Or, how to bump your Centris 650, Quadra 650 or Quadra 800 to 40 MHz with minimal effort.
A 40 MHz Quadra 650 (“Current System”)
Traditionally, the way we have overclocked the Wombat family of Macs (the 650s and the 800) is to desolder the CPU clock “G3” and replace it with a faster part, or alternatively, use a commercial product that “clips” onto the clock, disables it, and injects its own clock signal. Through a mixture of experimentation and, let’s be honest, the sudden availability of the schematics, I made the most awesome discovery, I’ve found that 40 MHz is easy without messing with the clock. This is great news, because…
The clock is a surface mount part, meaning if you want to swap between speeds, you need to make a socket / adapter.
The clock is right up hard against a NuBus slot meaning you have to be very careful when desoldering and soldering in the area.
It actually turns out that the logic board in these machines has a built in function that lets you select the clock source, either from G3, or a clock derived from the NuBus clock. Selecting the NuBus clock as a source just straight up runs your computer’s bus at 40MHz without having to do any particularly fiddly mods.
There are a number of “levels” of effort you can put in, each with different advantages and disadvantages. You can either just hard set your computer to 40MHz and overclock in terms of the timings (because the ROM still runs with 25 or 33MHz timings), change the timings to the correct ones for 40MHz by swapping resistors, or by adding an extra resistor and two jumper headings, make it that by just removing or fitting two jumpers, you can select (independently) the RAM/ROM/VRAM timings and the CPU clock speed (between either 20 and 25 MHz, or between 33 and 40 MHz).
Basic 40 MHz Overclock
This is the simplest overclock with the minimal effort.
Advantages – This is effectively an overclock, so you’ll get super fast video performance as well as fast RAM and ROM.
Disadvantages – This uses whatever timings you already have, so almost certainly be unstable if you have a machine that shipped as 25 MHz, and possibly unstable if you have a 33 MHz machine.
How to do the Basic 40 MHz Overclock…
Remove the logic board from your computer and place it on a heat-proof and static safe surface suitable for soldering.
Find the pair of pads next to the CPU called “J29” on the silkscreen.
Bridge the two pads, either with a blob of solder, or a little length of wire soldered between them. Alternatively, fit a two pin male header (2.54mm pitch) for a jumper.
Reassemble your computer and test to see that everything works.
J29 on the Schematic
The modification works because U71 is wired to select between two clock sources depending on if J29 is fitted or not. As shipped, J29 is always open, ClkSel is always “high” and so the clock signal (half the bus speed) is always taken from G3. If you short J29, ClkSel goes low and the chip switches to using C20M – a 20MHz clock used by things like the SCSI and EtherNet chip, sourced from the NuBus circuit. Whichever clock is selected is doubled by the MC88916 chip before being fed into the CPU. Don’t worry that R235 isn’t fitted – it isn’t needed.
The Location of J29
I recommend leaving a CPU in the socket while soldering to protect the socket from slips. In the photo above I’ve used solder wick / braid and flux to remove the solder that was in the through holes. You only need to do this if you want to fit a jumper (so you can select between the stock speed and 40 MHz by removing and fitting a jumper. The following shows my setup.
J29 with a Header and Jumper Fitted
If you don’t already have a CPU heatsink, consider fitting one.
Adding 40 MHz Timings
This is a more stable solution that requires a little more soldering. This mod is required if your computer originally shipped as a 25 MHz machine.
Advantages – This reduces the overclock on systems such as RAM, ROM and VRAM, likely making the system more stable at 40 MHz. Doing this change will make a 25 MHz machine work at 40 MHz.
Disadvantages – Benchmarks will score a bit lower than using 33 MHz timings at 40 MHz (in line with a real 40MHz machine if Apple had ever released one). The machine ID (gestalt) will change to an unreleased Mac – 7.6.1 will boot OK (but might not install without tricking the installer), but 7.1 won’t boot unless you can find a hacked System Enabler. (Update – I believe I’ve managed to make the required changes and have the magic System Enabler working now. Drop me a message and I’ll send you a copy).
How to select 40MHz timings…
Remove the logic board from the case and place it somewhere suitable to do some soldering.
If there is a resistor fitted to R152, remove it.
If there is a resistor fitted to R151, remove it.
Save any resistors in case you want to put them back.
Reassemble your computer and test to see that everything works.
Setting 40 MHz Timings
The way I remove 0805 surface mount resistors is by putting a little flux on them, and then placing the iron on one end for about 3 seconds, before quickly moving the iron to the other pad and pushing the resistor off. Sometimes it takes a couple of tries, but it is easier than faffing with other methods and I usually already have my iron set up. It is especially helpful if there is plastic parts nearby on the board that hot air might damage.
The Ideal Setup
This last modification is closest (see the “Other Tweaks” section) to how I have my system configured as it gives the best flexibility. It lets you have jumpers to select CPU speed and timings. Removing both jumpers gives you the stock CPU speed (whatever it is on your machine), with (always) 33 MHz timings, while fitting both jumpers gives you 40 MHz and 40 MHz timings. You can also mix and match.
Advantages – Flexibility – you can change the CPU clock between stock (hopefully 33 MHz) and 40 MHz, and also the timings between 33 MHz and 40 MHz settings. All without even removing the logic board from the case (once the modification has been done).
Disadvantages – If you have a machine fitted with the original 12.5MHz clock for 25MHz, it will be slower than as shipped when you pick 25MHz by jumpers, because you will only be able to select timings designed for higher clocks.
If you have a 25MHz machine, I recommend just doing the first two mods above and basically hard wiring it to 40MHz. Make sure you have a heatsink fitted and consider finding a 33MHz grade CPU.
Dual Jumpers
To fit dual jumpers…
Remove the logic board to somewhere you can do soldering.
Make sure no resistor is fitted at location R151.
Fit a 1.2k 0805 resistor (usually labelled 122) at location R152.
Fit a 330 ohm 0805 resistor (usually labelled 331) at location R233.
Fit a header at location J28.
Fit a header at location J29 if you didn’t previously.
Reassemble your computer and test to see that everything works.
J28 in Situ – This Selects 40MHz Timings when Fitted
R151, R152 (I didn’t have a 1.2k Resistor so used a 1.5k) and R233
At this point, as described above and if you started with a 33MHz board, you will be able to select 33 MHz with 33 MHz timings by removing both jumpers (I store them by placing them so they are on only one pin of the header), or 40 MHz with 40 MHz timings by fitting both jumpers. If you fit J29 and not J28, you will get a 40 MHz bus / CPU speed, with 33 MHz timings. This is an overclock with regards to the timings, so you will have boosted video, RAM and ROM performance. Most noticeable in the video benchmarks like the following :
Video Benchmark Score at 40 MHz with 33 MHz Timings
Note that while 40 MHz timings are selected, you will have the machine ID (gestalt) of an unreleased Mac, and System 7.1 won’t boot as mentioned above. If you want to boot in 7.1, just remove jumper J28.
Other Tweaks
Fit a 40 MHz Grade CPU and a heatsink.
Fit a socket for trying other clock speeds.
Fit a 25MHz clock to aim for 50 MHz.
Spoof a Quadra 800 power LED by shorting the appropriate two pins on the LED connector location. This makes your computer think it is a Quadra 800, which shipped with faster RAM, so I believe the ROM sets up slightly faster RAM timings.
Install the Wish I Were control panel to override the machine ID gestalt and reduce (but not completely) the number of issues from running a machine with an unreleased ID.
Replace the MC88916DW55 with an MC88916DW80 as this makes 50 MHz more likely in combination with a faster clock at G3. Fitting J29 still selects 40MHz, but removing it selects the faster clock at G3. Remember G3 is half the bus speed.
Remove onboard RAM and fit a single stick of very fast RAM (50ns EDO).
Fit a PowerPC Upgrade Card. Some will run at 80 MHz using the modifications described on this page.
I have personally done all the mods listed above to my machine, although I can’t promise they’ll all work for you, or are worthwhile.
Further Research
I worked out how to modify the System Enabler so that 40MHz works with System 7.1! Drop me a message if you want a copy. I might have previously posted it on Tinker Different – I forget.
Additionally, if your board shipped from factory with J28 and or J29 fitted, and especially R233 fitted, I’d be interested to hear from you. It’s likely that such a board would be a prototype and it would be great to check some component values.
This page is making a record of information I originally posted on 68kmla.
When retro Macintosh computers such as the IIci, Performa 475 and even early (beige era) Power Macintoshes were current, RAM was extremely expensive. As a kid trying to push the capabilities of what I had, I sometimes would be in a situation where I just didn’t quite have enough RAM to run a program, for example because the software requires a specific, newer, version of the OS (with a larger footprint).
Today, RAM for these machines is effectively cheaper, but still expensive for some types, hard to find for others (such as the IIx) and lastly, some computers such as the SE or LC have fairly low hard limits.
Choose Your OS Carefully
The most obvious solution is to use an older OS as the further back you go, the less memory the Finder and System tend to use. System 7.1 tends to use around 1 to 1.5MB, while if you’re able to run System 6, requirements plumit into the low hundreds of kilobytes especially if you avoid using the MultiFinder.
If you’re short of memory on a PowerPC, try running a version of Mac OS 7 – the only beige PowerPC that won’t run Mac OS 7 in some form that I know of is the Beige G3. Some machines will even run slightly older versions than the official documentation suggests – for example the 9600 will boot 7.5.3. Note you might need a “System Enabler” for some combinations of computer and OS. These can be found on Macintosh Garden.
Setting up multiple disk partitions or disks, with older and newer supported systems, can be a great way of switching between more feature rich / compatible systems that consume more resources, and smaller, older, more lightweight systems to maximise available RAM.
Optimise
Sometimes you can gain some memory by uninstalling certain software installed in the System Folder. You might consider removing your screensaver, for example.
Suggested things to remove…
Always remove QuickDraw GX – it isn’t needed and uses a lot of RAM.
AfterDark or other ScreenSavers.
Printer Drivers you don’t need (and desktop printing).
Modem drivers.
Ensuring the hard disk cache size is sensible – if you’re short of memory on a sub 10MB machine, ensure the hard disk cache in the Memory Control Panel is set to 128k or 96k.
If you are using a Power Macintosh, you will often find that enabling 1MB of Virtual Memory in the Memory Control Panel will actually free significant amounts of RAM, especially in Mac OS 8.* and 9.*. Some software refuses to run if Virtual Memory is enabled, other software won’t run without it. Generally, avoid using greater amounts of Virtual Memory – if you really need more RAM, buy more RAM. Real RAM is many times faster than Virtual Memory.
So You Have a II or IIx, or Perhaps a IIfx?
Getting RAM for these machines can be a nightmare, because they need special RAM. “PAL” RAM in the case of the II and IIx (plus complications with respect to ROM versions and memory controllers on the II), and unique form factor RAM for the IIfx.
This is the one and only situation where I would recommend using virtual memory (perhaps also the LC II, Classic II and ColourClassic), but I would specifically recommend the software “RAMDoubler”. Ignore the jeering PC enthusiasts who dismiss this sort of software as snake oil – their platform has a history of scams in this area. The RAMDoubler product is an impressive piece of engineering that attempts to Ballance available RAM, CPU and hard disk. Default settings will give you 100% extra RAM, by compressing the contents of a portion of your RAM. This increases your CPU load, but means the most likely to be reused parts of memory are normal speed while the remaining is still faster than hard disk based virtual RAM. I’m writing from memory, but I believe that the software also performs more agressive memory tidying and may move less frequently used items to disk… But I suggest reading the manual.
At Boot-Time
Another simple solution would could be to boot with shift held down to disable Extensions, but this wouldn’t always work if your program or game required certain extensions to function. Starting up with Extensions disabled reduces the RAM your OS uses to the minimum.
When your software mandates some extensions, you can create and save a minimum extension set using the Extensions Manager. This can be easily accessed by holding space during boot on System 7.5 and later (when it is installed), but it can also be added to earlier versions of System 7, and older versions exists even for System 6.
Extensions Manager allows you to add and remove Extensions and Control Panels and then save “sets”. You can create several, for each of your favourite RAM hungry applications or games. It is also useful for troubleshooting crashes during boot, as it loads early (before most Extensions), loads the interface by holding spacebar, and allows you to remove everything to confirm the OS boots with a minimum config (otherwise the issue is likely the core OS, hard disk or other hardware) add then things back few at a time until you identify what is causing the issue.
Who Needs the Finder Anyway?
The last trick I used to use as a child was to write AppleScripts to quit the Finder and then launch the Application you want. This only works in marginal cases, but has got me out of a hole a few times, including when an Installer application would not launch any other way on a particular machine. Only minimal AppleScript support is required because all applications understand the “open” apple event, but best AppleScript compatibility is with Finder 7.1.2 (so a copy of System 7.1 with Finder 7.1.3 or 7.1.4 installed is a good configuration) or later as this will enable you to do more extensive scripts if you want. A basic script might be as follows :
tell application “Finder”
quit
end tell
tell application “MemoryHog Pro 3.0”
activate
end tell
Save the script as an Application.
To produce a more extensive “Launcher” type AppleScript, you might consider using FaceSpan, which allows you to produce AppleScript based Applets with fancy GUIs using a nice IDE.
Summary
Hope this page is useful to someone. I remember having difficulty getting Escape Velocity to run due to memory restrictions, as well as trying to maximise RAM available for Director 4 (to increase the available Cast size) and for recording sound to RAM.
If anyone has suggestions for what Extensions and Control Panels are memory hogs, or for any other memory saving tips, send me a message via my contact form (under the About drop-down).
As well as the really nice graphics speed benchmark in the menu, the Mac OS version of MDK also has a built in frame counter! This can be handy for comparing the performance of different video cards on computers in this type of workload (pushing pixels to the screen – raw graphics performance not using QuickDraw) on PowerPC Macs. This game is surprisingly well optimised for a port to the platform. Genuinely, hats off to the programmers.
To enable the FPS display, type “frate”. If your computer is a bit slow, don’t type it too quickly, type it with purpose.
I swear I read about this feature on a website once, but I just spent half an hour trying to find this tip, even searching ‘MDK “frate”‘ and I didn’t get useful results. Hopefully this post will help someone else out.
MDK Frame Counter
The counter appears to give the FPS as “<current> (<average>)” in the bottom right, under the health HUD.
If you haven’t also seen the graphics benchmark, it runs routines manipulating on screen pixels and compares your computer to the performance of a 100MHz 601, a 145MHz 604e, and a 275MHz 603e. Basically you might say, an 8100/100, a 9500/150 and a 6400/275. Yes, I can’t find any 145MHz 604e based Macs, I’m not sure they existed, even that 9500 is possibly a 604, not a 604e. The benchmark is available at Main Menu > Options > Performance.
MDK Graphics Benchmark
Yes, I have been testing an underwhelming computer 😉
In the classic Mac operating system, files could have two forks, a conventional data fork, and a resource fork. The resource fork was sort of like a file system embedded within the file, with types and individual objects. It was an excellent way of storing structured data within files and applications.
Some of the Resources in the Lemmings “Music” file
One of these data types sometimes embedded into files was the music for computer games, such as Prince of Persia and Lemmings. These games used a sample based MIDI player to provide their music. As a kid, I noticed that these songs were stored as a resource type “MIDI”… moreover, if you *cough*, crudely pasted the content of these individual MIDI resources into the data fork of an empty file and typed it correctly as a MIDI file… they played in Cubase on our Yamaha TG 100.
While I’ve lost the original files I extracted years ago, I thought I’d re-extract them and make them available in case anybody is interested.
Note that these are the raw MIDI files, and in-situ, they were alongside INST and SONG resources which described which sample based instruments should be used for which individual tracks, as well as some tempo and maximum note length data. Additionally, some of the samples are percussion, and might not even be tuned to the same pitch (although this should be corrected in the “INST” resource, so I doubt it will be an issue in practice). This has a greater impact on the Lemmings songs, which seem to contain no instrument data themselves, while it seems that the Prince of Persia songs do at least try to select sensible instruments.
Downloads
The MIDI files here are straight from the games and so are sometimes a little weird – like include hanging notes instead of pauses, because game settings mean that sounds decay quickly. The sound samples are all the sounds stored as resources from the games, such as sound effects and instrument samples. I’ve provided them as both WAV files and in ResEdit Resource files. The latter wont work on linux or PCs and will be likely corrupted if you extract them from their .sit.hqx wrapper – this said, there isn’t anything you need in there that isn’t available as a WAV file elsewhere in the download.
I looked at extracting the songs from Prince of Persia 2 as well, but need to spend a little more time understanding how they’re formatted as they appear to have concatenated all the songs together in a small number of files and individual songs don’t appear to be terminated in the usual way? I’m not sure what exactly is going on, but it can’t be too complicated as MIDI itself is just a stream of commands to be interpreted.
Editing the Songs
It feels like it might be a fun and simple project to swap in some alternative MIDI songs and sound samples into a game. Other than a small amount of fiddling to set up instruments, it should be pretty much just a copy/paste job to swap different music into Lemmings enabling perhaps a custom version with your favourite instrumental tracks instead of the default music. With Lemmings this is especially easy since the music is stored in a separate file.
There are a number of classic Mac OS benchmarking tools available. Different ones have various benefits and personal preference is a major factor in choice. I’ve always used Norton’s System Info, which is provided as part of the Norton Utilities suite for Macintosh. I believe that it was first included with version 3 of Norton Utilities, but I need to verify that. As a kid, this was simply the only benchmark we owned, I like that it makes it clear what it is doing to benchmark (tests are named for the low level function that is being tested, often Macintosh Toolbox routines), and it uses a photo of two cats for the graphics tests that is clearly just a photo of a programmer’s pets. Its not even a very good photo, but this little human touch cheers me up every time I see it, and that’s worth a lot.
Lots of Norton Cats
System Info requires System 7 to run, and so an alternative benchmarking tool such as Snooper would be needed if you are running System 6. The newest versions of Norton Utilities claim to need Mac OS 8.* but it isn’t actually true.
Over time, there were few obvious changes within the System Info application. Included benchmarks and System used as the baseline (100 in tests) varied through time. I… think it started as the SE, but haven’t seen this version lately… but more common versions were the Quadra 700, and later the Power Macintosh 6100/60. The following shows System Info’s default window on launch. Later versions show a splash screen while launching, earlier versions did not.
Norton System Info Default Window
Features include
Native 68k and PPC support.
An extensive collection of included benchmark results files for both stock and some accelerated Macintosh computers.
Top level CPU, 2D (Quickdraw) graphics, disk and FPU scores.
Ability to select PPC or emulated 68020 performance on a PowerPC, and physical or software FPU where available.
Ability to select which video card and which disk.
Advanced mode which shows more detail during tests, and gives more detail in results. Graphical and tabulated results available, with the ability to show tabulated results in relative or absolute terms.
Results can be easily shared as saved individual files for each test run.
A detailed description of the computer and settings automatically saved with the results file.
Warns about some system settings which may impact performance, such as AppleTalk, Virtual Memory, video bit depth and Disk Cache.
A tiny application.
Ability to export tab separated values (TSV).
Disadvantages
New users don’t notice the need to enable advanced features and assume the tool is more simplistic than it is.
As is often the case, the performance benefits of a relatively small cache are possibly exaggerated as the tests seem to be repetitive and fit within the cache (difficult because caches give significant real world advantages, but not in all instances).
Tests are very much function based and not representative of real world activities.
Graphics tests are only QuickDraw 2D, and don’t include any QuickTime, RAVE, QuickDraw 3D, Glide or OpenGL tests.
Not scriptable?
The CPU speed shown in the main window in MHz isn’t always correct, especially if the accelerator is only enabled during boot, or is able to change speed.
Tips and Tricks
Even if the version of Norton you’re using is too new for the computer you are using, if you go into the “Norton Tools” folder and run “System Info” from there, it will launch in most circumstances, as long as you have System 7 or newer.
To enable advanced options within the software, select “Show More Choices” from the “Benchmarks” menu. This provides a number of additional drop down menus for fine tuning and seeing results in a greater level of detail throughout the application, as well as some additional menu items.
Selecting Show More Choices
System Info Default Window More Choices Shown for Volume and FPU, With the Latter Selected
Note that generally, the higher the video bit depth you run, the slower graphics performance will be, except where acceleration is only provided at higher bit depths. It was fairly common for video cards to not provide acceleration at bit depths below 8bit, and a small number of cards only provided 24bit acceleration.
Screen resolution doesn’t appear to have a significant impact on video scores for many machines, although running at 640×480 is probably best to ensure consistency with built in benchmarks. This said, Computers that use main memory as VRAM like the IIci, IIsi and 6100 can see significant CPU performance changes depending on video settings.
When in the “System Ratings” window, selecting a system from the list and clicking the “Get Info” button (see image “All Results Windows” below) will show a description of the system the benchmark was performed on. Details are given of both hardware, software and some system settings.
Additional results supplied, but not shown by default, can be added to the various results windows using the “Add Systems To List” menu item from the “Edit” menu.
Add Systems To List
Running a Test
Use the “Selected Suites” checkboxes to enable and disable each of the CPU, video, disk and FPU test suites and if in advanced mode, select between the real and emulated CPU, video card, disk (in reality, the list given is mounted partitions) and FPU or software. Not all of these options will be available at all times, depending on the computer under test.
Once the options are as desired, click the “Run” button. Note that at this point Norton will inform you of any System settings it has detected which will impact the scores and which it thinks you should change. Unless you’re doing specific testing, such as comparing 24bit mode between video cards, it is generally advisable to follow this guidance. A restart may be required for some changes (for example, disk cache, and AppleTalk on some System versions).
After this the tests will start. Depending on the performance of your computer, they may take quite a few minutes to run, especially the disk tests. In the advanced mode, you will be given greater detail of what test is currently being undertaken and what each test scores. This can help with the boredom of waiting for the tests to complete.
Once the benchmarks have completed, System Info automatically opens the System Ratings window, and lists the tests just undertaken as “Current System”. It is probably a good idea to save the results at this point if you plan to keep them for comparison.
Reviewing Results
All Result Windows
Note – The screenshot above incorrectly shows zeros for all results. This is an oddity of running the software within some emulators. Something I did for convenience when getting screenshots. On real hardware you will see actual scores for your and other computers.
Viewing Overall Rating / Test Category / Individual Tests
At the top left of the “System Ratings” window, there is a pop-up menu labelled “Show”. This allows you to select each of the four test categories, CPU, Video, Disk or FPU, as well as the default overall summary “System Rating”. The numbers to the right of each system name represent a score, relative to the baseline system (e.g. scaled where a Quadra 700 equals 100, or a 6100/60 equals 100 etc. dependant on the version of System Info you are using. The baseline system is identified in the list by the text “(reference system)” following its name). The bars to the right of these numbers (ordinarily) vary in length and graphically represent the score for each system.
If all tests were not run, the Current System will show as a dash “-” for the tests which were not completed, as well as for the overall “System Rating”.
If you are in advanced mode (i.e. with “show more choices” enabled), an additional drop down menu is available by clicking the downwards pointing triangle to the left of the “System Rating” column heading. This menu allows you to select individual tests for the category currently selected, for example, if you have selected “Disk”, then you will be able to choose to display specifically the random write, or perhaps 1K read test.
It is worth studying the results at this level, especially with respects to video accelerator performance, because it can be interesting to see how different cards perform in different areas. A card that scores well overall might be weak when it comes to rendering text compared to another card that scores lower overall. In this case, if your primary use case is… rendering text – the supposedly “slower” card might be the best card for you as it performs better in your application that mainly renders text.
Video Individual Test Results MenuFPU Individual Test Results MenuDisk Individual Test Results MenuCPU Individual Test Results MenuEach Menu for Individual Tests
To assist in comparing results on a test by test basis, clicking the “Show Details” button presents tabulated results. Within this window, a drop down menu enables you to select between “Absolute” and “Relative” results. I find that generally, “Relative” is more usable with the exception of disk scores, as the “Absolute” scores for the disk results are given in KB/s, which is meaningful.
If you manually select a subset of results in the “System Ratings” window before clicking “Show Details” only details for those, plus the baseline system, will be shown.
Tabulated Results
Exporting Results
While in the Detailed Ratings view, pressing Command-C will copy a summary of all results to the clipboard. This text can then be pasted into another application, such as for example, in the following screenshot, a text document. To remove unwanted entries either select a subset of results in the “System Ratings” window before clicking the “Detailed Ratings” button, or alternatively quit System Info and move unwanted benchmark files in the “Benchmark results” folder into the “Hidden Results” folder. Any results in the hidden folder wont show in the results lists / export. Moving them back afterwards will make them re-appear next time you relaunch the program.
Results Copied from the Detailed Ratings View, and Pasted into the Text Editor BBEdit
Note that the exported text includes all result types (CPU, Video, Disk and FPU), not just the results that were currently being viewed. The data is Tab Separated Values (TSV) and should be fairly trivial to paste into a Spreadsheet such as Excel or ClarisWorks.
Moving data to a modern computer can be easily achieved with BBEdit. I recommend BBEdit, because it will run on 68k Macs, and is aware of the three common line ending types. By pasting in the data, selecting save, and then clicking the “Options…” button, you will get the following dialogue box, where you can select your preferred line ending type. I’m picking “Unix”, which is best for Linux and modern versions of Mac OS, but if you are on Windows, you will likely want to pick “DOS”.
Saving for Other Platforms
Name the file, including a “.txt” or “.tsv” file extension and save. Once you have moved the file to your modern computer, you should be able to trivially import the data. You may have minor issues if there are unusual characters, due to the classic Mac OS Roman text encoding. The following shows my example exported collection being opened in LibreOffice Calc.
Importing Norton System Info Results into LibreOffice
Note – You should probably uncheck “Comma” and “Semicolon” if you have used these characters in your system descriptions, or if you are using continental European number formats.
The following screenshot shows the text file imported with no further manipulation, other than creating a chart based on the “Current System” and “Overall” columns of the first table of data.
Working with Norton System Info Data in LibreOffice Calc
Description of the Individual Benchmark Tests
The following is an extract from the Norton Utilities 3.2 manual.
Manual Extract for Norton System Info Benchmarks
Variations in test results of up to 1 percent are normal. For disk tests, variations of up to 7 percent are normal due to disk drive operation.
CPU Benchmarks
Some CPU tests actually test more than the CPU itself. For example, some tests access memory (RAM), which tests not only the CPU, but the system bus, memory speed, and caching (if any). These factors can influence test results substantially. For the purposes of benchmarking, these factors are all considered to be part of the CPU.
BlockMove Aligned and BlockMove Misaligned: Measures the rate at which the Macintosh can move memory using the ROM trap BlockMove. On all Macintosh computers, BlockMove is highly optimized and thus is a good measure of memory bandwidth (how much data can be moved per unit time). An aligned and misaligned case are used because on some Macintosh computers performance is much poorer when data is not aligned properly. Newer models, such as the Power Macintosh computers, have excellent performance in both cases.
Memory Read and Write: This is also a memory bandwidth test, but the benchmark accesses memory in the way a program normally would and does not make use of any special machine instructions.
Function Call: Measures the overhead associated with making a series of function calls with arguments. Most application programs make large numbers of function calls.
Bit Shifts: Measures how fast bits can be shifted in a machine word. This type of instruction is used frequently by a wide variety of software.
Multiply and Divide: Measures speed of integer multiplication and division. The tests are slanted more towards multiplication. Application programs frequently use this type of instruction to access arrays.
Branches: Measures how fast code containing conditional branches can run. Branches are on of the most common types of instructions used and can substantially affect performance on some CPUs. Power Macintosh computers have a special processing unit dedicated to processing branch instructions.
Instruction Overlap: Measures the ability of the CPU to execute integer (not floating point) instructions at the same time it accesses memory. The benchmark is written in such a way that memory access should be able to execute at the same time all intervening computations are performed.
Sort: A high-level benchmark that sorts a large array using a quick-sort and insertion-sort hybrid algorithm that makes heavy demands on memory and looping constructs.
Tree: A high-level benchmark that builds a large, sorted binary tree by repeated insertion. Good caching and/or memory bandwidth can speed up this test considerably.
Search: A high-level benchmark that performs a sophisticated string-searching algorithm that makes extremely heavy demands on memory in a partially-sequential way. This test is designed to access memory in a way that most machines cannot cache effectively without a substantial amount of cache memory.
Video Benchmarks
The Video benchmarks test the speed at which common Quickdraw operations can be performed. The benchmarks are designed to measure frequently used operations – not esoteric transfer modes or rarely utilized Quickdraw routines. For example, scrolling operations are emphasized in the overall video rating because they are heavily used by all users. Note that as the video bit depth is reduced, video performance generally increases. If you compare video performance on two different systems, make sure to test at the same video bit depth.
Rectangles, Round Rectangles, and Ovals: These tests cycle through painting, erasing, filling, framing, and inverting the respective shapes with different fill patterns and colors, where appropriate.
Lines: Draws lines of varying slopes and colors. The video circuitry of some Macintosh computers is highly optimized for drawing lines.
Picture: Draws a bitmapped picture at the same bit depth of the screen. It tests the ability of Quickdraw to decode a PICT format picture and place its bits on the screen.
CopyBits: Measures how fast bits can be moved around on the screen using the CopyBits ROM routine, which is heavily used in screen drawing. The small cases are 32 by 32 (the size of a large icon). The small cases are significantly impacted by the overhead of calling Quickdraw and its setup. The large cases minimize this overhead and, thus, more accurately test the actual speed at which bits can be moved. Aligned (to a 32 pixel boundary) and misaligned cases test optimal and non-optimal situations.
DrawText: Measures how fast a sentence can be drawn on the screen. Primary emphasis is placed on drawing plain text, although bold and italic text are tested as well.
Scrolling: Measures how fast bits can be scrolled on the screen. Emphasis is placed on vertical scrolling, but some horizontal scrolling is also performed. Results of this test vary widely depending on the bit depth of the screen.
Disk Benchmarks
By design, the disk tests do not isolate disk speed alone. The benchmarks include factors applications normally experience, such as CPU speed, SCSI bus speed, impact of the disk cache, and fragmentation.
Random Read and Random Write: These benchmarks read or write data into a file in three 1MB bands. The data size read and written varies from 4 bytes to 4 kilobytes. These tests reflect seek time, the effects of the disk cache, and how fast a drive can process numerous small requests to read or write data. They are a good indicator of overall disk performance.
Sequential Read and Sequential Write (1K, 4K, 16K, 64K, and 256K): Measure how fast the drive can read/write data when requested in a specific size chunk. Some drives can transfer data efficiently only in large chunks, others are efficient with many small chunks, but slow down with large chunks. These sequential tests intentionally bypass the built-in disk cache (to eliminate cache overhead) and provide a better indicator of performance of the disk itself. Smart, disk-intensive applications such as Norton DiskDoubler Pro bypass the disk cache, when appropriate, to provide maximum performance. Few current applications employ this technique, however.
FPU Benchmarks
All FPU tests are directly comparable among themselves. For example, you can directly compare the speed of a multiply to an addition, square root, or cosine.
A single-precision floating-point number is 32 bits. Because of differences in the floating point formats supported on various machines, a double-precision number varies in size. For double precision, the fastest format available that is at least 64 bits in size is used.
Multiply, Divide, Add, and Subtract: Measure basic FPU performance in both single and double precision. On some machines, double precision may actually be faster than single precision because single precision numbers must first be converted to double precision before the calculation occurs.
Integer To Single and Single To Integer: Convert an integer number to a single precision floating point number and vice versa. Some machines have hardware instructions for both operations, some have neither, and some have just one. This accounts for dramatically different results that you may encounter, depending on which way the conversion goes.
Sine, Cosine, Tangent, and Arc Tangent: Measure trigonometric performance on double-precision numbers. These results can vary dramatically from computer to computer. Some older models of the Macintosh can actually outperform the new Power Macintoshes on these functions because the Power Macintoshes implement these routines in software.
Absolute Value, Square Root, and Log 10: Measure the absolute value, square root, and logarithm in base 10, respectively, of a double-precision number.
Vector: Measures how fast an array of numbers (vector) can be manipulated. Each element in the array is multiplied by a constant, a constant is added, and the result is stored back into the array (X’ = aX + B).
Getting Norton System Info
Norton System Info was provided as part of the Norton Utilities for Macintosh suite of software. The best way to get a copy of this is by buying a genuine copy on eBay, although I’m quite sure there are copies on Macintosh Garden.
Feedback on this Post
If you have any additional information which you wish to suggest is added to this page, or would like to correct any errors / inaccuracies, please use my contact form to let me know.
I childishly ordered a Nakamichi MJ-5.16, 5 disc SCSI CD-ROM changer from the US a few weeks back, not sure if it would work in my beige beasts. It arrived earlier today, so I popped it into my 9600 (after a false start where I forgot to set the SCSI ID). Other than the SCSI ID, no other jumpers were required for my setup, although there is a Termination option if required.
As expected, the drive wasn’t recognised by the stock Apple driver, so I dropped in the the 5.3.1 “works with almost everything” version, which recognised the drive and mounted the disk inserted into the first of the 5 positions only. This was promising, but not good enough, so I removed it and continued onwards.
The drive uses “LUN” to provide a number of sub devices under a single SCSI IDs. Macs have supported this since SCSI Manager 4.3 and the Quadra 840av, but it doesn’t appear to have been implemented in many drivers, so I set about testing every driver I could find. After a couple of hours I finally found that the drive fully worked with FWT CDT 4, specifically after installing the Jukebox optional software.
The software gives a few options and an extra menu next to the Applications Menu and seems to work exactly as required.
The only issue is that ejecting 5 discs takes a couple of minutes and is quite an involved process!
The strangest thing about this drive is that they can be bought at reasonable prices, despite being a bit of a novelty, even though SCSI drives usually demand a premium. Sadly the listing I got mine from is gone and alternatives are double the price. Happy hunting.
The Highest speed of Sonnets “PCI” (as in, for most PCI based beige Macs, rather than it fits in a PCI slot) G4 upgrade cards used a different PCB and Motorola’s 7455 G4 Chips. These boards can be a little unstable and I’ve never had any luck using them with the actual Sonnet drivers. These cards were available with a 700, 800 and 1000MHz processor.
The most stable solution I have found is to use the PowerLogix CPU Director drivers, with Speculative Addressing disabled.
To speed up the process of getting this set up, I have created an installer for Mac OS 9 which installs the PowerLogix software, as well as a system preference which has the settings already enabled as required.
The download is a .bin file, which can be opened in Stuffit Expander. Inside the .bin file is an installer application called Sonnet7455.
The process…
With the previous CPU still in the computer, install this software, once done, power down the computer.
Remove the previous CPU and install the Sonnet Upgrade.
Make sure the CPU card is firmly seated.
Seriously, double check it – they can be quite stiff and if they’re not properly seated you might kill the computer or card.
Really – check it again, I lost a 500MHz G3 like this. Stupid thing resisted so much I thought it was in properly, and it was one of the ones without the tall bracket so I didn’t notice that it was higher than it should have been in my 8600.
Press the “Cuda” button. Usually a little red tactile switch. Press it once, for about one second. Do not long press it. Do not press it lots of times.
Close up the case and boot the machine up. You will have lost all your mouse / date and time etc. settings due to pressing the Cuda button, so they’ll need resetting.
At this point you should be good to go, although these upgrade cards are known for having incompatibilities. I have seen the following issues :
Sometimes it doesn’t like my OEM Twin Turbo card. It works fine with my Radeon 7000.
It doesn’t like my ethernet card and will crash if I start transferring a larger file. It works find with the built in ethernet.
There are more compatibility issues listed on the Sonnet FAQs.
I recently picked up a non-functioning Macintosh IIx which appeared to have suffered from battery and capacitor corrosion damage. After recapping it and some trace repair, I found the hard disk was in the process of failing, although I was able to boot it from the SCSI2SD from my Macintosh SE (4MB, 40MB, 800k). This wasn’t a long term solution as the SCSI2SD really belongs with the SE!
I decided to find a second hand SCSI hard disk and started searching eBay. Following a discussion with a seller (conistonmal) who was selling a bunch of SCSI 3, 68 pin drives and me asking them would they sell the disks in smaller quantities, I discovered that they were actually also looking to clear out a stock of 50pin SCSI 2 style drives. These drives fit straight into most traditional (pre G3) Apple Macs. Given the collection of machines that my family has, I took the opportunity and bought a load, especially as the seller was fully testing each disk before selling them.
Connecting the Drives
I decided that since there was so much room inside the IIx (it was designed for a 5.25″ hard disk), I would make an adapter bracket to take two hard disks side by side. This required making the physical bracket, matching the interfaces to what was already there, as well as making up my own double headed SCSI cable of the correct length, and obtaining a Molex extension and “Y” splitter. I have also ordered a small but quiet 40x10mm fan that I believe will be able to encourage airflow through the rear vents and over the disks as one of the disks gets quite warm (it is a 10,000rpm 18GB drive).
Secondly, as I had a few drives to format, I decided a needed a faster / more elegant way to connect and disconnect the drives quickly to format and test each one. I bought a dual voltage mains connected Molex Power adapter and pinched the DB25 to 50pin IDC adapter from my SCSI2SD board.
I placed a fan I usually use to cool my overheating work laptop next to the disk to make sure it had sufficient cooling.
Drive Settings
How sensitive to incorrect termination settings your setup is depends on the specific Mac – you should set it correctly, but you might not notice sometimes as you seem to get away with it on some machines. The main jumper settings that needed attention on the disks I was using were…
Termination enabled (on the last internal and external device)
Termination power from bus
SCSI ID
Specific to the model of Seagate drive I was using (the ST39216N), I used J6 to set the SCSI ID and J2 for various (mostly termination related) settings. The LED output was also on J6, but I did not use this. Unusually, these disks have a built in activity LED – I don’t think I’ve seen that before. The manual for the drives I was using can be found here. Remember to search for the drive type that you have to find the correct settings, as it is unlikely that everything on your disk will be identical. If your drive does not include termination, you may need to ensure that there is a device which does on the end of the SCSI chain, or buy a stand alone terminator.
My Disks Jumper Settings
Following the details shown above from Seagate, I set my disk to “Term. Power from SCSI Bus” and “Terminator Enabled”. The termination power setting is the one I understand to be correct for Macs, and I enabled the built in terminator because the disk was the only device connected externally to the SCSI bus. Only the final item on either the internal, or external SCSI connector needs to be terminated, terminating other devices or no devices will potentially cause issues. Note that the IIfx is weird so look for additional instruction if you are lucky enough to have one! The following photo shows my settings on J2. J2 and J6 both use 2mm jumpers on this disk. They can be bought from eBay if you need any.
Software
You will need the following software (alternatives are available) :
Macintosh System 7.5.3 Disk Tools
Drive Setup 1.7.3 (Patched)
System 7.5.3 combined with Drive Setup 1.7.3 allows you to format a disk in 4GB partitions and should run on anything back to a Mac Plus if it has enough RAM (possibly earlier machines with some creativity). While I wouldn’t pick it for a daily driver OS on something like my SE, it is fine for running the disk tools. From my limited testing, these 4GB partitions appear to be bootable and usable from System 7.1 (with Finder 7.1.3 installed to get the Scriptable Finder and a few other features), which covers my use case as this is my preferred system for the IIx. If you’re planning to frequently use older operating systems, I recommend keeping partition sizes at less than 2GB. Note that limit on the number of partitions is quite high – they are treated as open files and you are allowed up to something in the region of 170 open files. Extensions and Control Panels are also counted as open files, so you don’t want to get too carried away, but you are more likely to run out of room on your desktop in terms of pixel real estate, than hit the partitions limit with 2GB partitions.
I have created a disk image that boots Most Macs that have SCSI and automatically launches the Drive Setup utility to allow you to format your new disk. I believe it will boot machines from the Plus up to the Power Macintosh 8500. The only pre-G3 machines I don’t believe it will boot are the 7300, 8600 and 9600 – I’ll see if I can solve this at a later date! Note that these machines all have CD drives so this is less of an issue as you can easily write a boot CD.
The version of Drive Setup is patched and allows the software to work with non-Apple disks. The version selected allows you to up to 4GB partitions even on older (e.g. 68030 and probably older) machines. You can either write the disk images to floppy disk and boot from it, or just fetch the patched version of Drive Setup from it and use your own OS. Note to use this version of Drive Setup you’ll need to be running at least System 7.1.2, or possibly even 7.5, and you can’t format the disk that you have booted or launched the application from.
You can download my disk image from here as a Mini vMac / DOS style disk image :
Or here as a Disk Copy image :
Formatting
Prior to starting up your computer, make sure your jumper settings on your SCSI disk are correct, connect it to the SCSI bus (internally or externally), then connect the disk power (the Molex connector). Make a note of the SCSI ID you have used so that you can select the correct disk to format. Then, either boot your Macintosh from a floppy of the Bootable Drive Setup image, or otherwise from your existing System install if it is new enough.
Note : When booting from the Drive Setup floppy, you will not see the desktop or Clock in the upper right. This is normal.
Once you reach the desktop, the new disk may appear if it has been previously formatted, don’t worry if it doesn’t. If the drive to be formatted does mount, unmount it by dragging the disk icons for all of its partitions to the Trash / Wastebasket. Launch the patched version of Drive Setup.
The new disk should appear in the list, identified by its SCSI ID which you most certainly made a note of earlier. It will also be identified as “<not initialized>”. Select this item (make absolutely sure you select the right one or you could delete your files!). Click Initialise (“Initialize”).
Click on the “Custom” button.
From the drop down menu, select the partitioning scheme that best suits your needs. I chose to divide my disk into the smallest number of equal parts that resulted in partitions les s than 4GB. This resulted in 3 partitions of approximately 3GB each on my 9GB disk. You can drag the little black tab on the selected partition to resize it.
Confirm your choices (if you are sure).
Drive setup will then format the drive, and mount the partitions as “Untitled” and a number (you won’t see this if you’re using the bootable floppy). Feel free to rename them.
At this point you are basically done. You can install your software on the drive and boot from it if needed. I particularly like that this method uses the Apple driver rather than a 3rd Party disk driver.
If you booted from the Drive Setup floppy, quit Drive Setup and you will be asked if you would like to shutdown or restart.
Appendix A – Using Apple HD SC Setup v7.3.5 (Patched)
It is also possible to format your drive using the other formatting tool “Apple HD SC Setup”. I don’t recommend this for a number of reasons.
I find that it sometimes crashes with large disks
It is very slow – it takes 15 minutes to do with what Drive Setup does in 20 seconds
When you create a completely custom partition scheme, it doesn’t seem to enable the mount on startup functionality on your partitions, or format them.
Do not be mistaken in thinking that the application has crashed unless it throws a System Error – it is really slow when you format a 18GB disk with it.
Where Apple HD SC Setup seems to provide more functionality than Drive Setup is in the flexibility it gives in creating partitions. You can easily (if slowly) delete, create, size and locate individual partitions, and it offers more formatting options (AUX) than Drive Setup. I have used Apple HD SC Setup to modify existing partition schemes and then used Drive Setup to re-enable mount on boot. Note that the menu option in Drive Setup is picky. I found that I had to restart, then unmount all partitions of the drive, then launch Drive Setup and select the disk before the appropriate menu item for setting the correct flag appeared. It may be worth using a 3rd Party utility to do such work instead.
Finally note, there have been times when I’ve broken the driver on a disk and it will not mount. Selecting Update Driver in Drive Setup has not helped… but doing the same in Apple HD SC Setup has. This is a useful thing to know when trying to save yourself after you’ve trashed a disk.
Appendix B – Booting from Individual Partitions
I have been having some issues booting from individual partitions on my Mac IIx. When it first arrived it would only boot from the first partition on any physical device. I later installed some software including the Startup Tuner system extension as I understood this fixed such issues and for a period of time I was able to boot from any extension… until yesterday. I do not know what has changed (other than there are now two devices on the SCSI bus), but I can no longer select a specific partition in Startup Disk (it selects, but if I quit and return, all partitions on that drive are now selected).
Any advice welcome – I can’t remember if this is a disk driver issue, a ROM issue or a System software issue. I believe that the functionality wasn’t always there, and that you have to do specific things to enable it on older machines like the IIx… but my memory of the specifics fails me!
UPDATE! I’ve found that using “Startup Disk” from Mac OS 8.0 or 8.1 solves the issue. It doesn’t need to be installed in the System Folder, so I just keep a copy on my desktop.
