The Lounge - CodeProject

If you like being stuck with a 6GB pipe, keep with the old, comfortable magnetic.

If you want to abuse a 32MB pipe, you have to go with an M2 SSD. This is where they make magnetic media look like dinosaurs.

You can pull data off of a dead SSD, it just requires somewhat more specialized equipment/software.

Thanks for the reminder, btw. I'm overdue to create a system image at the house.

"There are three kinds of lies: lies, damned lies and statistics."
- Benjamin Disraeli

You're fixating on the wrong numbers (and I don't just mean that you've written "MB" instead of "GB" for PCIe bandwidth D'Oh! | :doh:

, or that the "B"'s really should be "b"'s since you're talking about bits not bytes with those numbers D'Oh! | :doh:

, or that you're comparing SATA3 before encoding overhead to PCIe after subtracting it D'Oh! | :doh:

).

For sequential IO spinning rust might be 4-10x slower than flash but is fast enough that it works just fine. Where SSDs crush HDDs is random IO where latencies drop from ~10ms to ~10uSec (1000x faster) and total throughput goes from ~1MB/sec to ~50-500MB/sec (50-500x faster); a situation where spinning rust is painfully slow. When your SSD equipped computer seems much more responsive than your old spinning rust crippled one it's the random IO performance that makes the difference. Cool | :cool:

Did you ever see history portrayed as an old man with a wise brow and pulseless heart, waging all things in the balance of reason?
Is not rather the genius of history like an eternal, imploring maiden, full of fire, with a burning heart and flaming soul, humanly warm and humanly beautiful?
--Zachris Topelius

Training a telescope on one’s own belly button will only reveal lint. You like that? You go right on staring at it. I prefer looking at galaxies.
-- Sarah Hoyt

Silly fat fingers and lack of proof-reading.

The comparison came from a different source, a Tom's review if I recall correctly. It's entirely probable that I misread Gb as GB in the article.

"There are three kinds of lies: lies, damned lies and statistics."
- Benjamin Disraeli

Just to clear up some facts:
- the '6GB pipe' appears to be referring to the SATA 3 data bus, which is limited to a theoretical I/O throughput of 6Gb/s (750MB/s) of raw data, or about 4 Gb/s (500 MB/s) of user data. That is still a lot more than what current mechanical drives achieve. Moreover, the SATA bus in no way limits the speed of random data access, which is the main advantage of SSDs, not the data throughput! For everyday use, a SSD easily speeds up any drive operation related task by a factor of 10-100. And this, for the most part, does not depend on the bus being used!
- the '32 MB pipe' may be referring to PCIe 3. IIRC, PCIe does not use data bus, instead it uses DMA, and therefore is not affected by the same limitation as SATA. I have no idea what the limits for DMA, are, but I suppose the '32MB' should probably read '32 Gb/s'.
- 'M2' devices use either SATA or PCIe as data interface. So when you advise picking M2 SSDs, you are probably referring to the improved performance you specifically get from M2/PCIe devices, not M2/SATA! You can get PCIe SSDs without M2 however, so 'M2' is not the deciding factor here! Moreover, M2 is currently not well supported (the Microsoft driver is abysmal), so I would advise against it without first picking up more detailled information.

One additional thing to consider is just how much data do you tend to read and write over the course of a normal day? Megabytes? Gigabytes? Terabytes? If you get to the Terabyte range, then a PCIe SSD may be well worth the investment. Otherwise, chances are that you hardly notice the difference to a standard SATA SSD! Other than solid drivers, and double the space for the same money. Just saying Wink | ;)

GOTOs are a bit like wire coat hangers: they tend to breed in the darkness, such that where there once were few, eventually there are many, and the program's architecture collapses beneath them. (Fran Poretto)

Thank you muchly for the detailed explanation.

"There are three kinds of lies: lies, damned lies and statistics."
- Benjamin Disraeli

So in all of this discussion so far no one talked about the limited read/writes. I haven't done any homework so maybe it is writes only or that no longer exists. I have an old win 7 laptop that the hd light is almost always on. I have removed a lot of software and narrowed the hd access to some windows system file. Can't recall the name off hand. I'm assuming they are attempting to index the drive in some fashion, even with the indexing service disabled or maybe they are reporting all I do back to Microsoft, who knows, but it bugs me.

So windows is hitting the hd a lot, wouldn't that cause an ssd drive to fail faster? Maybe I'm the only one who has this happen and it really isn't an issue.

To be fair, I don't know that I care if my laptop boots faster. (My old Commodore took 20 minutes to load a game from tape.) I might care when actually taxing the system, but I have too much crap for a tiny ssd to support. I know there are other probably better ways but I'm an old dog.

Jack of all trades, master of none, though often times better than master of one.

Hmmm, that is my attitude about the boot up time, if you have enough memory so stuff doesn't get chached to the hd there is no real advantage and a possible disadvantage.

I will disagree 100% on this.
I had 32GB of memory and a non-ssd boot.
Booting took a while. Serveresque machine, Oracle, etc.

So, I throw in an SSD, and my boot time is 30-40 seconds, which is easily 3-5 times faster.
In fact, it is SO fast to boot, that I stopped using Hibernation! It actually took longer to hibernate (and the hibernation file required 32GB of my boot SSD).

I wondered if SSDs would fail, and did a BUNCH of analysis of the data out there. I am a grey beard. My first HD was an RK06 on a DEC PDP-11/34a... From there a 5MB TRS-80 drive, and then eventually my own 5MB, 10MB all the way to todays 4TB drives.

I have worked with EMC where they explained that when building RAIDs they cannot use too many drives from the same batch of drives because they tend to fail at about the same time!

I have lost a few HDs in my day. And I have had 2 SSDs go bad. One was a Crucial, and it was a firmware issue where they said after X,000 hrs start turning off after like 90 seconds. They replaced it under warranty.

I have put them in TiVo boxes just to see. I took an Older SSD I had no use for, plug it into a TiVo 4 years ago, and it has not failed yet. That is a LOT of continuous writing.

Boot time, and operating speed. I can't buy this performance with a processor.
Sound: Really quiet!
Battery Life improved as well.

They have been around for some time. Engineering principles have been applied. Once they made them and were willing to put them into RAID configurations, your fears should have dropped a bit.

And finally. yep, restoring from them AINT EASY when they die. It may not even be completely possible. But think back. How long after HDs came out did we get to the white rooms of today? It will get there. Companies will figure it out, get some patents and open up a company.

In the meantime, my machine backs up weekly (full), differential (to full) nightly.
To a NAS device that I have been thinking would be fun to upgrade to all 4TB SSDs when the prices come down...

You'll find that since the drive can be written to and read faster, that light will be on much less of the time. Only the occasional blink.

If you have room for an HDD, and you have low RAM like 4GB or something, I'd recommend putting your page file on the HDD if you want to prolong the life of your SSD.

SSDs can be read indefinitely. It's only the writing that's limited. However, there's little to fear, as this article shows! The first SSD failing in that endurance test failed after writing more than 700 Terabytes of data successfully. Even if you write 100 GB of data every day, that's close to 20 years of expected lifespan!

That said, there are a couple of settings in Windows (and presumably other OSs as well) that you really should adjust. I've seen an article about just that topic, but can't recall it's title, nor can I locate a bookmark, if I even took one.

But the gist of it - to the extent that I remember - is that that you should disable anything continuously performing write operations; and you need to really think about what a specific process really does in order to find out who are the culprits:

(A) Anything meant to speed up searching or indexing - e. g. Windows Searchindexer - may often write to the disk to update its index, but is pointless on a SSD, because it reads so much faster. It's less of a problem when the index is kept in memory, but it's still pointless.

(B) Defragmentation is a SSD killer, and likewise entirely pointless for an SSD. The main point of defragmentation is to avoid the slow random lookups on mechanical drives. It is a workaround for the implicit weakness of these drives. But SSDs don't have that problem to start with!

(C) Antivirus software likes to write logs about just about anything. Make sure it is set up to only report relevant stuff so it doesn't continuously write to the disk! (or, alternately, if you have that option, you could tell it to write the log to a mechanical drive instead) Normally this shouldn't be a problem, but I mention it because at work our antivirus software is configured to log unholy amounts of information all the time. I can't see just how much it really is, but I have a strong feeling it's worth checking.

GOTOs are a bit like wire coat hangers: they tend to breed in the darkness, such that where there once were few, eventually there are many, and the program's architecture collapses beneath them. (Fran Poretto)

I installed them on a couple of my desktops, after reducing the system drives to fit and cloning them across.

I only use the system drive to store the system and a few programs that act up if they're not in %PF% (which is a really bad place to put programs, anyway). Everything else is stored externally, so if the SSD fails, I can just plug the old system drive back in, with minimal fuss.

I wanna be a eunuchs developer! Pass me a bread knife!

There's a 850 PRO 512GB in my lap top. I got it from Amazon cheap because it allegedly had a damaged package. The only hitch was that Samsung didn't yet have cloning software for Windows 10 at that time. This was just after Windows 10 was released. One of Samsung's help desk people helped me find cloning software for 10.
No problems so far, but this machine is gets moderate use. My top reason for getting the SSD was because a laptop gets banged around and the SSD is more forgiving of bumps.

I've replaced all the hard drives in my systems with 'em ... so far not a bit of trouble, and the machines run like greased lightning. Fast, quick upgrade. If they have any problems I've not seen them.

I'm running a terabyte Crucial drive in my laptop. I'll second what everyone else has said: Nice speed boost, no extra heat, better battery life, no real sensitivity to physical shocks. With prices dropping like they are, there's no reason not to try one out.

Most of the problems with them wearing out have been mitigated by randomizing where writes go. My understanding is that writes to what would normally be the same track/sector in a spinning drive, go to different locations each time with an SSD. Usage is statistically modified so no one area of the disk is overused.

Also, on the subject of data recovery from the chips, most of the modern SSDs now use encryption as part of that randomization scheme. If you don't encrypt the disk, you can plug it in any system and read it, but the bits and bytes are still encrypted on the other side of the disk's interface electronics. On the plus side, this makes full disk encryption quick, the data is already encrypted, so you're just adding your key to the existing key (I know it's much more complicated than that, just trying to keep things simple).

I do still keep my backups going, but I did that with the spinning drive as well.

SSD optimize performance on my desktop machine, but it I keep its HDD on it. SSD had problems right after installation - there was a bug with NCQ (whatever that is), which I was able to turn off modifying an option in linux kernel (unfortunately, no windows on SSD for now). Moreover, it wasn't plug-and-play, I had to turn-off command ordering (an elevator algorithm used to speed up head movement on HDD). Finally, I had to put only operating system files on the SSD, because warranty expires not only with time, but with amount written (70TB), which is achievable in less than 6 months with my data usage. Using it on a system files I was able to speed up installed programs and boot-time, which is great, but SSD are definitely not suitable for complete replacement.

Here's what my crappy work ssd looks like after a few years of windows 7 and visual studio:
ssdlife report

Runs like brand new and visual studio is able to max out the cpu when compiling. When I switched to it from the old spinning rust, the build time dropped from 50+ minutes to around 15.

Just like a lot of electronics these days, it seems that if it doesn't crap out on you in the first couple of weeks it'll be fine.

I've got to this post while looking for a SSD online... The one in my fabulous laptop is dying after less than 6 months of light usage! Before I decided to jump into the 21st century, I had a modest Toshiba Satellite, now over 8 y.o. That magnetic drive, as slow as it might be, has gone thru several OS-s, restores and thousands of installed programs over the years without a glitch! And it’s still doing its job faithfully to this day…

My new system is a Dell Precision M4800, equipped with an I7@2.8GHz and 32GB RAM, and came with a 256GB LITEON IT SSD drive. Impressive performance! I’ve replaced the DVD drive (useless) with a second (magnetic) 1TB drive, for data, some programs, virtual machines, etc. On SSD – just the Win7/64 and a handful of programs. By the way: what is the point to have the system booting in 10 seconds and then having the pagefile, most programs and redirecting access to a slower drive?!

I’ve noticed recently the system becoming unresponsive, while furiously accessing the HD… After days of research (the S.M.A.R.T. system proved to be a S.T.U.P.I.D. joke!), it turned out to be a failing SSD, which gave I/O errors. I’ve only used this laptop for probably 6 months, took all precautions to reduce usage of SSD. Spent days doing this – an unnecessary process on a regular drive – while enjoying the luxury of having a second drive in my laptop…

I’m not sure if I should buy a new SSD or a magnetic one… I’m old school: if I have to choose between reliability and speed/performance, I ALWAYS choose reliability!
It’s not about loss of data: I have backups, and most of it (even programs) reside on the magnetic drive anyway. It is about the loss of a drive, a drive deemed to die by design! It is about trust and the uneasy feeling that using the drive is like burning a candle!
This is in tune with where everything else is going (i.e. Win10, Office 365, etc.): you buy something, but not really own it (not for a long time, anyway). You pay a subscription. It would be a subscription, this time for hardware. It is made to expire, and every few years (or months) you have to renew it, by buying a new (expensive) drive.

Anyway, I might be subjective and maybe just an unlucky, not a typical case.

Luschan wrote:
what is the point to have the system booting in 10 seconds and then having the pagefile, most programs and redirecting access to a slower drive?!

You're right: you'll boot up at most maybe five times a day, if that much, so the total time saved doesn't add up to much. However, accessing the file system or calling system functions really add up to a significant amount of time!

As for page file: don't put it on the SSD if you can avoid it. Either up your RAM to the point where you don't need one, or move it to a mechanical drive.

More generally, if you don't want to fully switch to SSD, here's a guide to setting up your Windows system to properly manage your SSD. You can find many similar guides on the web.

GOTOs are a bit like wire coat hangers: they tend to breed in the darkness, such that where there once were few, eventually there are many, and the program's architecture collapses beneath them. (Fran Poretto)

So I was just doing some online clothing shopping when I noticed a shirt of over €500!
An actual Versace.
The shirt goes well with other Versace clothing priced well over €1000.
Washing instructions: chemical.
So we have a pretty basic (dare I say ugly) t-shirt that's impossible to clean and which costs over ten times more than a regular t-shirt (up to fifty if you go for a basic t-shirt).
And people are actually happy with it.
People are weird Unsure | :~

Visit my blog at Sander's bits - Writing the code you need.
Or read my articles at my CodeProject profile.

Simplicity is prerequisite for reliability.
— Edsger W. Dijkstra

Regards,
Sander

Versace died in 97, so, nooo, it's not an ACTUAL Versace.

If it's not broken, fix it until it is

Yeah, but the stitching, man!

I am not a number. I am a ... no, wait!

Frankenstein's Monster agrees Smile | :)

Visit my blog at Sander's bits - Writing the code you need.
Or read my articles at my CodeProject profile.

Simplicity is prerequisite for reliability.
— Edsger W. Dijkstra

Regards,
Sander

Versace is nothing compare to Tom Ford. The shoe prices are so expensive makes you think is this forever that can't be worn-out.

And he is alive too. Smile | :)

To be fair, that's the most beautiful (and expensive) sandal I've ever seen[^] WTF | :WTF:

Visit my blog at Sander's bits - Writing the code you need.
Or read my articles at my CodeProject profile.

Simplicity is prerequisite for reliability.
— Edsger W. Dijkstra

Regards,
Sander

Looking at the page, that looks like $1090 for one sandal! Big Grin | :-D

And most people have to outfit two feet! Cool | :cool:

Welcome to the Lounge