The ATA/66 or ATA/100 bus is already moving data as fast as it can with an ATA/100 drive attached.
A SATA to IDE adapted drive can only move data as fast as the bus can.
So, a SATA to IDE adapted SATA 6.0 Gbps drive on an ATA/66 bus only gets about 50 MBps (an average max ATA/66 data rate of 66 MBps minus system overhead) or .4 Gbps (about .7 Gbps on ATA/100).
If you want SATA 1.5 Gbps speed, get a PCI controller.
Otherwise, all the adapter allows is the ability to use larger, more readily available drives; SATA drives.
OK, then there is something I completely misunderstand from the beginning. I thought throughput (transfer rates of data) is different from acces time. I thought, if I have a faster rpm drive (or SSD) compared to a slower rpm drive, the accesss time gets faster (i.e. opening folders, applications will be visual more imediately).
I did connect a sata drive to my internal Sonnet SATA-PCI controller and now the folders open quicker.
Though has nothing to do with my above question I'd like to add another observation. I have to say, when I copy like 500GB from one drive to another and count the time, I get about FW400 speeds (I had between 35-38MB/s, from my calculation: 500GB divided by time needed to completion).
Another observation: I never heard someone complain, but I have to be very carefull with the cables. Because, fi I close the door, the power cord on the mainboard blocks. It touches the upper drive of two drives. I find it interesting that no one ever mentioned that that (and I have read some threads on SATA-PCI-Cards in other forums).
I thought throughput (transfer rates of data) is different from acces time. I thought, if I have a faster rpm drive (or SSD) compared to a slower rpm drive, the accesss time gets faster (i.e. opening folders, applications will be visual more imediately).
Access time would be seek time; movement of the head to the data location on the platter.
All are fast, measurable but imperceptible.
After that, the data must be read, moved to RAM or other cache, processed by the processor, returned to cache and used.
In between, you have the ATA bus, PCI bus, USB bus or firewire bus. Then, there is the memory bus.
All the data travels through all of this, with the weakest link being the slowest device/bus/process in the chain.
Here are some tests that I did of the above 4 bus data rates:
In the case of the GE G4, the ATA/66 bus is a choke point.
SSD are the fastest (on a PCI controller) but may be considered wasted in a G4.
My test results:
I have to say, when I copy like 500GB from one drive to another and count the time, I get about FW400 speeds
As per the above explanation, there is more going on that a simple read.
I never heard someone complain, but I have to be very carefull with the cables.
The case interiors are very compact. Complaints are a mute point.
Fill all 4 PC/AGPI slots, stuff in 4 drives and a fan or two and you have something that you don't want to open and close much.
are you copying from two drives on the same cable, or are the drives on separate channels on separate cables?
I think if you are copying between Master/Slave drives on the same cable, the speed will not be as fast since read and write opeartions are sharing the channel. And from one channel to another channel speed is limited to the slower channel speed.
thanks. I remember that I stumbled upon that site of your's some time ago, long before I subscribed here. Great website (used the graphics card list for OS 9 and X a lot to compare cards).
Heat wise: did you try, if one gets better cooling keeping the drive cage out? Because I see, that there are holes in the bottom of the case, that are covered, if one uses the drive cage of the PowerMac. Also, I tried quietening my G4 and changing the PSU fan with a newer one, with the same airflow rating (from noctua, but rated 17dB(A) vs 31dB(A)) did a big improvement. When I replaced the case fan with a Xigmatek fan, that was said to have 25dB(A) contra 37dB(A) of the stock G4 case fan, I had the feeling that at full speed the stock fan was quieter.
I finally somehow got to the hypothesis, that lowering the noise of the case fan, by keeping the max airflow rate the same, is impossible (mostly because there are only a few fans matching the criteria of same airflow and lower noise). So I think one would have to use a less powerful case fan and an additional 80mm fan on the bracket where the airport card can be installed. (don't want to go that far, as you did
I took the fan data from here http://infohost.nmt.edu/~holstien/g4info.html
I know one can't just take the numbers and go ahead, because every manufacturer has its own testing conditions, but as a very vague guideline, I thought it is ok.
@ Glen Doggett:
I have the Velociraptor and a Raptor as a boot drive sitting on the IDE master/slave cable on board via SATA-IDE-adapter.
I was speaking about two 1TB drives, both having a SATA cable that is directly connected to the Sonnet SATA-PCI-card. Moving 500GB from one drive to the other got 35MB/s.
@ OM 617.952:
Not quiet right...
PowerMac G4: 2000
Velociraptor: 2009 (faster than IDE WD 120GB and 80GB from 2003-2004)
WD 120GB IDE: 2003 (slowest)
Hitachi 1TB 7k1000.B: 2009 (same or bit slower than Velociraptor)
In fact I copied 20GB from a Velociraptor 300GB GLFS to HLFS (2008 and 2009) and it was faster than from a Samsung F1 to another F1 and faster than from a Hitachi 7K1000 (...SLA360) to a Hitachi 7K1000 (KLA330). All on the Sonnet SATA-PCI Card, from SATA-port1 to SATA-port2.
What do you mean: SSD on IDE or SATA?
So, why then recommend people so often "put a SSD in your ibook", if all the above counts and seek time doesn't matter since ATA-66 should be the limit here, too?
By the way are we talking about SATA-SSD or IDE SSD?
I heard that IDE SSD is slow. Also 1,8" SATA drives are said to be slower than 2,5" mechanical drives.
Anyway for the PowerMac, I would assume after what Japamac has said, that an SSD would not be faster on the ATA-66 Bus than all other drives. SATA HDD (which includes the Velociraptor) on SATA-PCI would be faster and SSD on SATA-PCI should be even faster. Because the ceiling of the SATA-Bus would be higher. But that goes for transfer rate, I wlways thought.
Why are people always talking about boot time and the time Applications and folders take to open and that this would be faster with a Velociraptor or SSD? Do the before mentioned not fall under seek time? Ah, ok I understand. Differences in Seek time (head moving, 5400rpm vs 7200rpm vs 10k rpm) isn't visible to the Eye. What we think as faster access (seek) is actually the data transfer rate. So is it a myth for PC people that want fast head movement or SSDs for Gaming to shorten times?
The Thing I always thought about that is, that once the System is fully booted or the Applications are open, you don't notice a difference in respinsiveness. But opening a folder (and showing its content) should not be that big data, so ATA-66 should open a folder quickly (aren't this like KB big sizes?). Though from practise, I noticed folders open faster over SATA-PCI.
Still, I get about 35-40MB/s moving big files over the SATA-PCI Bus from one 1TB to another 1TB and the files are on the beginning of the platter.
From ATA-66 to ATA-66 30-35MB/s. I wonder why you (japamac) get near 100MB/s over SATA-PCI.
Given the luck I normally have with trying things, I bet, if I would try a SSD (not that I would even conmsider this) I would get only the next 5MB/s more over the others.
lime-iMacG3, you completely missed the point that the Velociraptor is junk.
Its seek time is 4ms. Even the cheapest SSD today has a seek time of 0.1ms or less and will very easily saturate an ATA/100 bus. Thats 40x faster seek speed.
Seek time is completly independent of the bus speed.
Why are people always talking about boot time and the time Applications and folders take to open and that this would be faster with a Velociraptor or SSD?
As said before, the Velociraptor is an obsolete antique. It was made for a time before SSD drives became consumer affordable. SSD drives are better in every way except for purchase price.
What we think as faster access (seek) is actually the data transfer rate.
Seek time is exactly what its called, time it takes to access two different files. It has nothing to do with transfer rate, its a mechanical limit of the drive. An SSD has no moving parts so there is no delay accessing them and file fragmentation has zero impact on drive performance.
if I would try a SSD (not that I would even consider this)
And that is why you're stuck in the past with slow antiques.
<Edited By Host>
Now that you have deleted your bad talking about Japamac, I'd like to answer
It's a bit hard to call it junk and obsolete antiques, I find. Also, can't you imagine, that there might be software, that runs on PowerPC, that I want to stay with and since I don't need anything faster and do not have a facebook account or watch youtube, I don't see the need in investing in new stuff.
The Velociraptor was not bought new (buying new would have been totally stupid given price-performance-relation, they were allready overpriced, when there was no SSDs).
What I was aiming was to just try it and playing arround with the hardware I got. Since I got the Velociraptor cheap, why not try.
My assumption was not getting a new class lightning fast machine, but to see a visible difference in the pure action of a folder poping open. I do see no difference to IDE.
But why, do I see folders opening faster using the SATA-PCI-Card and a SATA drive, which is not a SSD? My assumption: the SATA-to-IDE adaptor on the ATA-66 BUS, must have problems handing the information though.
Mind = be aware! differentiate between:
a) SATA-Drive (Velociraptor or 7200rpm) on SATA-to-IDE bridge, which is connected to the onboard controller
b) SATA-Drive connected via SATA-PCI-Card to Mainboard.
And why did I see a differenc ein my ibook G4:
a) Toshiba 40GB 5400rpm from 2005, stock drive
b) WD 320GB 5400rpm drive from 2009
b) was faster.
My explanation (and excuse for you) SuperDuper does rudimentary defragmentation, while copying. So the copied install from my old drive was faster on the new, due to defragmentation. It might have been more defragmentated than MAC OS X is said to do with every startup.
I do not consider SSDs pricewise and because of longevity. Yes, now you might say: they are not mechanical, how can they break? HDDs have head crashes, motor defects, etc.
Well, I have dirves from my oldest Macs, that are 13 years old and they live on. I had no defective HDD so far, that got defective. Only on the software level and that can happen with flash Drives, too.
I fear flash Dirves, because I have bad experiences with USB-sticks (dongles you migth call them in english).
The best SSDs can are seek time and read transfer rates (being only slightly better in write transfers to HDDs, but still better in access time). The most that harms them is reading data, I often hear. This might have changed with current generations. But the bad feeling remains. Also, I hear a lot of Firmware issues or features not supported under Mac, which cause conflicts and problems.
Again, my aim was not to turn my G4 into a lightning fast machine, I was just trying and observing, if better access times would be visible. And they are on SATA-PCI, but not on ATA-66.
Apart from that access time discussion, there must be something wrong, still. Though I get folders opening faster on SATA-PCI, the transfer rates of a 2009 1TB SATA drive, with the data sitting at the beginning of the platter, is still 35MB/s. This can't be ok. Since others have faster transfer rates via SATA in their G4s. (or they are just telling what a virtual benchmark tells, I did stop the time, with a stop watch.)
Any of them. Since data rate is limited by the ATA100 bus, it doesn't matter which one you get as any of them will easily saturate the bus and respond as fast as the ATA controller can issue commands.
You can either connect it exactly as the VR with an adaptor or you can get a bootable PCI SATA card to take advantage of the drive's speed.
There is nothing special about how SSD drives work. They merely lack the pyscial heads and rotating media that restricts HD drive performance.
Transfer rate from an avaerage SSD is 600Mb/s (approx), the internal bus will limit that to 66Mb/s and a pluging PCIe card to 124Mb/s. At that rate they will be no faster than an HDD. Why suggest the expense of an SSD when the performance will be better than an HDD (which will be much larger for the same price)