Most of the problem is that the CPU is involved in almost all bus transactions. Sound stuttering was common from CD-ROM when doing other things like HDD access because the CPU would have to halt what it's doing, and process audio from the IDE/SCSI Channel to send it through the processor just to go the the audio chipset. Causing almost all it's doing to stop. Which was partly the reason why just playing a CD would cause the cursor in a text program (appleworks/etc…) to crawl and/or lag. It was a half-assed design to save cost on chips by almost using the CPU as the main chipset, as well as the CPU. These days, you have a north/south bridge or UniNorth (combined chipset. the nForce 630 on a PC board is an example where it has the north, south and a GPU all on the same chip. I have one of these boards and it works fine as they segregate the die of the chipset into 3 areas to do the same thing and route all the lines from the board into it)
Anyways, they released an updated board and OS 9 would not load unless you had it. It essentially caused the timings of certain components to take a backseat so certain high-data transfering could occur without waiting for other things, while it upped the priority of things like the ADB Bus (so typing wasn't affected).
This was the reason why the PM 5xxx/6xxx checker app was included on 8.5/9 discs as it checked to see if the board was RMA'd and replaced with a modified.
it wasn't technically a 68k board as much as it was a re-use of the chips from the 68k era, combined with money saving techniques to reduce the chip, and put as much on the processor as well. It was almost an indication that apple just wanted to reduce it's stock from the cross-over to the PPC world, while making people think they got the next gen, and it ended up ticking people off. (I seem to recall someone on the boards saying their friend bought one of these and took it back to the store and got a different machine because whenever they typed the CD ROM audio would stutter/and or the typing would be dog-slow)
It was a road-apple for this very reason. Now keep in mind, the 5400/6400/5500/6500 didn't have these design issues as they put a proper chipset. Also I seem to recalled the 5200/5300/6200/6300 all had either a PDS or a Nubus slot as the old chipsets had those in mind. Any time you see one of these machines with a PCI slot, you are almost guaranteed a better experience as the PCI required a full PPC chipset that took care of all the bus transactions.
I find it interesting that even the latest G4 didn't utilize bus-mastering on PCI, only the G5 seemed to from what I recall. This allowed certain things to occur on the bus without requiring an IRQ interrupting the processor. An example would be a TV Tuner calling directly to a PCI video card to output video or encode it without requiring the CPU and/or northbridge to process the data and re-direct it to the video slot. While cool in itself, it was very rarely utilized in PCI, but the parallel PCI bus made it possible (Impossible with PCI-E as those are tired directly to the Northbridge or processor directly and those handled it, but no PCI-E slot can see any other directly).
DMA didn't seem to come into play until the G4 series either. Which would reduce drastically the amount of CPU time involved in processing something as simple as access to an IDE/ATA Harddrive. Why apple waited so long (the Pentium Pro chipsets already included this as well as some later ISA chipsets) leaves me guessing, as DMA on the PCI bus would've made macs much faster, Especially with PCI capture cards and the like. Multi-processing while doing video capture anyone?
Anyways, they released an updated board and OS 9 would not load unless you had it. It essentially caused the timings of certain components to take a backseat so certain high-data transfering could occur without waiting for other things, while it upped the priority of things like the ADB Bus (so typing wasn't affected).
This was the reason why the PM 5xxx/6xxx checker app was included on 8.5/9 discs as it checked to see if the board was RMA'd and replaced with a modified.
it wasn't technically a 68k board as much as it was a re-use of the chips from the 68k era, combined with money saving techniques to reduce the chip, and put as much on the processor as well. It was almost an indication that apple just wanted to reduce it's stock from the cross-over to the PPC world, while making people think they got the next gen, and it ended up ticking people off. (I seem to recall someone on the boards saying their friend bought one of these and took it back to the store and got a different machine because whenever they typed the CD ROM audio would stutter/and or the typing would be dog-slow)
It was a road-apple for this very reason. Now keep in mind, the 5400/6400/5500/6500 didn't have these design issues as they put a proper chipset. Also I seem to recalled the 5200/5300/6200/6300 all had either a PDS or a Nubus slot as the old chipsets had those in mind. Any time you see one of these machines with a PCI slot, you are almost guaranteed a better experience as the PCI required a full PPC chipset that took care of all the bus transactions.
I find it interesting that even the latest G4 didn't utilize bus-mastering on PCI, only the G5 seemed to from what I recall. This allowed certain things to occur on the bus without requiring an IRQ interrupting the processor. An example would be a TV Tuner calling directly to a PCI video card to output video or encode it without requiring the CPU and/or northbridge to process the data and re-direct it to the video slot. While cool in itself, it was very rarely utilized in PCI, but the parallel PCI bus made it possible (Impossible with PCI-E as those are tired directly to the Northbridge or processor directly and those handled it, but no PCI-E slot can see any other directly).
DMA didn't seem to come into play until the G4 series either. Which would reduce drastically the amount of CPU time involved in processing something as simple as access to an IDE/ATA Harddrive. Why apple waited so long (the Pentium Pro chipsets already included this as well as some later ISA chipsets) leaves me guessing, as DMA on the PCI bus would've made macs much faster, Especially with PCI capture cards and the like. Multi-processing while doing video capture anyone?