TR Forums

So I am planning on moving to SSD for the system drive with my new build and I am wondering whether to go with a traditional SSD or one of these Revo PCI-E drives. There seems to be a shrinking price difference.

Anyone using the Revo series? Anyone expect the claimed speed difference to be noticable? Anyone think the Revo's lack of TRIM will be a problem?

Thanks

Here are links...
http://www.newegg.com/Product/Product.aspx?Item=N82E16820227579&cm_re=ocz_revodrive_x2-_-20-227-579-_-Product
http://www.newegg.com/Product/Product.aspx?Item=N82E16820148349&cm_re=crucial_c300-_-20-148-349-_-Product

I dont know much about that crucial but it looks like sataIII so the revo will crush it
I use Trim myself, but if you have this much money to invest in an SSD running a PCIE or raid is not a bad option
you might lose a little of it's life expectancy, but it will still last a few good years and then they should be much cheaper

if I were you, you could get two or three cheaper Sata drives and raid them or get a revo drive
(get the revo drive, comes with total elitism bragging rights as well)

I'm using a Revo right now and coming from a Velociraptor, there is a big difference, like boot times in the thirty second range. I cannot compare it to a regular SSD as the Revo is my first, so my amazement could just be the regular HDD to SSD difference. I did expect a little more then I got, however, mostly because I never see that 500MB/s speed that's on the box, and with the Revo, you have to deal with RAID and the lack of TRIM support, and if you use it as a boot drive, there is a little bit of setup required. Personally, I like the Revo over an SSD just to have fun bragging that I have a hard drive on a card.

Do people really use boot times as a benchmark these days? And if it makes a handful of seconds difference, does that have any bearing on sequential reads or writes? No.

computron9000 wrote:

Do people really use boot times as a benchmark these days? And if it makes a handful of seconds difference, does that have any bearing on sequential reads or writes? No.

I brought it up because it's a tangible thing I noticed. I can boot firefox in less than three seconds but the difference between the 5 to 3 seconds for firefox is not as big to me as 45 to 30 seconds for a boot. I also tweak my computer a lot, so the time really adds up when you decide to change something and end up rebooting 6 times an hour.

It's hard to quantify the speedup when your program loads in ½ second instead of 2 seconds. It's easier to see and measure the difference with a longer task that is bottlenecked by I/O. When your desktop loads in under 30 seconds compared to over a minute, for example.

JustAnEngineer wrote:

It's hard to quantify the speedup when your program loads in ½ second instead of 2 seconds. It's easier to see and measure the difference with a longer task that is bottlenecked by I/O. When your desktop loads in under 30 seconds compared to over a minute, for example.

But you see, booting is usually not 100% bottlenecked by I/O

Until newer SSD's with yet unreleased controllers arrive, I break it down like this:

256GB and you have SATA 3.0: C300 256GB
Everything else: Sandforce something

Remember that the smaller versions of the C300 are slower, and not worth the price premium.

JustAnEngineer wrote:

It's hard to quantify the speedup when your program loads in ½ second instead of 2 seconds. It's easier to see and measure the difference with a longer task that is bottlenecked by I/O. When your desktop loads in under 30 seconds compared to over a minute, for example.

I get that, but do you think it's apples to apples there? Plus once the system is online, if you use a respectable OS, you get boosts from keeping things in RAM.

I guess to me it's like starting a car. At first it grinds a bit and then turns over properly and starts to go and you can drive around: if it's cold outside, it heats up to the proper temperature it runs at before performing at optimal levels, ultimately, for the user. Same with modern operating systems. You turn the computer on, you boot your OS, you wait for it to finish its business, and eventually it is "ready."

I don't think the "OS boot-up speed" of hard-drives is particularly relevant in the last 4-5+ years. I'm more interested in the sustained read performance for big files I want to load, or the ability of the drive to handle fragmentation from downloads or other adverse things.

I do not believe that boot-up times correlate to real-life performance.

The boot sequence involves both large sequential reads and smaller fragmented and writes reads with multiple parallel requests. Hence, boot performance corresponds directly to real life performance.

A drive that doesn't perform well during boot, won't perform well afterwards. Conversely, a drive that performs well during boot should perform well afterwards.

BlackStar wrote:

The boot sequence involves both large sequential reads and smaller fragmented and writes reads with multiple parallel requests. Hence, boot performance corresponds directly to real life performance.

A drive that doesn't perform well during boot, won't perform well afterwards. Conversely, a drive that performs well during boot should perform well afterwards.

Emphasis added.

Generally, you're right- the boot up sequence is a functional test of general SSD performance, and is usually added as a test when SSDs are reviewed. It is not, however, an indication of how an SSD will perform in particular situations, and therefore shouldn't be over-emphasized .

grantmeaname wrote:

Boot up times are a kind of real life performance.

I'll have to disagree quite vehemently. I reboot my computer once every month, if that much. It's 2011, are you shutting down/restarting every day?

Yes, because it takes less than ten seconds.

So in other words, you are doing something that's useless (rebooting often), but are happy because an SSD makes that useless task fast.

This is not a knock on SSDs themselves, I'd like to have one as they make the overall computing creamy-smooth. But using boot times as a real-world work benchmark is pretty ridiculous.

I'm trying to figure out why DarkStar's post and my response are being ignored here- I almost want to just quote both of you and say 'read above'.

Boot times are a measure of performance, and I agree with DarkStar's explanation; but they shouldn't be used as the primary, or only, metric for which to measure an SSD. Isn't this obvious?

It's not a benchmark. It's an aspect of the way I use my computer, and my usage model has changed because of having an SSD.

Why do you get to decide what aspects of my computing are useless and which ones are useful? I'm the one using the computer! For one thing, it's a laptop, so shutting it down increases the lifetime of the system. As a dirt-poor kid paying his way through college, that's anything but useless. For another, it's light pollution and I like to be able to fall asleep without LEDs. I could go on, but I guess since those things don't fit into your usage model they're useless.

Airmantharp wrote:

I'm trying to figure out why DarkStar's post and my response are being ignored here

My post was a direct response to them, and did not ignore them. Look:

BlackStar wrote:

boot performance corresponds directly to real life performance.

Airmantharp wrote:

... boot up sequence ... test ... test ... reviewed ... indication

grantmeaname wrote:

Boot up times are a kind of real life performance.

Do you see the difference?

Boot times are a measure of performance

Yes. But they are also an actual aspect of the way in which having an SSD affects the use of the computer in real life, not just a benchmark.

but they shouldn't be used as the primary, or only, metric for which to measure an SSD.

Did I ever argue that? For that matter, did morphine ever argue that? Nope.

grantmeaname wrote:

Why do you get to decide what aspects of my computing are useless and which ones are useful? I'm the one using the computer! For one thing, it's a laptop, so shutting it down increases the lifetime of the system.

The OP is not talking about a laptop. That's why I said booting was far from a priority.

grantmeaname wrote:

As a dirt-poor kid paying his way through college, that's anything but useless. For another, it's light pollution and I like to be able to fall asleep without LEDs. I could go on, but I guess since those things don't fit into your usage model they're useless.

One word: hibernation.

Airmantharp wrote:

Boot times are a measure of performance, and I agree with DarkStar's explanation; but they shouldn't be used as the primary, or only, metric for which to measure an SSD. Isn't this obvious?

I agree with this, I'm simply stating that looking on boot times as an important, all-encompassing benchmark can lead you to a false result, and TR's tests show exactly that. The fastest drives for booting are the Intels, yet they're not the fastest at most other tasks.

morphine wrote:

hibernation

That's brilliant! Obviously I would never have stumbled on that solution myself if you hadn't pointed it out to me!
It's not really any quicker than booting up. It puts a big messy file on the drive at all time. No benefits, no thanks.

Edit: your response did not refer to the OP. You said "you", as in "me", and I do have a laptop.

grantmeaname wrote:

morphine wrote:

hibernation

That's brilliant! Obviously I would never have stumbled on that solution myself if you hadn't pointed it out to me!
It's not really any quicker than booting up. It puts a big messy file on the drive at all time. No benefits, no thanks.

Edit: your response did not refer to the OP. You said "you", as in "me", and I do have a laptop.

My God, you're nitpicking (and threadjacking!) here. Can we get back on topic? Do you have anything to add to it that would be of use to the OP?

My apologies.
I've stated my opinion about the performance issue, which you've ignored, in order to point out how I'm not responding to the thread. That's as counterproductive, if not more.

You misconstrue what I say about the topic, and when I explain why I feel that's the case you ignore it in order to point out that I'm not being productive.

The discussion I was having with morphine was about using SSDs. That's at least a step closer to what the OP asked than your discussion about benchmarking them.