In what areas is an SSD SIGNIFICANTLY superior to a Momentus Hybrid XT?

Update: The poll should read "What is the lowest level where the jump up no longer translates to significant/noticable performance gains in everyday tasks?"

Please explain the poll option you picked below. Thank you.

------------------------------------------------------------------------

I've read a lot of reviews of the Seagate Momentus Hybrid XT and they all compare boot times of programs and state that while a Hybrid XT isn't as good as an SSD, programs still boot very quickly on it.

But isn't there more to SSDs than just boot times? What are some areas where SSDs do well and the Hybrid XT does really poorly in?

Would it be in installing programs, or copying/torrenting files?

Or, for every day tasks (gaming, copying files, starting photoshop), would you really not notice much difference between the Hybrid XT and a 1st gen SSD and a 2nd gen SSD?

Basically, I'm asking, how does the stark ten fold difference in the 4K speeds between a Hybrid XT's benchmarks...



and a Indilinx Controller SSD (Crucial M225)



and a Sandforce Controller SSD (Vertex 2)



translate to in terms of real world performance?


Does the leap forward from a Hybrid XT to a 1st gen SSD translate to anything in terms of real world performance on every day tasks?


Does the leap forward from the 1st gen SSD to a 2nd gen SSD translate to anything in terms of real world performance on every day tasks?

 

The overall scenario where an SSD is significantly faster than an XT is when used in a completely random pattern - the XT's nand cannot predict what you'll do next so it won't be able to offer an appreciable speed advantage over a pure mechanical HD - whereas the SSD doesn't need to 'run through' it a few times to be as fast as it is.

In my opinion, 4K speeds translate to zilch - when you want to map them to real world performance.

Just like a 7200 RPM drive can be good (Hitachi 7K500) or bad (Seagate 7200.4) in performance, the 4K 'score' is not an indicator of one SSD's superiority not only over another SSD, but also to a mechanical HD too.

Where SSD's currently have no peer is simply in read speeds - at writing though, even though the benches say otherwise, the real world speeds leave much to be desired and can still be beat by mechanical HD's when the writing is heavy and sustained.

Gets even more complicated when different drives are made to read and write concurrently - that is where a lot of optimizations still need to be made to SSD's so that they can earn a gold star for overall performance above and beyond what mechanical drives do already.

This is a good a thread as any to state what I have seen for the first time:

My VRaptor based desktop was timed installing a CS5 update today: a 19MB download was clocked at just under 39 seconds (just the install time).

My Inferno based notebook was timed installing the same update: it completed at just under 33 seconds.

This is the first time I have seen a 'win' for an update by the SSD!

What changed? I can only guess that the latest IRST drivers are much better suited/optimized to real world use with an SSD installed (v10.1046).

I'll be curious to read other's thoughts on this topic...

 

Thank you for that informed post.

That's very interesting to hear. Based on the benchmarks, the ONLY area where the Sandforce controller SSDs outperform the Indilix controller SSDs to a significant degree is in the 4K speeds!

So are you saying that in the real world, there is really no actual advantage to getting a Sandforce SSD instead of a Indilix SSD?

 

Oh no! The SandForce SSD's are the fastest in real world use that I have tried/experienced - but that does not mean their speed is directly proportional to the higher 4K numbers that they can 'score'.

Just wanted to add to my post above:

The funny thing about the SSD finally 'beating' my VRaptor is that when I actually installed the 10.1046 version of the IRST, the VRaptor is the one that installed and rebooted first (not the SSD...).


Also, in the spirit of full disclosure:

I know the platforms I'm comparing against are not identical and I'll be told soon enough that its not scientific or even valid (my comparisons).

But, in my defense (of my logic) I'm not here (on earth) to objectively benchmark components - I'm comparing a working/running/optimized workstation with all the hardware and 'tweaks' that I know of to an identically (OS/apps) configured and also highly optimized (JJB and stamatisx tweaks) current platform that should theoretically be faster than a two+ year old setup is.

The good news is it's finally getting there! Slowly, but surely.

And, I thought I would throw it in here because this is exactly the kind of information you're asking/looking for? Right?

 

Thing is we are comparing what we can get in a notebook here, the site name kind of gives some validity to this. For what you can do in a laptop the Vertex 2 or C300's are about IT right now.

You can argue about some instances of the XT in raid0 but overall real world should not trounce a SSD. Especially a SSD and a HDD for storage in that same dual bay system. Like it or not SSD's have the storage performace crown for notebooks etc.

As we get smaller and lighter systems looking into 1.8" format along with SATA III in the future this will become even more prevelant a trend.............

 

TANWare, you are 100% correct.

I am working to be able to do a proper comparison for NBR.

I will test the Hitachi 7K500 vs. the XT vs. the Inferno in the same system and setup exactly the same on each drive.

I just thought that my ramblings in my post above might be relevant because on some things, the XT is faster than my desktop based VRaptors. Sorry, I know I should have stated that explicitly, but in the forum here I sometimes forget that what I answer in one thread does not get read by everyone in another thread (even if they are very similar topics in my mind).

Hope to be caught up in another week (give or take a day or two) and finally be able to say with certainty the rankings of the HD/SSD's I have available to me.

I agree with your assessment that the SandForce based drives are the top performers right now - I just want to know by how much they are superior to HD's too (in my specific use scenarios).

I'll find out soon and post my findings for others here.

 

Actually, I'm curious to know how the Sandforce based SSDs are superior to Indilix based SSDs in real world tasks.

In what day to day tasks for a moderate to heavy user would you see a performance difference between say an M225 and a Vertex 2?

 

During heavy multi tasking the Vertex will be (roughly) twice as fast.

Keep in mind that 10 browser windows, Word, utorrent and MSN simultaneously is, from a storage perspective, very light multi tasking.

Installing Windows Updates with utorrent running and copying files simultaneously is heavy multi tasking.

As to your first post, the review in my signature shows you exactly how the XT compares with SSDs.

 

Thank you Phil. That's very helpful info.

I read the review in your signature. Unfortunately, in the review, there doesn't seem to be an Indilix based SSD.

Which of the drives being compared is the performance equivalent of the Crucial M225? It basically seems to be WD Black vs. Momentus XT vs. several newer Sandforce based SSDs.

 

The difference between C300 64GB on SATA II and a 256GB Indilinx will be impossible to notice I expect.

C300 64GB is included in the review. The Indinlinx has higher write performance, the C300 higher read.

PS. I think this is the fourth thread you created with a similar subject The only real way to find out how a SSD or XT performs is to actually experience it.

 

I'm surprised nobody mentioned IOPS (I/O's per second).

It is basically a measure of how many read/write operations a drive can handle simultaneously, and is a measurement of how well the drive multitasks. An SSD will absolutely smoke a mechanical hard drive or the SMXT in this area. To me, the ability to multitask to an almost infinite degree without ever feeling performance lag is the value of the SSD - more than transfer rates, boot times, or app load times.

Here is a video I made showing an extreme example of this: Windows 7 boot + loading 27 apps at boot time on an SSD, in about 1 minute.
YouTube - Why I love my SSD - Windows 7 boot + loading 27 applications in about 1 minute.

A real-world example of when you'd actually want this:
You are installing a game/app in the background, and launch a web browser. Even simple tasks like web browsing will be noticeably slower and feel laggier, because the install process already maxes out the IOPS of a mechanical drive.

 

That video is awesome. Glad to see your m1330 works, mine burned away.

 

Yeah, awesome video

After reading this review...

Benchmark Results: Access Time : Tom's Hardware's Summer Guide: 17 SSDs Rounded Up

I feel like level 3 is the correct choice.

 

Level 4 i think.

 

For those picking level 4, what real world normal use tasks do you think a Vertex 2 significantly outperforms the Vertex 1/M225 in?

 

In benchmarks... in most real world situations, it's hardly noticeable.

 

Agreed, except for heavy multi tasking.

 

Most users will never get too that point. You would need quite a specific scenario to see a difference and even then unless side by side it may be an unnoticable one.

While you will notice various improvements from an XT to any of these SSD's you would most likely never see the difference between the SSD's themselves. The only time you will really see it is in synthetics.

I belive the same is true of even when we start getting to higher end SATA III drives in the future. The state of the art now has made the BIG improvements. Only minor noticable improvements are left. This is not too say storage mediums at SATA III and SSD will not one day be a bottle neck like old 4,200 RPM HDDs are today, but that day is a ways off........

 

Interesting link. OCZ Vertex 2 is claimed to have 0.1ms (100 us) access time. Using that utility I get ~ 140-160 us on average.

 

Another area where Sandforce performs better is when it's 80% filled.

 

Interesting that you say that.

Approximately what is the highest percent filled where the Indilix still performs very well? I'll keep my drive below that level.

 

I'm not sure about it, it's something I read on this forum.

Might want to check it yourself.

 

The rule of thumb is about 20%. I have found this figure in several places on the net, including this Anandtech article.

The general idea is that a drive will use free/unused blocks of memory for wear leveling and performance improvements. If you are below 15% - 20% of free space, then the algorithm that the drive uses for this will start to become less efficient.

Some drives address this by simply allocating 20% of the total storage space as non-user-facing, which is why you have some drives that show up as 50GB / 100GB instead of 60GB / 120GB. However, many of the newest SandForce drives have algorithms that are so efficient, that they are able to operate at peak performance even below 15% - 20% free memory.

In short: Indilinx likes 15% - 20% free. This does not matter on SandForce drives, because the manufacturer either pre-allocated free space (if you have a 50GB / 100GB drive) or is using a controller revision that is efficient enough not to need this (newer drives with 60GB / 120GB usable).