I have to say it's the shoes.. It's gotta be the shoes..
No it's do to the fact ATA/66 is cheaper and more widely available. There are not a lot of ATA 133 drive yet are there? Maybe Apple wants to skip ATA/100.
Maybe they are waiting for DDR333 to become standard as well.
my guess is that there are several factors in play here. first, since apple offers U160 drives for those who actually need very high drive performance, there's less incentive to offer slightly faster ATA solutions. BTW, for anyone who thinks ATA133 is way faster, read some of the good hardware sights. At this point the limiting factor in the ATA world is more often the drive itself (spindle speed, small cache, poor ATA chipset, etc).
So, apple, in trying to keep a relatively high level of performance, yet keep the price down, has to draw the line somewhere. ATA/66 for the 90% majority that don't really need anything faster yet, and U160 SCSI for true pros.
oh yeah, other tidbits:
there are plenty of companies that offer ATA/100 and ATA/133 PCI cards if you really want one. in fact, most also offer some type of RAID card too.
apple still uses 5400 RPM drives in some machines (yes, the NEW iMac). iBooks use 4200 RPM drives. only with the brand new PMs did apple switch to all 7200 RPM drives in that line.
(i'm not complaining, just pointing out that the 'why don't they use ATA/133 is kinda silly)
As I understand it, the only concrete advantage to going with ATA/100 or /133 is that they accomodate larger HD sizes - ATA/66 can't handle the new 160GB monsters that are rolling off the lines now. But then, Apple isn't shipping any of them in its machines.
As for why Apple hasn't adopted it: They have all their I/O consolidated into one big ASIC now, which greatly reduces their manufacturing costs, and allows for some cool optimizations (like FireWire, USB and Ethernet all getting dedicated bandwidth instead of clogging up the PCI bus). On the other hand, updating any one part means updating the whole ASIC, and Apple might well have decided that it's not worth doing until there are enough reasons to.
There will be soon. Apple is shipping 80GB drives, which are getting perilously close to the limit of what ATA/66 can handle, and the ink is drying on IEEE 1394b - just to name two technologies - and then there's Apple's eventual adoption of RapidIO and/or HyperTransport, which will also impact the design of the ASIC.
<strong>The only concrete advantage to going with ATA/100 or /133 is that they accomodate larger HD sizes - ATA/66 can't handle the new 160GB monsters that are rolling off the lines now. But then, Apple isn't shipping any of them in its machines.
As for why Apple hasn't adopted it: They have all their I/O consolidated into one big ASIC now, which greatly reduces their manufacturing costs, and allows for some cool optimizations (like FireWire, USB and Ethernet all getting dedicated bandwidth instead of clogging up the PCI bus). On the other hand, updating any one part means updating the whole ASIC, and Apple might well have decided that it's not worth doing until there are enough reasons to.
There will be soon. Apple is shipping 80GB drives, which are getting perilously close to the limit of what ATA/66 can handle, and the ink is drying on IEEE 1394b - just to name two technologies - and then there's Apple's eventual adoption of RapidIO and/or HyperTransport, which will also impact the design of the ASIC.
Exactly. Add to that DDR support (for whichever processor supports it), larger addressable RAM (rumoured up to 16 Gb), USB2, and whatever other goodies they're cooking up.
Actually the ATA controller is NOT on the big mofo of a northbridge Apple's come up with. The ATA controller is a separate chip.
The ONLY way to exceed the bandwidth of ATA-66 is to have 2 7200rev hard drives on the same channel executing a burst transfer at the same time. Until 10000rev ATA drives, you don't need more.
Although I wish IBM, WD, Maxtor etc. would get around to making some nice 7200rev copper FireWire drives and 10000rev fiber FireWire drives already...
Exactly. Add to that DDR support (for whichever processor supports it), larger addressable RAM (rumoured up to 16 Gb), USB2, and whatever other goodies they're cooking up.</strong><hr></blockquote>
sorry, but i have to nitpick (not trying to annoy you Programmer).
16Gb is 2GB. The older G4 could handle that much RAM. it may seem like a small thing, but there's a huge difference between a bit and a byte.
and just for reference, ATA/66 & ATA/100 both allow for volues up to 137GB, and ATA/133 (once the standard is finalized) will allow for up to 144PB (yes, 144 petabytes!)
<strong>As I understand it, the only concrete advantage to going with ATA/100 or /133 is that they accomodate larger HD sizes - ATA/66 can't handle the new 160GB monsters that are rolling off the lines now.
</strong><hr></blockquote>
Actually, neither can ATA100. 48-bit-addressing is only implemented in ATA133. Otherwise, you're completely right. The one and only other situation where UDMA-66 *might* be a bottleneck is when you have two harddisks attached to the same controller, and you access both disks concurrently.
[quote]<strong>
As for why Apple hasn't adopted it: They have all their I/O consolidated into one big ASIC now, which greatly reduces their manufacturing costs, and allows for some cool optimizations (like FireWire, USB and Ethernet all getting dedicated bandwidth instead of clogging up the PCI bus).
</strong><hr></blockquote>
This is only half true. The iBook and iMac do have a single ASIC for all their I/O (which is called Pangea). The PowerMac and PowerBook on the other hand have a more traditional "North and South"-combo. UniNorth contains the processor and memory interface, the ethernet and firewire controller, and the PCI controller. KeyLargo contains the remaining I/O (ATA, USB, Audio) and is connected to UniNorth via PCI (64 bit wide for the PowerMac, 32 for the PowerBook). The CardBus controller in the PowerBook is an additional PCI device of its own.
Note that this is not really a bad thing - the PCI bus connecting UniNorth and KeyLargo is just as fast as the connection between north- and southbridge on Intel, VIA and IIRC AMD chipsets, and those also have to run their NIC and FW traffic through it (which also happens to require the most bandwidth).
[quote]<strong>
On the other hand, updating any one part means updating the whole ASIC, and Apple might well have decided that it's not worth doing until there are enough reasons to.
</strong><hr></blockquote>
That's exactly the point, IMO. They'd have to do a new revision of the KeyLargo in order to support more than ATA66 on the PowerMacs/-Books. Since there's nothing else on the KeyLargo that needs to be upgraded, and especially since ATA66 is not a bottleneck for nearly all current ATA disks, it might just not be worth it in Apple's opinion.
[quote]<strong>There will be soon. Apple is shipping 80GB drives, which are getting perilously close to the limit of what ATA/66 can handle, and the ink is drying on IEEE 1394b - just to name two technologies - and then there's Apple's eventual adoption of RapidIO and/or HyperTransport, which will also impact the design of the ASIC.
</strong><hr></blockquote>
Yes, I believe there has to be a new northbridge in the not too distant future. Given the peak bandwidth of ATA133, especially in comparison to 100Mbit/s Ethernet, it seems quite likely that this might get another direct connection to the northbridge too. The same is true for USB2, in case Apple decide to support it.
sorry, but i have to nitpick (not trying to annoy you Programmer).
16Gb is 2GB. The older G4 could handle that much RAM. it may seem like a small thing, but there's a huge difference between a bit and a byte.
</strong><hr></blockquote>
Okay okay, I meant gigabytes. Nobody ever talks about memory capacity in bits anyhow -- they are generally reserved for serial communications where data is sent one bit at a time. Now that MacOS X is here Apple will undoubtably provide the capability to have more than a paltry 4 gigabytes in their machines, and when they have a 64-bit processor a single process will be able to address all of it.
Ask a simple question and get a FANTASTICLY informative thread!
With all the clamor over a potentially faster processor it seems that discussion of the overall plumbing of the PowerMacs has been overlooked. I look forward to seeing how the next true generation of these machines will respond to the new standards which loom large in the future of desktop computing. SCSI is a tenacious technology, but as ATA is a "for the masses" alternative, I'd suspect that Apple will bump up the speed within a year.
To sum up, there is no such thing as an existing ATA drive which can keep an ATA/66 bus fully saturated, let alone an ATA/100 bus, let alone an ATA/133 bus...I've yet to see *any* ATA drive that can exceed 40MB/s with any consistency for everyday tasks. Hence, why not use ATA/66 and keep the costs down until a much better drive exists that can utilize the extra bandwidth of faste ATA buses?
-ATA 133 controllers likely cost very little more than ATA 66. If the PC cheapo boards can afford it, so can Apple.
If something new can be had that obviously only brings benefits, we should have it. Otherwise you could say a Mac Plus is fast enough for typing, so why get a G4?
The real reason is that Apple didn't want to spend money on a Motherboard rev, if the next rev in the pipeline has all the fancy new tech anyway.
<strong>The reasons have been listed, and they are:
-peak rates CAN exceed 66MB/sec
-master/slave combos CAN exceed 66MB/sec
</strong><hr></blockquote>
Show me a link to an ATA drive that can exceed 66MB/s, even for short durations - I'd be interested to see it.
Master Slave Combo? You mean a RAID configuration? Having to separate Master and Slave drives has no bearing on performance unless you set them up for RAID...
Show me a link to an ATA drive that can exceed 66MB/s, even for short durations - I'd be interested to see it.
</strong><hr></blockquote>
Any drive that has a cache can exceed 66MB/s.
Of course, that's only for some fractions of a seconds, but well, he was talking about short durations anyway.
[quote]<strong>
Master Slave Combo? You mean a RAID configuration? Having to separate Master and Slave drives has no bearing on performance unless you set them up for RAID...</strong><hr></blockquote>
Even without a RAID setup, there can be an impact on performance when you access both drives concurrently. How often this actually happens in real life use is another matter, of course.
Comments
perhaaps it part of their G5 boards which keep getting delayed
perhaps they don't think anyone needs more than ATA/66
perhaps they are just plain cheap
who knows but its ridiculous
No it's do to the fact ATA/66 is cheaper and more widely available. There are not a lot of ATA 133 drive yet are there? Maybe Apple wants to skip ATA/100.
Maybe they are waiting for DDR333 to become standard as well.
Not sure
<img src="graemlins/oyvey.gif" border="0" alt="[No]" /> <img src="graemlins/oyvey.gif" border="0" alt="[No]" /> <img src="graemlins/oyvey.gif" border="0" alt="[No]" />
So, apple, in trying to keep a relatively high level of performance, yet keep the price down, has to draw the line somewhere. ATA/66 for the 90% majority that don't really need anything faster yet, and U160 SCSI for true pros.
oh yeah, other tidbits:
there are plenty of companies that offer ATA/100 and ATA/133 PCI cards if you really want one. in fact, most also offer some type of RAID card too.
apple still uses 5400 RPM drives in some machines (yes, the NEW iMac). iBooks use 4200 RPM drives. only with the brand new PMs did apple switch to all 7200 RPM drives in that line.
(i'm not complaining, just pointing out that the 'why don't they use ATA/133 is kinda silly)
[ 01-30-2002: Message edited by: concentricity ]</p>
As for why Apple hasn't adopted it: They have all their I/O consolidated into one big ASIC now, which greatly reduces their manufacturing costs, and allows for some cool optimizations (like FireWire, USB and Ethernet all getting dedicated bandwidth instead of clogging up the PCI bus). On the other hand, updating any one part means updating the whole ASIC, and Apple might well have decided that it's not worth doing until there are enough reasons to.
There will be soon. Apple is shipping 80GB drives, which are getting perilously close to the limit of what ATA/66 can handle, and the ink is drying on IEEE 1394b - just to name two technologies - and then there's Apple's eventual adoption of RapidIO and/or HyperTransport, which will also impact the design of the ASIC.
[ 01-30-2002: Message edited by: Amorph ]
[ 01-30-2002: Message edited by: Amorph ]</p>
<strong>The only concrete advantage to going with ATA/100 or /133 is that they accomodate larger HD sizes - ATA/66 can't handle the new 160GB monsters that are rolling off the lines now. But then, Apple isn't shipping any of them in its machines.
As for why Apple hasn't adopted it: They have all their I/O consolidated into one big ASIC now, which greatly reduces their manufacturing costs, and allows for some cool optimizations (like FireWire, USB and Ethernet all getting dedicated bandwidth instead of clogging up the PCI bus). On the other hand, updating any one part means updating the whole ASIC, and Apple might well have decided that it's not worth doing until there are enough reasons to.
There will be soon. Apple is shipping 80GB drives, which are getting perilously close to the limit of what ATA/66 can handle, and the ink is drying on IEEE 1394b - just to name two technologies - and then there's Apple's eventual adoption of RapidIO and/or HyperTransport, which will also impact the design of the ASIC.
[ 01-30-2002: Message edited by: Amorph ]</strong><hr></blockquote>
Exactly. Add to that DDR support (for whichever processor supports it), larger addressable RAM (rumoured up to 16 Gb), USB2, and whatever other goodies they're cooking up.
The ONLY way to exceed the bandwidth of ATA-66 is to have 2 7200rev hard drives on the same channel executing a burst transfer at the same time. Until 10000rev ATA drives, you don't need more.
Although I wish IBM, WD, Maxtor etc. would get around to making some nice 7200rev copper FireWire drives and 10000rev fiber FireWire drives already...
Barto
<strong>
Exactly. Add to that DDR support (for whichever processor supports it), larger addressable RAM (rumoured up to 16 Gb), USB2, and whatever other goodies they're cooking up.</strong><hr></blockquote>
sorry, but i have to nitpick (not trying to annoy you Programmer).
16Gb is 2GB. The older G4 could handle that much RAM. it may seem like a small thing, but there's a huge difference between a bit and a byte.
and just for reference, ATA/66 & ATA/100 both allow for volues up to 137GB, and ATA/133 (once the standard is finalized) will allow for up to 144PB (yes, 144 petabytes!)
I'm surprised that the ATA controller is by itself. I'll have to go back and find that motherboard diagram: I could have sworn it wasn't.
...which I should have done before posting. Ah well.
<strong>As I understand it, the only concrete advantage to going with ATA/100 or /133 is that they accomodate larger HD sizes - ATA/66 can't handle the new 160GB monsters that are rolling off the lines now.
</strong><hr></blockquote>
Actually, neither can ATA100. 48-bit-addressing is only implemented in ATA133. Otherwise, you're completely right. The one and only other situation where UDMA-66 *might* be a bottleneck is when you have two harddisks attached to the same controller, and you access both disks concurrently.
[quote]<strong>
As for why Apple hasn't adopted it: They have all their I/O consolidated into one big ASIC now, which greatly reduces their manufacturing costs, and allows for some cool optimizations (like FireWire, USB and Ethernet all getting dedicated bandwidth instead of clogging up the PCI bus).
</strong><hr></blockquote>
This is only half true. The iBook and iMac do have a single ASIC for all their I/O (which is called Pangea). The PowerMac and PowerBook on the other hand have a more traditional "North and South"-combo. UniNorth contains the processor and memory interface, the ethernet and firewire controller, and the PCI controller. KeyLargo contains the remaining I/O (ATA, USB, Audio) and is connected to UniNorth via PCI (64 bit wide for the PowerMac, 32 for the PowerBook). The CardBus controller in the PowerBook is an additional PCI device of its own.
Note that this is not really a bad thing - the PCI bus connecting UniNorth and KeyLargo is just as fast as the connection between north- and southbridge on Intel, VIA and IIRC AMD chipsets, and those also have to run their NIC and FW traffic through it (which also happens to require the most bandwidth).
[quote]<strong>
On the other hand, updating any one part means updating the whole ASIC, and Apple might well have decided that it's not worth doing until there are enough reasons to.
</strong><hr></blockquote>
That's exactly the point, IMO. They'd have to do a new revision of the KeyLargo in order to support more than ATA66 on the PowerMacs/-Books. Since there's nothing else on the KeyLargo that needs to be upgraded, and especially since ATA66 is not a bottleneck for nearly all current ATA disks, it might just not be worth it in Apple's opinion.
[quote]<strong>There will be soon. Apple is shipping 80GB drives, which are getting perilously close to the limit of what ATA/66 can handle, and the ink is drying on IEEE 1394b - just to name two technologies - and then there's Apple's eventual adoption of RapidIO and/or HyperTransport, which will also impact the design of the ASIC.
</strong><hr></blockquote>
Yes, I believe there has to be a new northbridge in the not too distant future. Given the peak bandwidth of ATA133, especially in comparison to 100Mbit/s Ethernet, it seems quite likely that this might get another direct connection to the northbridge too. The same is true for USB2, in case Apple decide to support it.
Bye,
RazzFazz
[ 01-30-2002: Message edited by: RazzFazz ]</p>
<strong>The G4 has been able to address 36bits of memory(64GB) since last year with the inception of the the PPC7450.</strong><hr></blockquote>
True, but what use is that without a memory controller that actually supports that too?
(UniNorth and Pangea certainly don't.)
Bye,
RazzFazz
<strong>
sorry, but i have to nitpick (not trying to annoy you Programmer).
16Gb is 2GB. The older G4 could handle that much RAM. it may seem like a small thing, but there's a huge difference between a bit and a byte.
</strong><hr></blockquote>
Okay okay, I meant gigabytes. Nobody ever talks about memory capacity in bits anyhow -- they are generally reserved for serial communications where data is sent one bit at a time. Now that MacOS X is here Apple will undoubtably provide the capability to have more than a paltry 4 gigabytes in their machines, and when they have a 64-bit processor a single process will be able to address all of it.
Ask a simple question and get a FANTASTICLY informative thread!
With all the clamor over a potentially faster processor it seems that discussion of the overall plumbing of the PowerMacs has been overlooked. I look forward to seeing how the next true generation of these machines will respond to the new standards which loom large in the future of desktop computing. SCSI is a tenacious technology, but as ATA is a "for the masses" alternative, I'd suspect that Apple will bump up the speed within a year.
D
-peak rates CAN exceed 66MB/sec
-master/slave combos CAN exceed 66MB/sec
-160GB drives ARE shipping
-ATA 133 controllers likely cost very little more than ATA 66. If the PC cheapo boards can afford it, so can Apple.
If something new can be had that obviously only brings benefits, we should have it. Otherwise you could say a Mac Plus is fast enough for typing, so why get a G4?
The real reason is that Apple didn't want to spend money on a Motherboard rev, if the next rev in the pipeline has all the fancy new tech anyway.
Expect that for MWNY.
G-News
<strong>The reasons have been listed, and they are:
-peak rates CAN exceed 66MB/sec
-master/slave combos CAN exceed 66MB/sec
</strong><hr></blockquote>
Show me a link to an ATA drive that can exceed 66MB/s, even for short durations - I'd be interested to see it.
Master Slave Combo? You mean a RAID configuration? Having to separate Master and Slave drives has no bearing on performance unless you set them up for RAID...
<strong>
Show me a link to an ATA drive that can exceed 66MB/s, even for short durations - I'd be interested to see it.
</strong><hr></blockquote>
Any drive that has a cache can exceed 66MB/s.
Of course, that's only for some fractions of a seconds, but well, he was talking about short durations anyway.
[quote]<strong>
Master Slave Combo? You mean a RAID configuration? Having to separate Master and Slave drives has no bearing on performance unless you set them up for RAID...</strong><hr></blockquote>
Even without a RAID setup, there can be an impact on performance when you access both drives concurrently. How often this actually happens in real life use is another matter, of course.
Bye,
RazzFazz