• Updated 2023-07-12: Hello, Guest! Welcome back, and be sure to check out this follow-up post about our outage a week or so ago.

68K/early PPC with onboard AAUI 100mbps ethernet?

omidimo

Well-known member
The list vs. street computer pricing is interesting, but how did 10Mb and 10/100Mb NIC prices compare across platforms I wonder?

What was the first Mac with built in 10/100Mb networking? Gigabit was waaaay late.


The first Mac with built in Fast ethernet was the bondi iMac followed by the B&W G3. The Beige G3s had a CTO choice. 

The 10/100 NuBus cards showed up around 1995 and as all things NuBus suffered from high pricing.  :/

 
Last edited by a moderator:

MOS8_030

Well-known member
I'm sorry, but this thread is fascinating! Reading all this is causing me to dredge up a lot Mac memories from the '90's when I was heavily into Macs.

I was reminded that I saved several hundred dollars by buying my new 840av from a grey market seller in Dallas in early '94.

I think I paid about $2800. I'm still using the keyboard I bought with it.

I bought a lot of stuff from Shreve and Small Dog and other more sketchy resellers back in the day.

shreve.jpg

 

Gorgonops

Moderator
Staff member
What was the first Mac with built in 10/100Mb networking? Gigabit was way late.
Huh? How was it "way late?". Apple introduced GigE in the July 2000 revision of the Power Mac G4, and I'm pretty sure it was one of the first "consumer/prosumer grade" machines on the market to include it. GigE wasn't really mainstream until at least three or so years later. (Trust me, I've been in charge of buying a *lot* of networking gear over the years.)

 

Gorgonops

Moderator
Staff member
But what I said was that all Macs were slot starved from day one as compared to PCs and that market penetration suffered badly for it.
What is your support for this other than the fact that you like slots, MOAR the better, and you're projecting that prejudice backwards into the heads of the majority of the computer buying public at large? I've ripped into a lot of computers over the years, and frankly the most common thing I find in expansion slots are cobwebs. The number of slots in a machine was, outside of people that had very specific needs in mind, *never* that much of a consideration, and it's also extremely dirty pool the way you keep handwaving the fact that a lot of the ***t that had to go in an expansion slot on a white box PC was already soldered to the motherboard of even the most basic Macs. An 8 slot PC isn't 8 slots, it's six at best, and realistically it's more like four once you add the connectivity that a base Mac comes with. (Three slots if the Mac has Ethernet onboard... heck, you can actually sort of give that to the Mac even without ethernet because all Macs had Localtalk, which is a perfectly serviceable network technology for small/simple deployments and a heck of a lot better than what a PC came with, which was *ZIPPO*.) You're also utterly ignoring the fact that quite a lot of the more premium-brand business PCs that more closely competed with Apple's products in terms of style *also* had fewer slots. For instance, Dell sold a line of cute little 486 machines in their "486P" line that only had three miserable ISA slots (which meant your options were *really bad* if you ever wanted to upgrade the video card), and they must have sold a bazillion of those things because they were dirt common on secretary's desks all through the 1990's. Again, if you're keeping score: add a Soundblaster card and a network card to one of those to match a Quadra 700 you're down to *one* ISA vs. two Nubus. Pretty sure the Mac wins that one!

In short, without evidence besides "SEE, FEWER SLOTS! MAC BAD!" I don't think you've made your case.

(Again, seriously, I think the bigger problem Apple had was in 1992 is that cute little Dell 486P had a retail price of around $3,000 for a config roughly matching the $4848 gray-market 8/105 Q700. That's something a prospective customer if far more likely to notice than any disparity in the number of slots.)
 

Point of fact, the lowest end PC had the option of Sound blaster Pro and 8-bit Stereo out (with a side of game port) from 1990 on. Had anything like it been available (PAS16 price points don't count) to a Mac user it would have filled the single slot allotment of any LC consumer level Mac until 1993.
Seriously? The ASC sound chip used in the Mac II family machines is essentially feature-comparable to a Soundblaster Pro (22mhz PCM with stereo out), and the Quadras up that to CD-quality. The *one* thing they're missing compared to a soundblaster card is the OPL synth chip for plinky chiptune goodness (Apple couldn't include a hardware synth thanks to their ridiculous legal settlement with the Beatles' record label), but you *could* easily attach a MIDI device like a Roland MPU box using the built-in serial ports. Basically *nobody* cared about the difference in specs between the two options, all they knew is that either was *infinitely* better than what you got out of a basic PC speaker... which is why, sure, back in the 386/486 era the *one* most common user-added card you're likely to find in a PC is a sound card. (And you can bet your sweet patootie that the person who installed it *absolutely hated* doing it. ISA sound cards are the *worst*, especially 16 bit ones like the PAS16.) Did anyone other than people *specifically* trying to do production work with sound *ever* buy a sound card for a Mac? Sure, I know they existed, but did any games take advantage of them in any form? Citation needed; without such I'm inclined to say that most people found the Mac's built-in hardware "good enough".

(Probably the second most common thing a PC owner of that era cracked the box for is to install a CD-ROM drive. And, again, every Mac has a port on the back for that. Even works on a Mac Plus.)

 
Last edited by a moderator:

Trash80toHP_Mini

NIGHT STALKER
In short, without evidence besides "SEE, FEWER SLOTS! MAC BAD!" I don't think you've made your case.

(Again, seriously, I think the bigger problem Apple had was in 1992 is that cute little Dell 486P had a retail price of around $3,000 for a config roughly matching the $4848 gray-market 8/105 Q700. That's something a prospective customer if far more likely to notice than any disparity in the number of slots.)
Fair enough, though I've been talking about perceived value. This isn't the place for it, but I've held onto the pages from the most even handed review I saw from 1984 comparing the 128K to the PC. It details the Mac's inclusion of the several slots needed to match what was included in the Steve's vision of computer as standalone appliance (more on that below) including out of the box networking capability.

Seriously? The ASC sound chip used in the Mac II family machines is essentially feature-comparable to a Soundblaster Pro (22mhz PCM with stereo out), and the Quadras up that to CD-quality.
Yes, seriously. Granted, ASC was capable and even implemented with stereo output on the jack in the back on higher end models. The problem was that Apple tied the two channels together on any Mac that had a chance to compete with the $3,000 Dell you mentioned. Limiting the LC, LCII, LCIII and CC to mono output was a big mistake IMO as that was the period in which the PC became the platform of choice for gaming. That was a bmajor factor after the release of Windows 3.0 which included Sound Blaster drivers.

The *one* thing they're missing compared to a soundblaster card is the OPL synth chip for plinky chiptune goodness (Apple couldn't include a hardware synth thanks to their ridiculous legal settlement with the Beatles' record label), but you *could* easily attach a MIDI device like a Roland MPU box using the built-in serial ports.
OK, that absurd lawsuit over right of use for a name from a different industy could explain a bit of the Mac didn''t have stereo output for so long. But Apple tied the ASC's channels together for the Duos right through the 2300c, did they do the same for the rest of the PowerBook line? Carrying headphones with a PowerBook for a little gaming in stereo on the road would have been a nice feature in the wonderful, dockable ultraportables, but even the dock was mono at the jack back on the ranch.

(Probably the second most common thing a PC owner of that era cracked the box for is to install a CD-ROM drive. And, again, every Mac has a port on the back for that. Even works on a Mac Plus.)
Here's where the PC became more akin to the SJ's vision of computer as appliance. The Tandy 1000SX semi-compatible I bought when the Mac was the cute little hobbit lookin' thing needed the serial port card I bought with it. I kicked myself up and down the block for not buying generic because the funky printer port implementation killed off any chance of running dongled CAD/CAM apps. The dang thing just wouldn't die, it became the heart of the telecom setup because its limited number of slots and second FDD bay made it an all in one box appliance solution for internal card based HDD upgrade and the FAX modem (early 1988 when such external boxes were yet to appear for the Mac) that turned customer's FAX machines into remote scanners and my paperless office setup.

The one external device I hooked up (before the faster Migent pocket modem arrived with the Compuserve Business Pack) was my main machine, the Mac SE/20/Radius16 graphics CAD/CAM workstation! The multitude of external boxen hooked up to the little hobbit lookin' thing looked nothing like SJ's appliance. That 1000SX was its backup hard drive over the SCSI connected QuickShare SCSI card in the PC that became its scanner replacement for the ThunderScan I'd been messing around with up to that point.

Whatever, I'll post the review that almost got me to buy a Mac before the FPU on the SE's accelerator made the platform ready for prime time CAD/CAM duty, far in advance of anything available the PC in 1987. The Logitech mouse for the Tandy came in the undongled Generic CAD box I was later able to buy for some functions unavailable turnkey MacSignmaker system that brought about love affair with the Mac that kept me one of the faithful through the dark days when Apple was doomed. [;)]

p.s. thanks for your correction on my GigaBit misconception, my bad.

p.p.s. Sound cards: archives

 
Last edited by a moderator:

NJRoadfan

Well-known member
Trash,

Apple wired the IIgs audio output to mono even though the Ensoniq 5503DOC (the chip that triggered the Apple Records lawsuit) could do multi-channel sound. Thankfully an expansion connector was placed on the board and 3rd party sound cards added stereo and audio input (the DOC could do mono audio sampling, some sound cards had their own stereo digitizer).

The SoundBlaster Pro's stereo output is technically a  44.1khz mono stream that goes thru a filter to alternate samples between the left and right channels.

 
Last edited by a moderator:

IPalindromeI

Well-known member
The Mac was never going to be a gaming platform due to Apple's apathy towards it post-II and the Mac's poor suitability towards gaming. Big framebuffers + no acceleration features like sprites + slow CPU + slow I/O between those = bad gaming platform. For the audio you got, mono was a silly limitation, but good enough and enough to compete.

 

Gorgonops

Moderator
Staff member
The problem was that Apple tied the two channels together on any Mac that had a chance to compete with the $3,000 Dell you mentioned. Limiting the LC, LCII, LCIII and CC to mono output was a big mistake IMO as that was the period in which the PC became the platform of choice for gaming. That was a bmajor factor after the release of Windows 3.0 which included Sound Blaster drivers.
The original Sound Blaster was also a mono device. The SoundBlaster Pro, circa mid 1991, rectified that, sort of (as NJRoadfan points out) but low-end mono 8-bit Soundblasters (and compatibles) were present on the market for a *long* time even after the Soundblaster 16 was introduced in 1992. So, again, if you're comparing "Apples" to Apples, IE, low-end machines, you're once again pretty much breaking even feature-wise when you compare an LC-series Mac to a low-end "cheap and cheerful" gaming PC.
 

I've been talking about perceived value.
Let's think yet a little more about "value".  There's no way an LC could be considered a competitor to that $3000 Dell; I picked that machine as an example because at the time it was a premium, stylish, small-footprint "Executive Workstation", IE, I tried to find something that you'd actually reasonably cross-shop against a Quadra 700. A Dell 486P/33 would *easily* run rings around an LC II, it's not even close. (16mhz 68030 *on a 16 bit bus* vs. 33mhz 486.) Let's figure out what a potential LC buyer in January 1992 would actually be cross-shopping against. (Picking that because I have the handy list of "street prices" from omidimo above.) I'll choose the base-most 2/40mb config at $1595 and add an Apple 13" monitor at $665, for a total entry price of $2260, and to whatever PC I pick out I'll add the price of a soundblaster card (Found an ad selling it $169, so that's our reference there.). I'll break this into two categories: PCs you could buy for the same *price*, and a PC that's actually roughly comparable in terms of features, and I'll also break it down into "brand name" verses "randomly chosen from one of those sketchy ads in the back".

In the "sketchy white box" category, from an outfit called "Microline Computers" I can get in the "feature match" category:

386sx/25 with 2MB RAM (expandable to 8mb), 52MB HD, 512k VGA card, Relisys 14" 1024x768 (interlaced) SVGA monitor, keyboard, mouse, MS-DOS+Windows 3.0+soundblaster: $1,434.

Best price match:
386DX/33 w/64k cache SRAM and 4MB RAM (expandable to 32mb), 120MB HD, 1MB VGA card, non-interlaced 14" 1024x768 SVGA, DOS+Windows 3.0 + soundblaster: $1964.

(Price match was a tough call, because for the $2260 target you get to choose between a really tricked out 386 or a stripped 486. I went with the 386 figuring you could stand to spend the money on more RAM. It also sort of reflects a more realistic cost of the LC if you bought a non-Apple monitor.)

For the "name brand" I went with Gateway 2000 because their ad actually had convenient prices printed on it, Dell's was a hot "call us" mess. From them, the feature match:

386sx/16, 4MB RAM, 512k VRAM, 40MB drive, 14" monitor, etc, etc, Soundblaster: $1614

And the price match:
 

386DX/33, 4MB, 1MB VRAM, 120MB HD, 14" monitor, etc, Soundblaster: $2314

(Yeah, this time I erred on the high side by $50. I picked it because it's the same config as the completely no-name machine; the roughly $400 premium for coming from a prettier ad also roughly reflects what you'd pay if you bought a "white box" from a local computer store instead of mail order. Of course, if you went to a local computer store for your Mac you'd probably be paying more than that gray-market ad's prices too.)

I think the thing that becomes readily apparently when you make a comparison like this is that not only is the price of entry significantly lower for a PC, the huge price gap that existed between Apple's bottom-of-the-barrel systems and their "mainstream" business Macs simply didn't exist in the PC world. To make the jump from an LC to a IIci (a machine that compares pretty evenly to the 386 I chose in the "price match" configs) nearly doubles the price of the machine, while in the PC world it's only about a 50% premium for a machine that'll easily be twice as fast. (This price-performance curve actually gets really bad if we chose the IIci as our comparison bait, because its entry price is deep in brand-name 486DX territory. You can get an EISA machine with a 340MB SCSI hard drive for about the price of the 8/240 IIci. Or, more reasonably climbing up the ladder from the 386 config, you can make that a name brand 486/33 with a 200mb IDE drive and undercut the 8/105 IIci by about $400.)

Apple's problem is simply that they cost too darn much for most people. I seriously doubt more slots would have tilted the scales much.

 
Last edited by a moderator:

trag

Well-known member
 this suggests it would actually be possible to build a RAM card to give yourself one GB of RAM in a Quadra 900/950. That'd be amusing, if utterly useless from a practical standpoint.)


I would be an interesting exercise for a number of reasons.   

FPM RAM is slowwwwww...., even compared to the old 20 - 40 MHz bus speeds.     I have not examined the 680x0 bus protocols in sufficient detail to be sure, but it looks like when the CPU does a read or write to memory it waits for a bus acknowledgement signal rather than just having some number of wait cycles programmed in.

So, if one were to build a very fast memory subsystem, based on a cheap DDR2 DIMM and one of the cheap Xilinx FPGA that now have built-in DDR2 controllers, and added 680x0 and Q950 compliant GLU so that when the CPU does a read or write to that RAM card, it gets back an ACK on the very next cycle, that might be much faster than built-in memory.

The trick would be figuring out some way to tell the OS that there's RAM in those addresses.   I'm not quite sure how the RAM detection routine at boot up would handle that, but I bet bbraun knows...

 

Trash80toHP_Mini

NIGHT STALKER
Connectix Virtual or RAMdisk+ might work, depending on the OS? Could the same kind of thing be done in the IIfx PDS as a 20MHz access SiliconDisk? Read somewhere that the IIfx PDS isn't half clocked for EVERY application. Don't recall where that one popped up.

@NJRoadfan: That Sound Blaster page says the Sound Blaster 16 came out in 1992, with "good enough" for CD playback quality. HiFi def was still two years off. I haven't found ANY info yet on my PAS 16s for NuBus and PDS. Wikipedia article author was probably a MacHater from back in the Platform War days.

Those would be the days when number of available slots was guns and ammunition. The warriors on both sides only managed to prove their lack of understanding of computing in general, eudi. Battle Ground skirmishes were fun to watch even after Y2K over on ars. [;)]

 

Gorgonops

Moderator
Staff member
that might be much faster than built-in memory
That's certainly a good point. Unless there's something in the PDS arbitration circuitry that would prevent it (and it doesn't look like there's a whole lot to that) you could probably supply a gig of ram RAM effectively as fast as the SRAM used in 1992-era CPU caches.

It would be interesting to know what you'd have to do from a software standpoint to get a RAM expansion card recognized. (Did Apple's firmware actually anticipate the possibility?) From a hardware standpoint, I vaguely recall that the later Mac II and Quadra machines omitted any hardware on their motherboards to remap memory in the SIMM sockets into contiguous blocks; IE, the socket banks were scattered across the available address space, and if a bank was occupied with SIMM(s) smaller than the per-bank spacing you ended up with "shadowed" copies of RAM over the bank area. (IE, if the banks were spaced, I dunno, 64MB apart, and a bank had 16MB of RAM in it you'd see 4 copies of that memory before you got to the next block.) Instead of fully decoding and serializing the banks in hardware they program the MMU to logically remap all the available chunks of memory into a contiguous block starting at $00000000. So... if that is indeed the case I suppose you'd have to design your RAM card so it doesn't stomp on any of the areas where the SIMM sockets actually reside. Hopefully in the case of something like a Quadra 950 the "dirty" area would be limited to, say, the lower 256MB, instead of smeared across the whole 1GB of area Apple calls "RAM" in the memory maps. (I imagine it's probably documented somewhere in the developer notes, etc, where the RAM sockets physically map in a given machine. The fact that the card designer's document says you *can* map to that area at least provides a glimmer of hope that it *is* actually possible to put something in there without conflicts.)

I guess super-ideally if you supplied the Mac with some near-zero-wait-state RAM like that you'd reprogram the MMU so the low-memory globals were stored in it instead of just piling it up behind the slow RAM. Maybe you could even unmap it entirely? (Another question actually then arises if the parts of a Quadra that do psuedo-DMA, like the Ethernet controller, would be able to deal with RAM mapped on the PDS.)

 

Gorgonops

Moderator
Staff member
That Sound Blaster page says the Sound Blaster 16 came out in 1992, with "good enough" for CD playback quality. HiFi def was still two years off.
Any mention of "Hi Def" audio gets me in the mood to snatch tinfoil hat wearers baldheaded. I can *kind* of grant that capturing at higher sample rates/bit depths might have some value in the production studio, but unless you're a vampire bat CD-quality (44.1khz@16 bit) is all the consumer will ever need. Although you do have me intrigued: who was doing "HiFi Def" audio cards in 1994? Audio cards started supporting 48k around the mid-90's because it was used for DAT and DVD soundtracks, but I don't recall seeing "Hi Def" 96k/24bit in "mainstream" consumer cards before the early 2000's. (Soundblaster Audigy 2? Of course, by this point most people had given up on discrete sound cards and were settling for AC'97 DACs built into their motherboards.)

I haven't found ANY info yet on my PAS 16s for NuBus and PDS. Wikipedia article author was probably a MacHater from back in the Platform War days.
I can tell you from owning one (a PAS 16 was my first sound card) that a PC PAS 16 was feature-wise basically a match for the Soundblaster 16. (IE, up-to-CD-quality 16bit stereo DAC plus a Yamaha OPL-3 synth chip.) Looking in the Mac PAS 16 manual it appears it's essentially identical, including exactly the same synth chip. So think of it as a Mac version of a Soundblaster 16, just without the game software support. It apparently came with a cute little mixing-board thing that the PC version didn't come with so it looks like it was trying to stake out some low-rent "prosumer audio production" niche. (Which is sort of at odds with its inclusion of a game port.) Kind of reads like a solution in search of a problem.

 
Last edited by a moderator:

omidimo

Well-known member
All this SoundBlaster talk reminds me of this travesty:

SoundBlaster-Live-Mac-box.jpg.72a81e501aa2002c869e99daa7951fc6.jpg


I rushed to get one only to discovery how crappy the drivers were. I promptly returned it without hesitation. I don't think it was on the market too long, It vanished from the catalogs as soon as it landed.

 

CC_333

Well-known member
It probably would've been fun to keep, as a collectible, if nothing else. I'm sure they're quite rare nowadays due to a combination of short time on the market and nobody buying.

c

 

NJRoadfan

Well-known member
I never realized the PAS-16 Mac had the Yamaha OPL3 on it. Did any software actually take advantage of it? Did MediaVision even bother to write a driver for it to make it appear as a General MIDI device to DAW software?

 

Cory5412

Daring Pioneer of the Future
Staff member
So, kind of going back to the worthwhileness of 100 megabit. ryaxnb21 in #68kMLA on irc.oshaberi.ne.jp (the official channel!) found this tidbit in a MacUser magazine:

IvxqGE2.png

The WGS80 is, as you know, a Quadra 800 and FDDI (if you don't know) is 100 megabit token-ring-like networking standard. I have no idea what the overall infrastructure of the test network.

Using normal AppleTalk, the results are disappointing. I don't know what the special file transfer utility does differently except perhaps, IDK, uses TCP/IP Instead of AppleTalk or uses the native FDDI protocol.

I'd still love to see tests with that ethernet card.

EDIT: It was July 1995, PDF is here. On the PDF, it's page 52. (these PDFs were taken from some kind of sampler or reference CD rom where the ads were removed.)

I have yet to totally read the article, It sounds like there's a few different things going on and they were testing for a few different things. The focus of the article is a little less about improving one mac's experience by switching a desktop to 10/100, and a little more about improving a 40-desk workgroup's experience by changing network technology or swapping out a server.

 
Last edited by a moderator:

Gorgonops

Moderator
Staff member
Using normal AppleTalk, the results are disappointing. I don't know what the special file transfer utility does differently except perhaps, IDK, uses TCP/IP Instead of AppleTalk or uses the native FDDI protocol.


44 seconds for a 30MB file is still pretty miserable. Regular old 10base-T should be able to beat that. (Should handle a bit over 1MB/s on a switched network; collisions will cut that down if you're using hubs, of course.)

I suspect the real villain in this story is the overall crummy design of the classic MacOS. As evidence I present this snipped from the AWS95 Server Tuning Guide:
 

Increased performance you can expect from AppleShare Pro

The following table shows the results of tests with AppleShare running on a Macintosh Quadra 950, and with AppleShare Pro running on the Apple Workgroup Server 95. As an example of the tests, the read operation involved opening a set of large files and sequentially reading from each one. The numbers in the table reflect the aggregate throughput of the server, which is the sum of all the server's processing for all of its clients.


Table 1-1
Comparison of performance of AppleShare 3.0 and AppleShare Pro (with 10 active clients)


 


AppleShare 3.0.1
on Macintosh Quadra 950


AppleShare Pro 1.0
on AWS 95


AppleShare Pro 1.1
on AWS 95


Sequential read operations


193 Kbits/sec.


851 Kbits/sec.


951 Kbits/sec.


Sequential write operations


160 Kbits/sec.


348 Kbits/sec.


594 Kbits/sec.


Enumeration (file listing) operations


90 items/sec.


132 items/sec.


295 items/sec.



The aggregate throughput of read and write operations performed by AppleShare Pro is four times faster than that of corresponding operations performed by AppleShare on a Macintosh Quadra 950. The improvement in aggregate throughput in AppleShare Pro as compared to AppleShare is largely due to the multitasking of the server; while the server is waiting for a request from one client, it processes a request from another client. (Tests of directory enumeration operations are discussed later in this chapter.)
Same hardware can blast almost four times the data through the Ethernet port under Unix than it can under regular System 7.

 

Unknown_K

Well-known member
I am sure the aws95 they tested has the PDS card with cache built in so that helped the processor quite a bit.

 

Cory5412

Daring Pioneer of the Future
Staff member
Not having read the article yet, one more anecdotal thing is that there was another server article in... gosh, I think it was September 1999 MacWorld. It compared ASIP, IRIX, Solaris, NT4, and a beta of OS X for workgroup file service. Perhaps predictably, ASIP lost. I'll have to reread that because I think that issue is on Archive.org.

 

Gorgonops

Moderator
Staff member
PDS card with cache built in so that helped the processor quite a bit.
CPU cache? I kind of doubt 256k of L2 cache is going to make that much difference in I/O bound processes. The faster SCSI ports on said card might certainly help overall but again, unless Apple was using *really* lousy hard drives in the Quadra 950 doing 1MB/second shouldn't be out of reach of the built-in SCSI chip.

 
Top