-
Hello MLAers! We've re-enabled auto-approval for accounts. If you are still waiting on account approval, please check this thread for more information.
Powerbook Duo SSD
- Thread starter theos911
- Start date
I'll wager that their OS "requirements" have more to do with their support of a limited set of Low Level Formatting Tools than the Drives or the specific OS versions themselves.
NOBODY seems to want to support Mac OS 9.2.2 or below, even Sonnet is beginning to cut the cord!
Makes sense business-wise but it's annoying just the same! :-/
NOBODY seems to want to support Mac OS 9.2.2 or below, even Sonnet is beginning to cut the cord!
Makes sense business-wise but it's annoying just the same! :-/
Cheap IDE-CF adapter plus CF card would be much cheaper....
I would think a SSD would be better at handling VM. The macbook air with SSD has been out for a while and there aren't many reports about them going bad and VM is always on with OSX. Two times ram for VM on a different Duo vs. an Air would be very different, 36x2=72mb or 2gbx2=4096mb. Swapping 4096gb would be worse than 72mb I think.
On a different point. I was thinking of doing something similar recently and found SSD drives and DOMs (they're more similar than different) from Kingspec all over Ebay. Might be a cheaper route and they're also IDE.
Either way, if you pull this off I would really like to hear the difference in performance.
On a different point. I was thinking of doing something similar recently and found SSD drives and DOMs (they're more similar than different) from Kingspec all over Ebay. Might be a cheaper route and they're also IDE.
Either way, if you pull this off I would really like to hear the difference in performance.
There are huge differences in speeds and numbers of program/erase (PE) cycles between CF cards. Speeds are well covered lots of different places, PE cycles tend to vary between 10,000 for the lower end cards to more than 1million at the high end. Check the rating of the specific card you're looking at.
Both SSD and to some extent CF manufacturers seem to avoid mentioning PE cycles directly, advertising an MTBF number instead. This is kind of understandable because the SSD/CF present a block level interface to the controller, with the card/drive managing the mapping of logical block to NAND cell, with wear leveling thrown in the middle. So you don't necessarily know how many NAND cells there are (there may be significantly more than the capacity presented to the controller, which will be used for wear leveling) which makes it a little difficult to extract meaning from a PE cycle count. Then there's MLC vs. SLC in SSD, with SLC having significantly more write cycles, and MLC having comparable write cycles to most CF cards.
In addition to all that, the wear leveling implementation can make a noticeable difference in usability as well. For example, with the cheaper (I particularly notice it with 133x and lower) CF cards will get reasonably decent write performance for a little while, then slow to a crawl as wear leveling kicks in, then jumps back up. On the other hand, higher end CF and nearly all SSDs don't have this problem (or at least isn't nearly as noticeable).
But, all this should be weighed against how it is being used. 100,000 write cycles in a 133x CF card might be pretty slow and seem short lived, but if you're putting it into a machine with a 5MB/s SCSI interface that you'll turn on once every 6 months, does it really matter?
Personally, I've been using SanDisk Ultra cards (plain old Ultra) and it has been fine for my uses. Most recommendations I've read seem to recommend Ultra II at a minimum and the Extreme cards are preferred. Those are much better cards, and I'd recommend them as well.
As for VM, there's nothing special about it from the card's perspective. It's just yet another IO operation. I do not know enough about the specific implementations of the various versions of Mac OS VM to be able to intelligently comment on how they would interact with CF vs. SSD.
Anyway, those are just my experiences and observations FWIW. If you read the numerous other posts here on the subject, there are a lot of people with different experiences and observations.
Both SSD and to some extent CF manufacturers seem to avoid mentioning PE cycles directly, advertising an MTBF number instead. This is kind of understandable because the SSD/CF present a block level interface to the controller, with the card/drive managing the mapping of logical block to NAND cell, with wear leveling thrown in the middle. So you don't necessarily know how many NAND cells there are (there may be significantly more than the capacity presented to the controller, which will be used for wear leveling) which makes it a little difficult to extract meaning from a PE cycle count. Then there's MLC vs. SLC in SSD, with SLC having significantly more write cycles, and MLC having comparable write cycles to most CF cards.
In addition to all that, the wear leveling implementation can make a noticeable difference in usability as well. For example, with the cheaper (I particularly notice it with 133x and lower) CF cards will get reasonably decent write performance for a little while, then slow to a crawl as wear leveling kicks in, then jumps back up. On the other hand, higher end CF and nearly all SSDs don't have this problem (or at least isn't nearly as noticeable).
But, all this should be weighed against how it is being used. 100,000 write cycles in a 133x CF card might be pretty slow and seem short lived, but if you're putting it into a machine with a 5MB/s SCSI interface that you'll turn on once every 6 months, does it really matter?
Personally, I've been using SanDisk Ultra cards (plain old Ultra) and it has been fine for my uses. Most recommendations I've read seem to recommend Ultra II at a minimum and the Extreme cards are preferred. Those are much better cards, and I'd recommend them as well.
As for VM, there's nothing special about it from the card's perspective. It's just yet another IO operation. I do not know enough about the specific implementations of the various versions of Mac OS VM to be able to intelligently comment on how they would interact with CF vs. SSD.
Anyway, those are just my experiences and observations FWIW. If you read the numerous other posts here on the subject, there are a lot of people with different experiences and observations.
Low Level Formatting Tools = partitioning tools that are already present for free in any of those OS's listed anyway? :approve:
. . . but they don't support the partitioning tools of Win'95, Win'98, early Linux Distros or the Pre-OSX Mac OS because it'd be insane to try to field support calls about the paleolithic OS choices of the users/collectors of neolithic hardware . . .
. . . IOW, us!
)
HP_Mini has a 16GB SSD as standard equipment and it even runs acceptably well off chiplets in adapters as the startup drive . . .
. . . but my expectations aren't high, it's just a supercharged Duo to me, that's why I fell in love with it at first sight! :beige:
. . . but they don't support the partitioning tools of Win'95, Win'98, early Linux Distros or the Pre-OSX Mac OS because it'd be insane to try to field support calls about the paleolithic OS choices of the users/collectors of neolithic hardware . . .
. . . IOW, us!
) HP_Mini has a 16GB SSD as standard equipment and it even runs acceptably well off chiplets in adapters as the startup drive . . .
. . . but my expectations aren't high, it's just a supercharged Duo to me, that's why I fell in love with it at first sight! :beige:
I have a 2300c with the cheapest Chinese adapter I could find and a 4GB CF card. It runs 8.6 acceptably, and a program like ClarisWorks 4 is very nimble on it. I have also run Office 2001, which works fine if a little slowly. I do, however, have the maximum RAM installed.
Alright, I've sent out payment for my duo(as some of you may know, my old one died), and I managed to get a Transcend 2.5" IDE 4gb SSD on eBay for 26$.(Here is the ad if you want one) So, I will soon have some nice boot-up speed comparisons and such. Wish me luck 
Great! I'll be a'waitin' the results! 
System requirements is probably more to do with TRIM support in the system's OS, which of course is not going to be present in any of our beloved machines. OCZ drives (which I'm looking at for my QS2002) use the SandForce controller which handles this at the drive level, not requiring any OS support.
Well, if the performance becomes lousy, I can always just pop it into the ide-usb adapter and run a garbage collector style mass trim on it. I couldn't find the drive controller info for this model.
It also appears that apple is only recently including support for TRIM in 10.6.6/7 and Lion, so it will be a while before(assuming we ever do) get any sort of third party(read:hacked) TRIM support in os 9.
It also appears that apple is only recently including support for TRIM in 10.6.6/7 and Lion, so it will be a while before(assuming we ever do) get any sort of third party(read:hacked) TRIM support in os 9.
Well, apparently the SSD is too new for the older ATA specs in the Duo. However, I promised comparisons and I do have them.
Since it cannot be used in the duo, the benchmarks reflect the FrankenLat. It is a 1.3 Ghz PIII-M, 392mb of RAM, Windows 2000. This is my laptop for when I must work in the windows world or need something that is better done in the FrankenLat's native 1600x1200 than the Duo's 640x480.
Using the stock 4,000 Rpm 4 gb drive, boot time from selecting the partition in PLOP to the time I have a cursor and can login is 1 Minute and 4 Seconds.
Using the New Transcend SSD 4gb drive, boot time from selecting the partition in PLOP to the time I have a cursor and can login is a LIGHTNING FAST 29 SECONDS! It also logs in almost instantaneously. Unofficial since I forgot to time it with the old drive, but it now shuts down in 10 seconds rather than about 30-45 previously. Applications open probably close to twice as fast.
It is FREAKING SWEET GUYS! I highly recommend these to anyone whose laptop can handle ATA-6. You will not regret it. However, a word of caution: The drive response times are too fast for low-end IDE-USB adapters(like the one I ordered from china for 4$ :
). G-Parted continuously had IO errors when it was on the usb adapter. I solved this by putting the SSD in the drive bay and the HDD in the adapter and copying with G-Parted like that.
Oddly enough, this laptop from 2002-03 had a 4,000 rpm drive. The duo from 1995 has a 4,900 Rpm drive. Odd huh? (Apple>Dell)
Lastly, it's plain weird having the laptop be totally silent. Well, at least until this model's industrial CPU fan kicks in...
Since it cannot be used in the duo, the benchmarks reflect the FrankenLat. It is a 1.3 Ghz PIII-M, 392mb of RAM, Windows 2000. This is my laptop for when I must work in the windows world or need something that is better done in the FrankenLat's native 1600x1200 than the Duo's 640x480.
Using the stock 4,000 Rpm 4 gb drive, boot time from selecting the partition in PLOP to the time I have a cursor and can login is 1 Minute and 4 Seconds.
Using the New Transcend SSD 4gb drive, boot time from selecting the partition in PLOP to the time I have a cursor and can login is a LIGHTNING FAST 29 SECONDS! It also logs in almost instantaneously. Unofficial since I forgot to time it with the old drive, but it now shuts down in 10 seconds rather than about 30-45 previously. Applications open probably close to twice as fast.
It is FREAKING SWEET GUYS! I highly recommend these to anyone whose laptop can handle ATA-6. You will not regret it. However, a word of caution: The drive response times are too fast for low-end IDE-USB adapters(like the one I ordered from china for 4$ :
). G-Parted continuously had IO errors when it was on the usb adapter. I solved this by putting the SSD in the drive bay and the HDD in the adapter and copying with G-Parted like that.Oddly enough, this laptop from 2002-03 had a 4,000 rpm drive. The duo from 1995 has a 4,900 Rpm drive. Odd huh? (Apple>Dell)
Lastly, it's plain weird having the laptop be totally silent. Well, at least until this model's industrial CPU fan kicks in...
Similar threads
