Originally Posted by Baz Trust me, using an SSD an an OS using superfetch are nothing alike. Superfetch might pre-load the program files of some apps and titles but it won't help when booting, loading larger apps or when browsing your system. A decent SSD really is a huge step up in performance - I promise!
agree, have to use it to believe it.
the earlier days were slow but simple, remember how the web browser don't have so many addons and and only load a single page at a time? remember how simple graphical user interface were? even those days we are still bottlenecked by the hard drive, it's just not as obvious as on today's quad-core, 8GB monsters.
"The SSD market has been really hotting up lately"
hotting up??? I taken you meant heating up lol
Anyhow good review but again Im waiting until that $1gb mark before I make the switch :) but it is certainly great to watch the progression of these drives
There's various standards of wear levelling, which is why I proposed the test.
Normal wear levelling distributes writes over the spare space on the disk. This is OK, but for most boot drives, the vast majority of the data stays in one place, so that the wear accrues only on empty cells, concentrating degradation massively into one area of the flash. Under these circumstances, 1000 or even 10,000 rewrites is a staggeringly small number.
Really good wear levelling shuffles data around regardless of whether it's involved in the current write, distributing writes across all of the storage by moving previously static data to other places. This helps enormously.
Bit-Tech should enquire of flash disk manufacturers what their approach is to this issue.
And I thought flashing my PSP all those years ago was tence. lol
But still, do you think it's possible if I were to sell it on or RMA it? Maybe the same process but flashing over G.Skills 1370 (asuming downgrades are allowed).
The manufacturing costs of a SSD from what I have seen must be lower than that of a HDD due to the fact that a SSD is only a PCB as were the HDD has both PCB and mechanical parts.
With memory being so cheap is it the controller that is the bulk of the high cost, or just a case of milking the must have techno nutters.
Originally Posted by Phil Rhodes There's various standards of wear levelling, which is why I proposed the test.
Normal wear levelling distributes writes over the spare space on the disk. This is OK, but for most boot drives, the vast majority of the data stays in one place, so that the wear accrues only on empty cells, concentrating degradation massively into one area of the flash. Under these circumstances, 1000 or even 10,000 rewrites is a staggeringly small number.
Really good wear levelling shuffles data around regardless of whether it's involved in the current write, distributing writes across all of the storage by moving previously static data to other places. This helps enormously.
Bit-Tech should enquire of flash disk manufacturers what their approach is to this issue.
P
agree, especially that of the Samsung one :)
although looking at it on O&O Defrag, it seems the location of files doesn't change
Originally Posted by wuyanxu but most importantly everything feels instant and you don't feel the program is being launched from the disk, it feels as though it came from RAM because the lack of drive seaking.
Y'know, superfetch does much the same thing for me in Vista =P
Trust me, using an SSD an an OS using superfetch are nothing alike. Superfetch might pre-load the program files of some apps and titles but it won't help when booting, loading larger apps or when browsing your system. A decent SSD really is a huge step up in performance - I promise!
Oh, I have no doubt, my only doubt is whether or not I'll hate myself in the morning for spending that kind of money on that amount of storage. It's all well and good to have the speed, but it does me no good if I can only fit half my apps on the drive.
although looking at it on O&O Defrag, it seems the location of files doesn't change
It won't, at the OS level, or you'd break the filesystem. This sort of messing about is done at the drive's firmware level, which is why the only way to figure out what's going on is to ask the manufacturers.
Because we all use an 8 year old OS rebooting thousands of times a day with a £300 performance part designed for modern operating systems? I think we should also throw in some Windows 98, 95 and even 3.1 testing too just for people to see the improvement an SSD will bring to that '96 Packard Bell.
The point is not what OS you use. Use Vista if MS are paying you to vocally support the steaming pile of...
The point is how much less long lived the flash is than the hard disk. I don't know how much less good it'll be, but I suspect it'll be less good. Possibly so much less good that it's questionably usable.
P
Don't be hatin' on Vista, it's not as bad any everyone gives it stick for. And Microsoft don't pay me to say that.
Considering the average MLC SSD rewrite endurance is anywhere between 1000 -10,000 cycles, I'm personally not too worried. Even worst case, that'd mean I'd have to write 128TB of data to the drive over it's lifetime, something that realistically just won't happen in a home user, or arguably even an enterprise environment. These drives also come with specific wear levelling algorithms on the drive controller to avoid over using a specific cell of the drive repeatedly, so you'll see fairly even wear too. Even then, wear isn't going to influence performance too much (until you hit the write endurance of course) - it's the defragmentation caused by partially filled cells that causes performance drops, - something Wiper.exe and hopefully TRIM support can happily resolve
The thing with flash is that even if you want to write 1B, you will have to erase and partially rewrite a whole sector, so programs that write/rewrite small files very often will wear down SSD very quickly. Sure, in desktop you can use HDD for that, or put more RAM, but in laptops, with limited RAM and space (no second drive unless 17" laptop) SSD probably wont last very long.
The thing with flash is that even if you want to write 1B, you will have to erase and partially rewrite a whole sector, so programs that write/rewrite small files very often will wear down SSD very quickly. Sure, in desktop you can use HDD for that, or put more RAM, but in laptops, with limited RAM and space (no second drive unless 17" laptop) SSD probably wont last very long.
We've not been able to find any documented evidence of any of the SSDs we've reviewed in the last year wearing down due to exhausting the drives number of rewrites, even under heavy loads. Unless someone can produce some concrete evidence, I think you're all making a big deal out of a non-issue.
Unless someone can produce some concrete evidence, I think you're all making a big deal out of a non-issue.
It's not a case of evidence, that's how the damn things work. You won't have had them for nearly long enough to exhibit any of these problems. What we should do is produce some evidence, which would mean stress testing one of these things until it starts to fall over.
Which is what I suggested back on page 1, if you recall.
Those of you comparing raptors and SSDs need to please stfu and would the real slim shady please stand up?
Firstly, you're comparing an 8-year old drive with an SSD (and I know you are because you're commenting on noise). Secondly, the Raptors are SATA I, which means poopy speeds. And the platter density on the Raptor is made from ass. So yes. An SSD vs. a Raptor is no contest.
However, Western Digital has the 150GB Velociraptor on the same size/form factor platter and it's inaudable. The Vrap features SATA II and a balls-to-the-wall average latency. Put two Velociraptors in a stripe (Raid0) and then show me an SSD that can compete with that speed, capacity, and reliability all while staying in the same price range. It's an exponential curve in favor of Velociraptors (or SCSI for that matter) when comparing to SSDs. The simple fact that every GB is cheaper on a platter drive makes it an uphill battle for the SSDs. I can afford to stripe multiple drives at SSD prices and gain similar and much more reliable performance.
Just to confirm, you 'll need to find the OCZ firmware update that requires the drive to be jumpered. This causes the drive to appear in windows as a seperate entity (ylangylang device or therebaouts if i recall correctly), which then allows the older OCZ firmware to work.
The same process allows you to reinstall G.skills own firmware, as this still requires the use of the jumper, unlike OCZ's newer firmware updates that do not need, and in fact will not work, with the jumper.
Comments 26 to 44 of 44
Replythe earlier days were slow but simple, remember how the web browser don't have so many addons and and only load a single page at a time? remember how simple graphical user interface were? even those days we are still bottlenecked by the hard drive, it's just not as obvious as on today's quad-core, 8GB monsters.
hotting up??? I taken you meant heating up lol
Anyhow good review but again Im waiting until that $1gb mark before I make the switch :) but it is certainly great to watch the progression of these drives
If things go wrong you'll have a bricked drive - no half measures when it comes to firmware updates!
Normal wear levelling distributes writes over the spare space on the disk. This is OK, but for most boot drives, the vast majority of the data stays in one place, so that the wear accrues only on empty cells, concentrating degradation massively into one area of the flash. Under these circumstances, 1000 or even 10,000 rewrites is a staggeringly small number.
Really good wear levelling shuffles data around regardless of whether it's involved in the current write, distributing writes across all of the storage by moving previously static data to other places. This helps enormously.
Bit-Tech should enquire of flash disk manufacturers what their approach is to this issue.
P
You can say either, hotting up is more slangy though if you're going to be a gram-nazi about it :P ;)
But still, do you think it's possible if I were to sell it on or RMA it? Maybe the same process but flashing over G.Skills 1370 (asuming downgrades are allowed).
With memory being so cheap is it the controller that is the bulk of the high cost, or just a case of milking the must have techno nutters.
Technologically, DRAM is a pain. The only reason we use it is that it's extremely fast and extremely cheap.
P
last graph: meters per second?
although looking at it on O&O Defrag, it seems the location of files doesn't change
Oh, I have no doubt, my only doubt is whether or not I'll hate myself in the morning for spending that kind of money on that amount of storage. It's all well and good to have the speed, but it does me no good if I can only fit half my apps on the drive.
It won't, at the OS level, or you'd break the filesystem. This sort of messing about is done at the drive's firmware level, which is why the only way to figure out what's going on is to ask the manufacturers.
P
We've not been able to find any documented evidence of any of the SSDs we've reviewed in the last year wearing down due to exhausting the drives number of rewrites, even under heavy loads. Unless someone can produce some concrete evidence, I think you're all making a big deal out of a non-issue.
It's not a case of evidence, that's how the damn things work. You won't have had them for nearly long enough to exhibit any of these problems. What we should do is produce some evidence, which would mean stress testing one of these things until it starts to fall over.
Which is what I suggested back on page 1, if you recall.
P
Firstly, you're comparing an 8-year old drive with an SSD (and I know you are because you're commenting on noise). Secondly, the Raptors are SATA I, which means poopy speeds. And the platter density on the Raptor is made from ass. So yes. An SSD vs. a Raptor is no contest.
However, Western Digital has the 150GB Velociraptor on the same size/form factor platter and it's inaudable. The Vrap features SATA II and a balls-to-the-wall average latency. Put two Velociraptors in a stripe (Raid0) and then show me an SSD that can compete with that speed, capacity, and reliability all while staying in the same price range. It's an exponential curve in favor of Velociraptors (or SCSI for that matter) when comparing to SSDs. The simple fact that every GB is cheaper on a platter drive makes it an uphill battle for the SSDs. I can afford to stripe multiple drives at SSD prices and gain similar and much more reliable performance.
It's just not the time for solid states. Not yet.
After reading this review, I order the disk, and tomorrow the shipment will arrive, so please tell me if this is the one I need?
The same process allows you to reinstall G.skills own firmware, as this still requires the use of the jumper, unlike OCZ's newer firmware updates that do not need, and in fact will not work, with the jumper.
-
« Previous
-
1
-
2
-
Next »
Discuss in the forums