#retrocomputing folks: I'm trying to get a sense of the proportion of people here who are into retrocomputing today but didn't experience certain classes of machines when they first came on the market.
-
#retrocomputing folks: I'm trying to get a sense of the proportion of people here who are into retrocomputing today but didn't experience certain classes of machines when they first came on the market. I want everyone's input! Please boost!
These questions are specifically about all 32-bit comsumer machines: PCs, Macs, RISC machines, you name it. Mid 1980s to 2005-ish.
(I'm also concurrently running a poll for 8 and 16-bit home computers and another for minicomputers. I'll probably have to do another for UNIX Workstations and LISP machines!)
Hat tip to @Foritus!
-
#retrocomputing folks: I'm trying to get a sense of the proportion of people here who are into retrocomputing today but didn't experience certain classes of machines when they first came on the market. I want everyone's input! Please boost!
These questions are specifically about all 32-bit comsumer machines: PCs, Macs, RISC machines, you name it. Mid 1980s to 2005-ish.
(I'm also concurrently running a poll for 8 and 16-bit home computers and another for minicomputers. I'll probably have to do another for UNIX Workstations and LISP machines!)
Hat tip to @Foritus!
@fluidlogic @Foritus Hell, I had access to a 24-bit computer that needed a software hack to go 32-bit.https://en.wikipedia.org/wiki/MODE32
-
@fluidlogic @Foritus Hell, I had access to a 24-bit computer that needed a software hack to go 32-bit.https://en.wikipedia.org/wiki/MODE32
-
@fluidlogic @Foritus Hell, I had access to a 24-bit computer that needed a software hack to go 32-bit.https://en.wikipedia.org/wiki/MODE32
If MODE32 works on it, then the hardware is 32-bit, but the ROM code isn't.
The problem is that said ROM code uses the upper 8 bits of pointers to store flags instead of address bits, and disables the upper 8 address lines.
This makes sense on the original Mac, whose 68000 has only 24 address lines and *always* ignores the upper 8 bits. But a 68020 or newer has 32 address lines, so this ROM behavior wastes the CPU's potential.
MODE32 patches the ROM to fix this.
-
If MODE32 works on it, then the hardware is 32-bit, but the ROM code isn't.
The problem is that said ROM code uses the upper 8 bits of pointers to store flags instead of address bits, and disables the upper 8 address lines.
This makes sense on the original Mac, whose 68000 has only 24 address lines and *always* ignores the upper 8 bits. But a 68020 or newer has 32 address lines, so this ROM behavior wastes the CPU's potential.
MODE32 patches the ROM to fix this.
There are also apps that do the same “upper 8 bits are for flags” thing. If you are in 32-bit mode (i.e. all 32 address lines are enabled) while such an app is running, the app will read/write the wrong memory locations, causing crashes or worse.
As I recall, apps were supposed to use the ROM memory manager rather than making up pointers themselves, and would therefore work in 32-bit mode, but some apps were naughty, presumably in order to run faster.
-
There are also apps that do the same “upper 8 bits are for flags” thing. If you are in 32-bit mode (i.e. all 32 address lines are enabled) while such an app is running, the app will read/write the wrong memory locations, causing crashes or worse.
As I recall, apps were supposed to use the ROM memory manager rather than making up pointers themselves, and would therefore work in 32-bit mode, but some apps were naughty, presumably in order to run faster.
Only a few Mac models ever had this problem: the Mac II, IIx, IIcx, and SE/30.
Earlier/lower-end Macs had a 68000 or 68010 and therefore couldn't do 32-bit addressing at all. In later Macs, Apple fixed the ROM and 32-bit addressing was always enabled.
-
#retrocomputing folks: I'm trying to get a sense of the proportion of people here who are into retrocomputing today but didn't experience certain classes of machines when they first came on the market. I want everyone's input! Please boost!
These questions are specifically about all 32-bit comsumer machines: PCs, Macs, RISC machines, you name it. Mid 1980s to 2005-ish.
(I'm also concurrently running a poll for 8 and 16-bit home computers and another for minicomputers. I'll probably have to do another for UNIX Workstations and LISP machines!)
Hat tip to @Foritus!
My family's first computer had a 486. The badge on the case proudly announced that it was 32-bit.
It ran MS-DOS and Windows 3.1, both of which were generally regarded as 16-bit. At the time, I lamented that the 32-bit capabilities of its CPU were wasted on this software configuration.
-
My family's first computer had a 486. The badge on the case proudly announced that it was 32-bit.
It ran MS-DOS and Windows 3.1, both of which were generally regarded as 16-bit. At the time, I lamented that the 32-bit capabilities of its CPU were wasted on this software configuration.
Little did I know that Windows 3.1 very definitely does use 32-bit addressing! Apps run in real mode and still use the 8086's weird segmented addressing, but the Windows kernel runs in 32-bit mode and maps memory in and out of that space, somewhat like swap on a modern operating system (except triggered by GlobalLock calls instead of page faults).
-
#retrocomputing folks: I'm trying to get a sense of the proportion of people here who are into retrocomputing today but didn't experience certain classes of machines when they first came on the market. I want everyone's input! Please boost!
These questions are specifically about all 32-bit comsumer machines: PCs, Macs, RISC machines, you name it. Mid 1980s to 2005-ish.
(I'm also concurrently running a poll for 8 and 16-bit home computers and another for minicomputers. I'll probably have to do another for UNIX Workstations and LISP machines!)
Hat tip to @Foritus!
@genehack raise your hand if you know it
-
If MODE32 works on it, then the hardware is 32-bit, but the ROM code isn't.
The problem is that said ROM code uses the upper 8 bits of pointers to store flags instead of address bits, and disables the upper 8 address lines.
This makes sense on the original Mac, whose 68000 has only 24 address lines and *always* ignores the upper 8 bits. But a 68020 or newer has 32 address lines, so this ROM behavior wastes the CPU's potential.
MODE32 patches the ROM to fix this.
@argv_minus_one @flargh @fluidlogic @Foritus TIL that there’s a worse option than 8088-style “segmented” memory addressing
-
#retrocomputing folks: I'm trying to get a sense of the proportion of people here who are into retrocomputing today but didn't experience certain classes of machines when they first came on the market. I want everyone's input! Please boost!
These questions are specifically about all 32-bit comsumer machines: PCs, Macs, RISC machines, you name it. Mid 1980s to 2005-ish.
(I'm also concurrently running a poll for 8 and 16-bit home computers and another for minicomputers. I'll probably have to do another for UNIX Workstations and LISP machines!)
Hat tip to @Foritus!
@fluidlogic @Foritus when 64-bit computers were new (around 2005) I worked for a company that used QuickBooks, and got to talk Intuit tech support through how to install the QuickBooks server on a 64-bit Linux host
-
@argv_minus_one @flargh @fluidlogic @Foritus TIL that there’s a worse option than 8088-style “segmented” memory addressing
Nah, at least 68000 pointers are flat and unambiguous. 8086 far pointers are just as long (32 bits) and can't even be compared for equality doing a bunch of math first. And 8086 near pointers can't be compared at all unless you know which segment they both point into.
-
#retrocomputing folks: I'm trying to get a sense of the proportion of people here who are into retrocomputing today but didn't experience certain classes of machines when they first came on the market. I want everyone's input! Please boost!
These questions are specifically about all 32-bit comsumer machines: PCs, Macs, RISC machines, you name it. Mid 1980s to 2005-ish.
(I'm also concurrently running a poll for 8 and 16-bit home computers and another for minicomputers. I'll probably have to do another for UNIX Workstations and LISP machines!)
Hat tip to @Foritus!
@Foritus @fluidlogic wait, when did 32bit machines' heyday end?
-
Little did I know that Windows 3.1 very definitely does use 32-bit addressing! Apps run in real mode and still use the 8086's weird segmented addressing, but the Windows kernel runs in 32-bit mode and maps memory in and out of that space, somewhat like swap on a modern operating system (except triggered by GlobalLock calls instead of page faults).
@argv_minus_one @fluidlogic @Foritus Windows 3.1 is weirder than that – in Enhanced mode, it's really a hypervisor running at least one virtual machine (which is running Standard mode Windows 3.1), the other virtual machines are DOS boxes if you're running any.
-
#retrocomputing folks: I'm trying to get a sense of the proportion of people here who are into retrocomputing today but didn't experience certain classes of machines when they first came on the market. I want everyone's input! Please boost!
These questions are specifically about all 32-bit comsumer machines: PCs, Macs, RISC machines, you name it. Mid 1980s to 2005-ish.
(I'm also concurrently running a poll for 8 and 16-bit home computers and another for minicomputers. I'll probably have to do another for UNIX Workstations and LISP machines!)
Hat tip to @Foritus!
@fluidlogic @Foritus What about 16- and 36-bit computers?
-
R ActivityRelay shared this topic
