I have a comment/question about SSD in MBPs.

[revelated]

And not the question you're likely expecting. Quite the opposite. And I'm going to ask it with an analogy.


With the airport backscattering technology, the government and the TSA say that they are perfectly safe and will not harm human tissue. Specialists say that repeated exposure to even light amounts of such direct skin radiation can be harmful to those with pre-existing conditions but the results won't show themselves for many years. Nothing might happen for all we know, but the point is it will take time to know the definitive answer.

I see a lot of people talking about SSD and TRIM and garbage collection. SSDs, at least in the mainstream sense, are fairly new as far as presence in laptops. There are a few models out, but they haven't been out enough to really put them to the paces in terms of performance degradation. To really know, one would have to use such a drive for literally years - 5-10 years - to get a solid enough picture. Then it's really user perception of speed degradation, rather than what some benchmark tells you. Because as we all know, if I'm looking at how long an app takes to open, I can tell when it's slow and when it's not. The general rule of thumb is, when my free space drops below 5GB, the drive will be somewhat slower until I free space.


I have an Intel X-25M 80GB SSD in my MBP17". I'm aware of its speed compared to old platter drives, including 7200 RPM platters. I know that if I were to have the exact same image on an OCZ, an OWC, a Kingston, a Crucial, a WD, or an ADATA SSD, I would see the exact same perceived performance. In effect the only thing I really need to care about at this juncture is the reliability of the drive itself which, again, isn't really known due to the newness of the technology in the mainstream as well as the fact that some of these manufacturers are fairly unknown in the SSD realm. What's more, the tech is improved almost daily, which means an Intel that you did buy 3 years ago might not be as reliable as one bought today, obviously.


Thus my question: Why are people so hung up about TRIM/garbage collection when we don't know if it really matters in a day-to-day usage situation and won't until we get extended usage reports from regular users having used a given SSD for 5-10 years?

 

[Thiol]

Thus my question: Why are people so hung up about TRIM/garbage collection when we don't know if it really matters in a day-to-day usage situation and won't until we get extended usage reports from regular users having used a given SSD for 5-10 years?

Depending on your read/write usage pattern, you can see degradation in a few months. It all depends on how heavily and in what way one uses an SSD. Yes, most people won't notice any changes, but some people will.

 

[Stvwndr219]

It's more a problem for people working with big files that they have to import/export over and over again (read photographers, filmmakers, etc).

Anandtech had some reviews when SSD's first came out that showed some pretty significant differences after benchmarking/read/writing all the time.

 

[revelated]

Let me try this again.

Ignore benchmarks and focus on what your eyes see.

We don't know at this juncture if there truly is a value for TRIM/garbage collection as far as the user's perception of speed degradation over time. I'm asking, why are people hung up on it now? It's still not clear whether it's something to care about as far as user perception. In any event, if you lose 1 second, will you really care? I submit that only the most retentive of folks will.

 

[Kenndac]

Beats me. I guess people really want an SSD, can't really justify the cost but are using TRIM/Garbage collection/etc as an excuse.

Personally, I've been using a first generation Intel SSD (X-25M) that doesn't support TRIM with Mac OS X, which also doesn't support TRIM, for well over a year. I recently moved it from my Macbook to my iMac, and my iMac is super duper fast compared to last week when it was running from a normal hard drive!

 

[tibi08]

Let me try this again.

Ignore benchmarks and focus on what your eyes see.

We don't know at this juncture if there truly is a value for TRIM/garbage collection as far as the user's perception of speed degradation over time. I'm asking, why are people hung up on it now? It's still not clear whether it's something to care about as far as user perception. In any event, if you lose 1 second, will you really care? I submit that only the most retentive of folks will.

What's your point - not to worry about something that might not happen, even though it has been demonstrated that it could happen? On that basis, I might not bother to turn up to work on Monday, as I don't know if my employer's perception of my attendance might ever be an issue....

 

[revelated]

What's your point - not to worry about something that might not happen, even though it has been demonstrated that it could happen? On that basis, I might not bother to turn up to work on Monday, as I don't know if my employer's perception of my attendance might ever be an issue....

Your analogy is faulty. A better analogy is this.

You have a job. Your shift is 8-5pm. You're hired and expected to work that exact shift down to the minute. For the first few months you're diligent, but it's entirely possible you might show up late a few minutes, or leave a few minutes early, or come back from lunch late, etc. Problem is, at the time you're hired, they have no way to know if you'll end up being a slacker or not. In fact, you might be a model employee for years. Assuming that one will end up being a slacker will get companies wasting money on employees they have no confidence in based on assumptions.

Even better, if a person is a few minutes after the buzzer, most companies have learned that it's really no big deal if their job performance has not dwindled.


What I am saying: People seem to be harping about TRIM/GC based on assumptions and third-party benchmarks that may or may not mean anything to them this immediate moment. Buy the product that meets what you want, use it for 3 years or so, then if you see degradation, reformat or replace it. But most aren't going to see any significant degradation.

 

[mrmister]

"Ignore benchmarks and focus on what your eyes see. We don't know at this juncture if there truly is a value for TRIM/garbage collection as far as the user's perception of speed degradation over time."

That simply isn't true--especially in the first generation of SSDs to be adopted, people saw measurable, day-to-day speed and performance issues once they filled the drive. I don't think it is nearly the problem that it was even a year ago, but if you're honestly asking why people get worked up over it, it is in part because it recently *was* an issue.

 

[revelated]

"Ignore benchmarks and focus on what your eyes see. We don't know at this juncture if there truly is a value for TRIM/garbage collection as far as the user's perception of speed degradation over time."

That simply isn't true--especially in the first generation of SSDs to be adopted, people saw measurable, day-to-day speed and performance issues once they filled the drive. I don't think it is nearly the problem that it was even a year ago, but if you're honestly asking why people get worked up over it, it is in part because it recently *was* an issue.

We're not talking about filling up the drive. I have already acknowledged the fact that once your free space drops to a certain point you will see what I can only describe as pathetic performance.

I am talking about the operating system performing garbage collection and TRIM and other operations to the drive in order to optimize it - similar to a defrag - which of course Snow Leopard does not support, but are being advertised on many SSDs as a feature. A lot of people are, IMO, getting too caught up in this TRIM/GC stuff when there is really no evidence that it will even affect a regular everyday consumer's perception of speed or performance. It really does hearken back to the early days when everyone claimed that defragging improved drive speed when really, the faster RPM drives that came into play basically made speed gains non-existent, because the drive never got slower with fragmentation as far as the user's experience. Same deal with TRIM/GC and SSDs.

Any drive that has no free space is going to be impacted by performance. I'm simply suggesting that pining over TRIM/GC at this stage is a pointless endeavor because the user experience does not change regardless of whether the drive does or does not support TRIM/GC. In other words, for now, RIGHT NOW, it just does not matter in the long run. The reliability of the drive as reported by users is really the only differentiating factor.

 

[InfinitySquared]

On an SSD, when you mark a file as erased, it still has to go through a few hoops to write to a part of the drive that ever had a file on it.

What this basically means is that, from a performance perspective, you are working with a full drive.

TRIM/GC makes headway against that problem. Is the performance degradation so intolerable as to render the drive unusable? Perhaps not. Will you suffer severe performance degradation, depending on your usage patterns? Perhaps not. But regardless of what you say about the value of benchmarks, this is the bottom line: If a drive can only write at half the speed it could when it was new, you're going to notice.

 

[barmann]

Let me try this again.

Ignore benchmarks and focus on what your eyes see.

It's still not clear whether it's something to care about as far as user perception. In any event, if you lose 1 second, will you really care? I submit that only the most retentive of folks will.

If perception is all you need, fair enough .
Perception will not tell you about a creeping performance loss, though.

So if actual performance is what one needs, a quick disk speed check once in a while won't hurt - and the occasional reinstall of the system from a clone.
After that, even my perception tells me things work faster .

 

[revelated]

But regardless of what you say about the value of benchmarks, this is the bottom line: If a drive can only write at half the speed it could when it was new...

But what I'm saying is HOW DO YOU KNOW the degradation will ever get that bad? You can't base it off of a benchmark, I'm sorry. The only way to know if that will be the case is to actually use it in the mainstream for a long period of time in order to properly assess the user experience over time. Benchmarks are flawed on their face because they assume facts that may or may not apply in a given situation, i.e. rapid read/writes of certain data that don't apply to regular consumers.

The only way to know for certain if it matters is to just use the damn thing for 5-10 years and see based on your individual usage pattern.

 

[barmann]

But what I'm saying is HOW DO YOU KNOW the degradation will ever get that bad? You can't base it off of a benchmark, I'm sorry.

If your harddrive takes only 2 weeks to loose 50% of performance , chances are you won't even notice .
Of course your productivity will suffer .
You claim to remember the speed of your mechanical HDD - no you don't.

But if you don't use your Mac for work, it really doesn't matter much.

 

[Thiol]

I used to have a first-generation Intel X25-M in a Macbook Pro. I'm a scientist, so I move a lot of large data files around. After about a year or so, it was easy to perceive some sort of degradation. Applications would take a couple more seconds to open or close. This didn't really affect my workflow, but I can personally say that it was noticeable. Just my subjective two cents.

 

[revelated]

If your harddrive takes only 2 weeks to loose 50% of performance , chances are you won't even notice .

Dude, that isn't even a possibility with the crappiest of drives. 50% performance loss could only happen with (A) mistreatment or (B) a full drive. Not regular usage where there is plenty of free space and then all of a sudden 2 weeks in you start to see apps load as slow as with a HDD - it's not realistic.

 

[fehhkk]

The degradation due to use in the X-25M G2 series is minimal and imperceptible under normal use.

 

[wirelessmacuser]

But what I'm saying is HOW DO YOU KNOW the degradation will ever get that bad? You can't base it off of a benchmark, I'm sorry. The only way to know if that will be the case is to actually use it in the mainstream for a long period of time in order to properly assess the user experience over time. Benchmarks are flawed on their face because they assume facts that may or may not apply in a given situation, i.e. rapid read/writes of certain data that don't apply to regular consumers.

The only way to know for certain if it matters is to just use the damn thing for 5-10 years and see based on your individual usage pattern.

After reading your posts, the question that comes to mind is what are you trying to accomplish?

I can only speak from personal experience. I've been using, testing, benchmarking performance of over a dozen SSD's I've installed in several MBP's & ThinkPads over the last few years. Respectfully nothing I've read here makes any sense to me.

 

[revelated]

After reading your posts, the question that comes to mind is what are you trying to accomplish?

I can only speak from personal experience. I've been using, testing, benchmarking performance of over a dozen SSD's I've installed in several MBP's & ThinkPads over the last few years. Respectfully nothing I've read here makes any sense to me.

I've already answered that question...twice. Which means you either didn't read the posts like you claim, or you read them wrong.

Perception of speed is all that matters to a casual user. Period. The perception of speed does not change in SSDs except, as I stated, due to misuse of the drive or a full drive. Normal use will reveal no noticeable performance degradation for at least 5 years or so, at which time newer, cheaper replacements would be available with better technologies.

What I am saying is - all of these threads worrying and pining over TRIM/GC are jumping the gun. It's a buzzword at this stage of the game that may or may not really mean anything. There's got to be a logical reason Apple did not see fit to include TRIM support in Snow Leopard. Maybe it also thought TRIM was not something a casual user needed to concern themselves with?

 

[barmann]

I've already answered that question...twice. Which means you either didn't read the posts like you claim, or you read them wrong.

You don't give an inch, do you ?

Alright, SSDs are the dog's bollocks, will outlast all of us, and can even grow facial hair .
Happy ?

 

[AttilaTheHun]

I've already answered that question...twice. Which means you either didn't read the posts like you claim, or you read them wrong.

Perception of speed is all that matters to a casual user. Period. The perception of speed does not change in SSDs except, as I stated, due to misuse of the drive or a full drive. Normal use will reveal no noticeable performance degradation for at least 5 years or so, at which time newer, cheaper replacements would be available with better technologies.

What I am saying is - all of these threads worrying and pining over TRIM/GC are jumping the gun. It's a buzzword at this stage of the game that may or may not really mean anything. There's got to be a logical reason Apple did not see fit to include TRIM support in Snow Leopard. Maybe it also thought TRIM was not something a casual user needed to concern themselves with?

Ok. just tell me witch SSD 256GB shall I buy to my MBP 17" i7 2.66