Jump to content

68-pin VRAM SIMM “shortage”?


Recommended Posts

  • Replies 70
  • Created
  • Last Reply

Top Posters In This Topic

On 2/27/2021 at 5:13 AM, jessenator said:

I read somewhere this was a limitation within the ROM, but you'd think as the 90s went on higher on-board color density at "desktop publishing" resolutions might've been desired, especially on the flagship or higher end machines that could handle it... But maybe there's another technical limit there I'm not considering.

Perhaps there was less demand for higher capacity VRAM on the system board as graphics cards became the 'standard' way of providing video, especially in the higher-end machines.  When I bought my Quadra 605 (new), I upgraded the VRAM, fast forward a few years and I bought a 7600 with a PC card and a video card. (for quite a bit more money) I also got one of the 1710 AV monitors - nice monitor when it worked ;-)

Edited by Juror22
Link to post
Share on other sites
19 hours ago, ArmorAlley said:

I'd say a lot higher. 512K VRAM SIMMS have become rare of late and in demand.

 

That's certainly possible.  I just threw out a rough estimate.  Are 80ns modules fast enough for all Macs?

Link to post
Share on other sites
  • 68kMLA Supporter
On 2/27/2021 at 6:13 AM, jessenator said:

According to the Apple Memory Guide, very few of the Quadra/Centris line utilize 512k modules, but a IIvx/vi can?? 605/475 also do. I read somewhere this was a limitation within the ROM, but you'd think as the 90s went on higher on-board color density at "desktop publishing" resolutions might've been desired, especially on the flagship or higher end machines that could handle it... But maybe there's another technical limit there I'm not considering.

 

The Wombat boards (650/800) can't use them, even an 840av can't use them... And that was primo, tip top definition of Flagship Macintosh.

 

The utilization... It's just baffling to mme but again, if there's a limit, like there wasn't a point to having 4 MB on-board, I'd like to know.

 

I mean Radius and RasterOps were definitely in that ballpark and beyond with their top model NuBus cards.

 

I'd hazard a guess the most popular use scenarios for these now are 575s/Mystic Color Classics (2x512KB to get Thousands of Colors at 640x480, assuming that mod has been performed) or the HPV cards from the x100 Power Macs (adding 4x512KB to a 2MB card to turn it into a 4MB card).

 

I haven't been able to find the 112 pin VRAM DIMMs for the later Power Macs for some time either, but I figure those aren't in as high demand because you can just toss a cheap PCI Radeon in those and get way better video than what was onboard.

Link to post
Share on other sites
On 2/28/2021 at 4:23 AM, Juror22 said:

Perhaps there was less demand for higher capacity VRAM on the system board as graphics cards became the 'standard' way of providing video, especially in the higher-end machines.  When I bought my Quadra 605 (new), I upgraded the VRAM, fast forward a few years and I bought a 7600 with a PC card and a video card. (for quite a bit more money) I also got one of the 1710 AV monitors - nice monitor when it worked ;-)

 

I think this is accurate.  But also, the onboard system VRAM was supplied for general purpose.  For specific needs, there were machines with NUBUS and PDS slots, where a video card could easily be put in.  If Apple's target market for a Quadra 650 was varied (some for server use, some for graphics workstations, some for home computers, etc.), then why increase the cost of the machine by requiring the buyer to have high-end graphics?  Supply what's needed for a general use case, and if the user needs more power, they can buy a third-party graphics card that has more of what they need.  That's what the logic was, I think.  The same went with soldiering on system RAM.  People's needs were varied, and by soldering on the basic amount required to use the computer, users could choose what amount they wanted and add it.

Link to post
Share on other sites
On 2/28/2021 at 4:19 PM, MrFahrenheit said:

The same went with soldiering on system RAM.  People's needs were varied, and by soldering on the basic amount required to use the computer, users could choose what amount they wanted and add it.

 

Up to a point yes, but I've always looked at soldered RAM as a way of hobbling Macs that should have been more capable, but would have competed with higher end machines. IIsi would be a case in point, by soldering 4MB in Bank A, Apple precluded a 128MB Loadout and competition with the IIci. Quadra 605 same deal, the soldered "Bank" of baseline RAM killed off expansion as was also the case with the follow-on Quadra 630 and its derivatives. Even the two slot 630 DOS variants were hobbled by soldered minimum capacity Banks of RAM on the logic board.

 

That said, VRAM is another thing, but limitations there also push dollars toward higher end machines. There were few (one?) 24bit VidCards for the LC/LCIII slot. If you wanted more pixels at full depth, it was up the product lineup or SOL.

 

I finally found a quartet of 512MB VRAM SIMMS for the 2/4 HPV slated for my Radius 81/110. But that was several years back and more than a few bucks over what I would have spent on any other machine.

Edited by Trash80toHP_Mini
apparently I can no longer type/spell.
Link to post
Share on other sites
On 3/8/2021 at 3:29 AM, Trash80toHP_Mini said:

 

Up to a point yes, but I've always looked at soldered RAM as a way of hobbling Macs that should have been more capable, but would have competed with higher end machines. IIsi would be a case in point, by soldering 4MB in Bank A, Apple precluded a 128MB Loadout and competition with the IIci. Quadra 605 same deal, the soldered "Bank" of baseline RAM killed off expansion as was also the case with the follow-on Quadra 630 and its derivatives. Even the two slot 630 DOS variants were hobbled by soldered minimum capacity Banks of RAM on the logic board.

 

That said, VRAM is another thing, but limitations there also push dollars toward higher end machines. There were few (one?) 24bit VidCards for the LC/LCIII slot. If you wanted more pixels at full depth, it was up the product lineup or SOL.

 

I finally found a quartet of 512MB VRAM SIMMS for the 2/4 HPV slated for my Radius 81/110. But that was several years back and more than a few bucks over what I would have spent on any other machine.


I don’t really think soldering the ram to limit the maximum memory configuration was the motive. Because even the Quadra 800 had ram soldered onto the board (and vram for that matter). I’m sure it was cheaper to solder on a basic amount of ram than it was to offer simm sockets everywhere. 
 

The IIsi had 1mb soldered to the board, not 4. This was mainly because of the vampire video, requiring system ram for video. It was actually a good idea considering the way the video was. What I never could understand is how the LC could have dedicated and upgradable video ram but the IIsi used system RAM and was stuck at its amount. Back in 1993 I had a IIsi and I frequently ran it in black and white for extra speed. 
 

When each machine model was released, the base amount of ram allowed basic use of that machine. The Plus and SE with 1MB and System 6 was still a usable configuration. The IIsi had 3mb or 5mb configuration, and also came with system 6 (I think by the time system 7 came out the IIsi was stock with 5mb). The IIsi could max at 65mb of ram. That was quite a lot in 1991-1994 during its usable life. 
 

The LC was targeted at the education and home market. It had 2MB soldered onto the board allowing system 6 to have quite some usability for that purpose. Apple sold this model with a hard disk or with two internal floppy drives for schools. One for booting the OS the other for running Apple IIe software loaded on 3.5 inch disks. The 2MB basic configuration was useless with system 7, which came out in March 1991 (only a few months after the launch of this machine). The 10Mb limit was reached with adding 2x 4MB SIMMs. 
 

The LC2 wasn’t released for another full year and by then Apple added 4MB to the onboard soldered ram instead of just 2MB. 
 

It took me a bit of consideration but I realized just recently why Apple hobbled the LC on a 16 bit bus. Partly so it wouldn’t compete with other models but I think mostly due to cost. A 16 bit bus allows SIMMs to be upgraded in pairs. One “pair” soldered onto the board, another pair in two sockets. This would have greatly decreased the cost of memory on the LC (and Classic II / Color Classic).  SIMMs and sockets were a lot more expensive than just integrating the base amount on the board itself. 
 

I worked in retail selling Macs between 1991 and 1995. Most people bought the machine as it came without adding RAM. Our target customer was the home, home office, and small business where the base amount of ram was likely sufficient. 
 

When Apple moved the LC series to the LC 3 platform, they soldered on a base amount of VRAM that allowed a 640x480 to run 256 colors. A single vram socket allowed upgrading to 768kb which supported 16 bit color at 640x480. The system came with 4MB soldered on, which was still enough for basic apps for the intended audience and a single SIMM socket limited the machine to 36MB, quite a lot in 1993. 
 

The LC class motherboard was always a smaller board and if you look it over, having 4 SIMM sockets would have been hard to do. I imagine there would also have been issues running that many traces but I’m not sure. 
 

The Quadra 800 had 8mb of ram soldered on, and came with 512kb VRAM soldered on. This allowed most of the basic users who purchased this machine to use it without upgrading anything. The video supported 256 colors at 640 x 480 and System 7.1 would have given the user almost 6.5mb of free ram. On this machine that would allow running PhotoShop and something like QuarkXPress simultaneously. A more high end user could choose to add vram on board or insert a NUBUS video card. With 4 SIMM sockets this machine could run 72MB when it was released. A huge amount for the day. 
 

One platform that was like going back to the 2 or 4MB onboard ram of the LC series was the early PowerMacintosh models. With 8mb on board this was like 4MB of ram on a 68k machine. At the time of release they were running System 7.1.2 but the system was basically 68k code. All of that 68k emulation required more System ram and hobbling it with only 8mb was truly a poor decision as was not including the cache on the base models. The 68k emulator really suffered without that added cache. 
 

In those days, Apple was competing with Wintel computers, and needed to keep costs in line with Pentium models. I’m sure releasing their base PPC systems with 16MB of RAM would have been so cost prohibitive that sales would have suffered dramatically. 
 

It’s always been the view of a company to manufacture the machine with only the base amount of ram so that it can be priced competitively and used immediately for basic functions. Most users in those days didn’t actually know what they were going to use their first computer for. Those were the days of people buying their first home/home office/small business computer. I personally setup first computer systems in many businesses in the 90s. Many still used typewriters, fax machines, and letter mail. When neither the manufacturer nor the user really knows the target use case, a minimum amount is best. That allows the machines to be sold at the lowest possible price while being usable immediately and yet allows the user to upgrade to whatever their needs require down the road. 
 

Also, things changed rapidly in those days. In 1989 1mb was pretty good. By 1991 4MB was good. By 1993 8MB was good. 1995 and the PowerMac 7200 brought 8mb expandable to ... 512mb of ram. HUGE. 

 

We look back at systems from 1989-94 and think the machines were severely restricted with “only” 4MB of ram. I used a Mac Plus with 1mb for years. I upgraded to a IIsi with 5mb. It was enough for me, on a severe budget.  By 1994 I had a Quadra 605 and 8mb of ram. I used RAMDoubler software because it was cheap, and allowed 8mb to be 16. I recorded music, burned CDs and developed computer software on that machine. 
 

Even today I have most of my vintage machines in my collection with mostly base amounts of ram. I’m starting to upgrade the ram in them mostly because I can. But I frequently boot up a LC3 with “only” 8mb of ram and I run what I want on it. If I need more ram I generally need a more capable machine to run it on. And that’s the point. 
 

Every 2 years machine specs almost doubled. By the time a user bought an LC and needed more than 10Mb of RAM (it’s limit) they likely were looking at the LC3, a machine at least twice as good. 
 

By 1995 the CDROM revolution hit, and a PowerMac 7200 offered 2-4x the capability of a LC3. I upgraded my machines with new ones almost every 2 years. 

Link to post
Share on other sites

Excellent points, this will make for a great thread in its own right if the mods will split it out and you're spot on about the IIsi having but 1MB soldered on board. Got confused, it can be easily upgraded to 4MB, easier even than taking the woefully inadequate 128K of the original Mac to 512K as was fairly common practice in the world of DTP. The addressing of Bank A in the IIsi has lines to support 8MB, but the IC package for that was never available.

 

I really don't think board real estate was a problem in almost any case however. The onboard ram  in the IIsi is as wide as the four SIMM slots and about 60% as long. That probably holds true until 72 pin SIMMs were implemented in the 605/475 era. Haven't measured, but the footprint of soldered RAM almost has to be greater than the socket for a 72pin SIMM?

 

To get directly back on the topic of VRAM SIMMs, your point about the LC architecture addressing but a single VRAM SIMM is very interesting. Such may explain why there was never a 1MB VRAM SIMM in hardware terms outside of the usual ROM limitations. If the supply of ICs for 512MB VRAM SIMMs should dry up, such might very well preclude using VRAM ICs on one SIMM with a simple jumper line setup connected to a "dummy SIMM?"

 

 

 

 

Edited by Trash80toHP_Mini
apparently I can no longer type or spell.
Link to post
Share on other sites
On 2/28/2021 at 1:49 PM, EvilCapitalist said:

 

I haven't been able to find the 112 pin VRAM DIMMs for the later Power Macs for some time either, but I figure those aren't in as high demand because you can just toss a cheap PCI Radeon in those and get way better video than what was onboard.

You need 4MB VRAM on the PCI Powermacs for video mirroring (using the A/V ports).

 

Those used to be so common and cheap I used to buy a bunch and strip the RAM off for PC video card upgrades (a long long time ago). Not sure what I have left of those.

Link to post
Share on other sites
22 hours ago, Trash80toHP_Mini said:

Excellent points, this will make for a great thread in its own right if the mods will split it out and you're spot on about the IIsi having but 1MB soldered on board. Got confused, it can be easily upgraded to 4MB, easier even than taking the woefully inadequate 128K of the original Mac to 512K as was fairly common practice in the world of DTP. The addressing of Bank A in the IIsi has lines to support 8MB, but the IC package for that was never available.

 

I really don't think board real estate was a problem in almost any case however. The onboard ram  in the IIsi is as wide as the four SIMM slots and about 60% as long. That probably holds true until 72 pin SIMMs were implemented in the 605/475 era. Haven't measured, but the footprint of soldered RAM almost has to be greater than the socket for a 72pin SIMM?

 

To get directly back on the topic of VRAM SIMMs, your point about the LC architecture addressing but a single VRAM SIMM is very interesting. Such may explain why there was never a 1MB VRAM SIMM in hardware terms outside of the usual ROM limitations. If the supply of ICs for 512MB VRAM SIMMs should dry up, such might very well preclude using VRAM ICs on one SIMM with a simple jumper line setup connected to a "dummy SIMM?"

 

 

 

 


The point I was making about the IIsi wasn’t space related it was in relation to the IIci. It was recommended on the IIci to have 4x256kb SIMMs in the first four sockets followed by 4, 8, 16 mb SIMMs in the last 4 sockets. Due to the vampire video. 
 

Apple likely looked at this situation and just skipped putting 8 sockets and instead soldered the required minimum 1mb so vampire video works efficiently right onto the board. 
 

With regards to the lack of space issue, I was referring to the LC, and it’s derivatives the Classic II and LC 520. There is not a lot of space on that board. Also, having 2 sockets and nothing soldered would have been an inconvenience for most people upgrading. They’d end up with SIMMs they had no use for. 
 

If Apple had used 4 sockets for SIMMs wouldn’t that mean it would need to have all 4 filled to be used?  If so, again the user would have to remove 4 good SIMMs and replace with 4 higher capacity SIMMs. What to do with the ram taken out?  It would be such low capacity so as to be useless. 
 

Better to solder on some ram and allow an upgrade with already empty sockets provided. 

Link to post
Share on other sites
  • 68kMLA Supporter

OK - UTSource have the memory chips available in quantity - i can manufacture them, just need the PCB, so if anyone has a dead 512k one, or wants to sacrifice one - i can do a sprint job on it. 

Link to post
Share on other sites
On 3/8/2021 at 6:58 PM, Kai Robinson said:

OK - UTSource have the memory chips available in quantity - i can manufacture them, just need the PCB, so if anyone has a dead 512k one, or wants to sacrifice one - i can do a sprint job on it. 

 

I have a very interesting VRAM module.  It's labeled 256KB on one side, and 512KB on the other side.  It has traces for chips on both sides, but if chips are only installed on top side, it's 256KB, otherwise if on both sides it's 512KB.  I don't know if there would be a clearance issue.  I'm thinking of desoldering two chips from another 256KB VRAM SIMM and soldering it on to see if it would work.

 

I don't have a dead 512KB SIMM, but if it's necessary for the project I could send you one.

Link to post
Share on other sites
On 3/8/2021 at 4:58 PM, Kai Robinson said:

OK - UTSource have the memory chips available in quantity - i can manufacture them, just need the PCB, so if anyone has a dead 512k one, or wants to sacrifice one - i can do a sprint job on it. 

 

Which chips are you looking at getting? I can put together a quick design, and then you can modify it if needed. I've already got a few designs, but they tend to vary based on chip.

Edited by SuperSVGA
Link to post
Share on other sites
On 3/9/2021 at 1:58 AM, Kai Robinson said:

OK - UTSource have the memory chips available in quantity - i can manufacture them, just need the PCB, so if anyone has a dead 512k one, or wants to sacrifice one - i can do a sprint job on it. 

 

I did this while working on my "HPV Riser Project", maybe it's useful. It's a 256Kb VRAM stick but this should be enough to get a rough idea

Datasheet : https://github.com/demik/oldworld/raw/master/EDA/HPV SRAM/Datasheet/V52C8_datasheet.pdf

Schematic https://github.com/demik/oldworld/raw/master/EDA/HPV SRAM/HPV_SRAM_V52C8128.pdf

 

If you want also the RAM to be useful with high resolution card such as the HPV card, you will need 60ns memory.

Link to post
Share on other sites
On 3/9/2021 at 6:56 AM, demik said:

 

I did this while working on my "HPV Riser Project", maybe it's useful. It's a 256Kb VRAM stick but this should be enough to get a rough idea

Datasheet : https://github.com/demik/oldworld/raw/master/EDA/HPV SRAM/Datasheet/V52C8_datasheet.pdf

Schematic https://github.com/demik/oldworld/raw/master/EDA/HPV SRAM/HPV_SRAM_V52C8128.pdf

 

If you want also the RAM to be useful with high resolution card such as the HPV card, you will need 60ns memory.


Would an old Mac LC work with 60ns VRAM chips though?  
 

Apple used 100ns on the original LC and 80ns on the 68040 Macs. Their information they put out around 1992/3 indicated you shouldn’t mix the different speeds on the same machine to avoid display issues, and you should use 100ns on the LC. 

 

I was challenged here to find proof that the speeds were important and that Apple actually said this. I can attest I myself have had issues with pixels along the top edge of the screen when using some 512kb vram SIMMs in an original LC. 

 

80ns shouldn’t be a problem for most machines. I know Apple did some 70ns on Quadra Macs and some say these are required on a 40mhz 68040 machine. I don’t know if 60ns would be fully cross compatible. 
 

I realize speeds of memory are always “this speed or faster” but I do remember back in the day it wasn’t always good to have ram timings that were a lot faster than original spec. 

Link to post
Share on other sites
  • 68kMLA Supporter
On 3/8/2021 at 2:05 PM, MrFahrenheit said:

The IIsi had 1mb soldered to the board, not 4. This was mainly because of the vampire video, requiring system ram for video. It was actually a good idea considering the way the video was. What I never could understand is how the LC could have dedicated and upgradable video ram but the IIsi used system RAM and was stuck at its amount. Back in 1993 I had a IIsi and I frequently ran it in black and white for extra speed. 

 

 

On 3/8/2021 at 7:14 PM, MrFahrenheit said:

The point I was making about the IIsi wasn’t space related it was in relation to the IIci. It was recommended on the IIci to have 4x256kb SIMMs in the first four sockets followed by 4, 8, 16 mb SIMMs in the last 4 sockets. Due to the vampire video. 

 

I hadn’t thought about the IIci like this. The old IIsi trick I know and definitely works is having enough “fast” 70/60ns? RAM in bank B where you can set the memory control panel disk cache to 1024KB to consume all of the slower bank A memory. This forces the “vampire” video memory to use the faster memory in bank B. (At least I think that’s what’s happening) My past tests show about a 10% CPU speed improvement:

 

 

I thought if you used “fast” 70/60ns RAM in both banks on the IIci it wouldn’t matter the video was being pulled from one bank but I think I’m following this logic. Similar but different from the IIsi’s fixed, slow, bank A on the IIci using that bank only for vampire video means bank B is never blocked from system/app use due to Video access. 

 

It would be neat to get some IIci 80/70/60ns 256k SIMMs to test this out...

Link to post
Share on other sites
  • 68kMLA Supporter
20 hours ago, Fizzbinn said:

set the memory control panel disk cache to 1024KB to consume all of the slower bank A memory.

There's also an INIT, IIsi RAM Muncher, which is designed to do this as well: suck up all the slow on-board memory. 

 

There's an old, old forum post (might even be way back in the archive) where a user took faster, higher density RAM from a 61xx logic board and replaced their failing IIsi board RAM. While I don't knock the exercise in principal, especially where one is needing to replaced bad memory anyway, I think a contemporary tactic would be very different: buy a graphics card. Even something like a Toby would likely be faster than on-board vampiric video. Even on a IIci, that might be a more sensible solution, if the Mac was being used for graphics—maybe not a Toby for the fastest contemporary graphics...but the point stands. The whole point of those machines was the convenience of not needing to suck up a NuBus (or an adapter + card) for something already on-board. But like I say, if you needed something faster at the time, plopping slightly faster RAM in the bank wouldn't net you as much gain as a RasterOps or Radius card.

The IIsi was cheaper than a IIci, and much more capable of doing office managerial type things than, say, and LC, which still gave you color. They were all built to a price point and covered a wide range, unlike today, where the gaps between models is actually very small for 80% of all tasks. If you're working that narrow 20%, you pay more for an HEDT or workstation class machine.

Here's a good visual example (sorry for lack of pixels) I appropriated from an eBay listing:
jdNOoDj.jpg

 

But like I say, we have fun with trying to juice our Macs to be the best they can be. Some folks are hardline against that, just want them to be what they be. And some folks like to tinker, have fun and so on.

Even contemporaneously people mixed and matched. The last IIsi I owned had an original owner working for Motorola, and did CAD on it, even though that chart says it's not meant for that. Maybe I'm undoing my whole argument...if I was ever arguing to begin with.

So hopefully I'm not deflating someone's balloon of fun with all of this.

Link to post
Share on other sites
  • 68kMLA Supporter
20 hours ago, jessenator said:

But like I say, we have fun with trying to juice our Macs to be the best they can be. Some folks are hardline against that, just want them to be what they be. And some folks like to tinker, have fun and so on.


Many aspects to this hobby, I think having fun with it, whatever that means for you, is the most important! :)

 

I think I like putting myself in the shoes of someone trying to make the best out of each of the machines in my collection from the timeframe prospective of when each one was just past its prime. Sometimes that’s neat optimizations, sometimes it’s trying to find the rare old upgrades (or recreations of them) that I could have never afforded back in the day...

Link to post
Share on other sites

I recently acquired a Performa 475, but have been at a loss to find 512K VRAM SIMMs for 16-bit colour at 640x480. Photoshop with 256 colours just isn’t the same.

 

I would definitely be interested in a couple of VRAM SIMMs if they can be made.

Edited by Flux
Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...

×
×
  • Create New...