I don't know the manufacturer and getting the MDD of of its quiet box is complicated.
The vendor ID# is  0x1106 ; Device ID# 0x3104 (if anybody has the same card.).
Although it works perfectly in OSX, it is NOT recognized at all by OS9. It doesn't affect boot at all.

[Protools5LEGuy]

I have a 4-port USB2 card in my MDD. I don't know the manufacturer and getting the MDD of of its quiet box is complicated.
The vendor ID# is  0x1106 ; Device ID# 0x3104 (if anybody has the same card.).
Although it works perfectly in OSX, it is NOT recognized at all by OS9. It doesn't affect boot at all.

I have 2 USB 2.0 cards on my machines. The NEC one acts as USB 1.1 on Mac OS 9. The VIA one is recognized from Panther onward, ideal for USB OSX only hardware for those OS X days.

NEC chipset cards are scarce and more expensive than VIA chipset ones. It would be great to develop a VIA chipset USB extension for OS 9. 

[geforceg4]

i also have an nec based pci card that works in both osx + os9.. but im fairly certain that it does not operate at usb2.0 speed in os9. VERY certain of that.

[Texas_RangerAT]

I got the card yesterday but won't have time to test it until the weekend. The box contains NO manufacturer/distributor info or model number whatsoever, just says Made in China.

[MacOS Plus]

  I have a 2-port card from Lava with a NEC D720100AGM chipset.  It shows up and works in OS 9.2.2 but is obviously not transfering faster than 1.1-mode speed.  I would think that any brand of card with this chipset would work the same way.

  As much as I'd love to see 2.0 speeds in OS 9.x I'm hard-pressed to believe that it's possible with the existing driver.  It doesn't seem to know anything about negociating the more advanced communication mode with any attached devices.  It's essentially the same thing with the FW800 ports on the 'unsupported' G4s only operating at FW400 speed.

[Mat]

Forget all the cards witht that NEC chip!
You can find a lot of informanions on the net about that chip and Mac OS 9. It always worked with 9.x as 1.1 USB.

If that user at macusers.de wasn´t joking us, and I doubt so as he made detailed informations about transfere speed and more, and even tried tha card in another computer, and also was stunned that it was so fast, you will need exactly his card, to check the speed!
If there is any chance that higher speed is possible it is connected to exactly that card he uses, not to the chip!

[MacOS Plus]

  If the NEC D720101GJ chip is capable of this feat I'd love to see it reproduced by other users.  All of these NEC chips are completely self-contained devices - there is quite literally no other active logic on the PCB they are soldered to (except in the unusual cases of combo cards with additional port types on-board).  In that sense the card IS the chip.  Unless this later NEC chip has a completely independent management processor that doesn't depend on the host operating system and drivers for its ability to negociate USB 2.0 speed, I don't see how it's possible.  I will certainly try to research this further.

  The whole point of USB was like all those horrendous WinModems of the past - the host systems had to provide most of the processing and management of the USB operations so that the USB controller itself is a cheap and relatively simple IC.  Even at higher speeds than USB 1.1, connected devices could never perform as well as those connected via other means because the host system had to expend a lot of processing capacity to manage them.  If you've ever compared the relative performance of a USB-based ethernet connection versus a true dedicated ethernet host controller, or a server-grade fully-managed ethernet controller with its own processor and cache RAM for that matter, you'll know what I mean.  My point isn't that USB is a resource hog, it's just that most consumer-grade USB controller chips are not responsible for the majority of the processing required to negociate and manage external connections.  I've also never heard of a USB chip that was actually capable of doing all that itself, nor did I think it was even possible within the USB design spec.

  Just to clarify about the card I was using - it has two external ports but also an internal pin-header for two more ports.

[Mat]

  In that sense the card IS the chip. 

I totally understand what you mean. But we have the proove that several cards with the NEC D720101GJ work as 1.1 device. We have even many many reports about the chip itselve and Mac OS 9. But we have just one single report about higher speeds witht he chip. So I really think our only option to get the proove (for working or not working) is if we can test exactly that card that the user has.

But of course go on, and test all cards containing the NEC D720101GJ, perhaps we find the right one by accident.

[DieHard]

I have been following this one, but have been a little busy to comment.  I am very convinced that the "USB 2.0" under Mac OS 9 is a Myth, since I personally tried over 3 dozen cards when this was important to me.  Yes there were clearly cards with NEC chip sets that performed well under OS9 as far as stability with the devices they connected, but at 1.1 speeds; VIA always made crap, the cheapest PC Logic boards (VIA chipset), worst audio (VIA), etc.  So it it wise to avoid them.  I also made a note about... "DO NOT use any ALI based cards", but like usual I did not elaborate, so I am unsure what the issue was.

The whole point of USB was like all those horrendous WinModems of the past - the host systems had to provide most of the processing and management of the USB operations so that the USB controller itself is a cheap and relatively simple IC.  Even at higher speeds than USB 1.1, connected devices could never perform as well as those connected via other means because the host system had to expend a lot of processing capacity to manage them.

Wow, that brought back memories and is 1000% spot on, We used to setup Fax servers and other modem intensive applications and "WinModems" were NOT usable.  You needed to use a high quality US Robotics (non-win modem) or intel that had a dedicated controller chip.  The results were awesome, handshaking without issues, and they never dropped the connections.

I also believe, that is why FW audio interfaces run flawlessly compared to USB even on Machines that support USB 2.0; the FW interfaces, when used with a good quality FW chipset appear to use less main CPU, but this is more of an overall observation, I never benchmarked the differences. Maybe, MacOSPlus can elaborate on this.

...or a server-grade fully-managed ethernet controller with its own processor and cache RAM for that matter, you'll know what I mean.

Yes, we would hang about 80 to 90 workstations on the same subnet with a Bus mastering Ethernet card in a Novell Server (Pentium II) and the packet stack never overflowed.  Basically, all network file/print requests were being handled almost in real time with 1 Ethernet card !

[Texas_RangerAT]

I tried the card today and it doesn't work. In fact it crashes 8.6 during startup, even with extensions off. The other card with the same chip works fine but obviously at 1.1 speed. I tried booting three times, twice the system froze during the early stages of the progress bar, before any extension logos showed, the third time I got a bomb (unfortunately I didn't note down the error code but I doubt there's much we could do with it).

[MacOS Plus]

I have been following this one, but have been a little busy to comment.  I am very convinced that the "USB 2.0" under Mac OS 9 is a Myth, since I personally tried over 3 dozen cards when this was important to me.  Yes there were clearly cards with NEC chip sets that performed well under OS9 as far as stability with the devices they connected, but at 1.1 speeds; VIA always made crap, the cheapest PC Logic boards (VIA chipset), worst audio (VIA), etc.  So it it wise to avoid them.  I also made a note about... "DO NOT use any ALI based cards", but like usual I did not elaborate, so I am unsure what the issue was.

The whole point of USB was like all those horrendous WinModems of the past - the host systems had to provide most of the processing and management of the USB operations so that the USB controller itself is a cheap and relatively simple IC.  Even at higher speeds than USB 1.1, connected devices could never perform as well as those connected via other means because the host system had to expend a lot of processing capacity to manage them.

Wow, that brought back memories and is 1000% spot on, We used to setup Fax servers and other modem intensive applications and "WinModems" were NOT usable.  You needed to use a high quality US Robotics (non-win modem) or intel that had a dedicated controller chip.  The results were awesome, handshaking without issues, and they never dropped the connections.

I also believe, that is why FW audio interfaces run flawlessly compared to USB even on Machines that support USB 2.0; the FW interfaces, when used with a good quality FW chipset appear to use less main CPU, but this is more of an overall observation, I never benchmarked the differences. Maybe, MacOSPlus can elaborate on this.

Sorry it took me such a while to respond to you:

  One of the key differences is in the underlying protocols and negotiation schemes.  USB is essentially a 'mutant' serial port with connect/disconnect auto-detection added and packet-based data bursts.  It doesn't have much redundancy available in the timing for devices that need stable data rates, like with audio or with CD burners.  It also doesn't get much priority within the system and often gets communication disturbed when some other process demands attention.  I have a very simple USB audio device from M-Audio that is next to useless for recording because it frequently drops samples without any notification.  Recording a track in sync with another, in other words an overdub, is impossible because every missing sample shortens the length of the content rather than being left in the file blank.  The driver has some part to play in this fault, but the nature of USB is the fundamental flaw.  I also recall blowing A LOT of CDR discs back in the day using a USB-based burner that didn't have a large enough buffer cache to make it through communication disruptions.

  As I understand it, Firewire operates more like the ethernet protocol.  This makes for a much simpler connect negotiation, particularly with external hard drives, and allows more intelligent handling of timing constraints for critical processes that have stability or high-bandwidth requirements.  The data protocol doesn't require as much control data per individual data packet either supposedly so the stream itself is more efficient.  The amount of stable audio and MIDI data that can be carried by a single Firewire connection is quite amazing, and even better when dedicated to the task as in MOTU's proprietary Firewire-based interconnect bus.

  Firewire was designed to handle more of its processing needs by the controller chip rather than the host processor, which is probably the only reason it works for carrying digital video reliably.  Video capture and DV camera control was the primary early target application for the interface, the exact reason I bought my first Firewire PCI card for my PC of the time.  I can't state any kind of percentage difference in host processing required but certainly found it noticeable.  It was far more important that Firewire communications were much harder to disrupt with other system activity than USB.

  One final note with USB vs Firewire as it specifically pertains to Apple.  The whole reason behind the long-standing lack of USB 2.0 on the Macintosh platform was a marketing decision within Apple to push Firewire as "the future" and the best performing of the two.  They had nothing to gain by supporting USB 2.0 when this would run counter to their plans for Firewire.  Remember what a big deal Apple always made about having built-in SCSI (even though it was a crap implementation)?  Well Firewire took its place within Apple in that same sense with a similar fervor.  As I remember it the rest of the industry balked at the licensing fees being demanded for producing/including Firewire in their devices and subsequently left it behind almost completely.  It kinda had the feel of the VHS vs Betamax 'battle' where the product widely deemed to be the lower quality actually ended up winning.  Cost of production and market exposure eventually made USB mainstream and Firewire somewhat 'niche'.  While a Firewire thumb-drive was an awesome rarity, the USB thumb-drives probably outnumbered them by the hundreds of billions.  By the time Firewire 800 appeared on the scene the damage was already done - it was already a lost cause before USB 3.0 came along.  Now every even remotely recent cellphone on the planet has USB.  If history had turned out differently the "Lightning connector" might have been based on Firewire and truly acted like 'lightning'.