Doing an ICH10R SSD RAID0 comparison, thoughts?

[lowfat]

All tasks were run on the following:
Core i7 920 @ 4.2GHz w/ HT off
4 x 80GB X25-M G1
6GB Aenon PC3-12800 @ 1600MHz 9-9-9-18
Gigabyte EX58-UD5
2 XFX HD 4980s
Asus Xonar Essense STX
1000W Antec Quattro

A system image was created. Then before each install on each setup, an HDDerase of all the drives was completed. I performed a reboot between every test. Superfetch & page file were disabled. All startup items were disabled through msconfig.

Results:






As was thought my many people here, RAID0 on ICH*R just doesn't benefit a lot of tasks out there. The best bet would to be to buy the fastest single SSD you can find. Buying multiple slower SSDs just shows no real benefits.

[Griff805]

Do you have a hardware RAID controller available to test with?

[Darth_Penguin]

great to here, I look forward to seeing the difference between the 3 setups. I take it you are using the X25s?

[lowfat]

Do you have a hardware RAID controller available to test with?

currently no. Which is why the thread title says ICH10R

great to here, I look forward to seeing the difference between the 3 setups. I take it you are using the X25s?

Correct.

[XS Janus]

Simultaneous file read and write performance!

Make a large 3GB+ Folder with lots of jpegs, and mp3s then measure the time it takes for it to duplicate each individual setup.
Timing how long it takes the setups to do the same with a large .mkv or .iso file would also be useful.

THX!

[lowfat]

Simultaneous file read and write performance!

Make a large 3GB+ Folder with lots of jpegs, and mp3s then measure the time it takes for it to duplicate each individual setup.
Timing how long it takes the setups to do the same with a large .mkv or .iso file would also be useful.

THX!

what do you mean by this?

Just copy a 3gb+ folder of pictures/mp3's. And paste them to the same drive?

[Mic]

+1 for the copy test
A large folder with a bunch of files in various sizes to same drive.

[Ao1]

I split my raid 0 set up around a month ago and have only just rebuilt it. That was long enough to get used to how a single drive performed and having gone back to raid 0 I have noticed zero difference. Now that drives are getting bigger I don’t see the point in raid 0 for desktop use, although I know some people say they see the difference. I’d be interested to know where the differences become obvious, cos I just don't see it with my usuage patten.

I'd also like to see if the slightly better performance of a bigger drive gives any benifit in real life. A 160GB G2 is technically slightly faster that a 80GB G2 plus the data/ free space ratio is likely to be better on bigger drives so I'm wondering if a 160Gb drive would actually be better than two 80Gb in raid 0?

[tiro_uspsss]

I'd also like to see if the slightly better performance of a bigger drive gives any benifit in real life. A 160GB G2 is technically slightly faster that a 80GB G2 plus the data/ free space ratio is likely to be better on bigger drives so I'm wondering if a 160Gb drive would actually be better than two 80Gb in raid 0?

the only thing the single 160-G2 *may be faster in is access time & thats it.. IOPs, MB/s etc, the 2x 80-G2 in R0 will totally lay waste to the single 160-G2

[Ao1]

Hmmm...outside of benchmarks I see zero difference between a single X25-e and two X25-e's in raid 0. Tomorrow I will have a 80gb G2 to compare against two X25-e's in raid 0. I somehow doubt I will be able to tell the difference unless I benchmark.

Edit: AS Benchmark scores improve by 60% with raid 0, but that does not translate to faster real life performance.

Edit 1: Rather than install multiple programs would opening say 100 Photoshop files simultaneously do the same thing if the files sizes varied?

[aamsel]

I split my raid 0 set up around a month ago and have only just rebuilt it. That was long enough to get used to how a single drive performed and having gone back to raid 0 I have noticed zero difference. Now that drives are getting bigger I don’t see the point in raid 0 for desktop use, although I know some people say they see the difference. I’d be interested to know where the differences become obvious, cos I just don't see it with my usuage patten.

I'd also like to see if the slightly better performance of a bigger drive gives any benifit in real life. A 160GB G2 is technically slightly faster that a 80GB G2 plus the data/ free space ratio is likely to be better on bigger drives so I'm wondering if a 160Gb drive would actually be better than two 80Gb in raid 0?

Especially with an onboard controller (I have ICH10R) I agree with you.
I just ordered a single 160GB G2 over the weekend for that reason. Also, I won't have to deal with array failure (should it occur) which I have had with RAID 0 in the past, and should get TRIM support under Windows 7 much quicker than I would for two X25-M's in RAID 0 on an ICH10R.

I have no doubt that two of them (or of the 80GB X25-M G2) will show much higher throughput and beat the single drive in most benchmarks, but doubt that I will see or feel any real world difference.

[tiro_uspsss]

Hmmm...outside of benchmarks I see zero difference between a single X25-e and two X25-e's in raid 0. Tomorrow I will have a 80gb G2 to compare against two X25-e's in raid 0. I somehow doubt I will be able to tell the difference unless I benchmark.

Edit: AS Benchmark scores improve by 60% with raid 0, but that does not translate to faster real life performance.

Edit 1: Rather than install multiple programs would opening say 100 Photoshop files simultaneously do the same thing if the files sizes varied?

then I'd say u dont do anything that maxes out the IOPs of a single SSD, hence why u cant 'feel' the diff of a 2nd in R0.. it definitely does translate to faster real life performance - if u require that much IOPs out of ur storage sub-system
as for the PS idea.. yeah could work - its a whole lot of reading, something SSDs specialize in - do it @ same time theres a lot of writing - then u'll have some fun!

[Ao1]

^^ There is a big difference between the iops of a vertex and Intel drive, yet the view is there is no noticable difference between the two drives. This is why I think that iops over a certain value do nothing for desktop use. If the difference between a vertex and intel drive can not be noticed why would two drives in raid 0 be noticed? I'm not saying that raid 0 is not faster in theory, but in practice, at least from my usage pattern, it is not.

[Baron_Davis]

Raid 0 is good for people who actually USE it. It's like someone complaining that their GTX 285 isn't that much faster than their old 8800 GTS in CS 1.6.

[cx-ray]

I went from an X25-M to RAID 0 X25-M G2 160GB and definitely feel that my system has gotten faster. For instance, there's a noticeable improvement when opening the Event Viewer in Vista 64 or when generating thumbnails in Windows Photo Gallery (very basic examples).

When I initially switched from a Velociraptor 300 to an X25-M I was a little disappointed. Sure things were faster, but it doesn't have the snap that my system had now.

[Rhys]

Raid 0 is good for people who actually USE it. It's like someone complaining that their GTX 285 isn't that much faster than their old 8800 GTS in CS 1.6.

Raid 0 only benefits large packet requests. i.e. when dealing with files at least double the size of the stripe. (You can see this more clearly with WBC off)
Thing is you won't perceive any benefit until you begin working with files that are large enough to be able to notice.

Also by some accounts raid slows down boot and app times slightly.

So lone wolf, with all your wisdom you spout, do you actually have any ssd's in raid 0?

Lowfat I would be very interested if you could make a large 10gb PSD file in photoshop and time how long it takes to save changes of the image.
And also how long it takes the image to load up by clicking it's icon from a cold boot with photoshop not open.

The above is where I would expect to see real world improvements.

Also just an idea, it may be worth doing some comparisons with wbc on and off to see which is really best and use the best method for the remaining tests.

[Ao1]

I’m not complaining What does USE it mean? One Hurtz has posted a time of 12s to open 40 or so apps simultaneously with 2 x25-e’s. Low Fat does the same test with a single X25-M and does it in the same time. Griff ran a load of real world tests between a single X25-m and two X25-m’s in raid 0 and the differences were negligible if anything. What exactly do you need to be doing to see raid 0 become a benefit and how often is that going to happen in real life?

That’s a rhetorical question btw, I’m hoping that it is something that Low Fat can work out when he runs his tests.

[lowfat]

That’s a rhetorical question btw, I’m hoping that it is something that Low Fat can work out when he runs his tests.

I'm open to some suggestions

[Baron_Davis]

I’m not complaining What does USE it mean? One Hurtz has posted a time of 12s to open 40 or so apps simultaneously with 2 x25-e’s. Low Fat does the same test with a single X25-M and does it in the same time. Griff ran a load of real world tests between a single X25-m and two X25-m’s in raid 0 and the differences were negligible if anything. What exactly do you need to be doing to see raid 0 become a benefit and how often is that going to happen in real life?

That’s a rhetorical question btw, I’m hoping that it is something that Low Fat can work out when he runs his tests.

RAID 0 is good for high sequential reads/writes...

IE. videos, working with massive images, doing a lot of work with massive files, like copying gigs of data from one folder to another, etc.

Also good if you need a lot of IOPS for example a web server where there are a lot of read requests off the DB, and this is where the Intel X25-M benefits greatly, because it excels at small random reads and writes, exactly what a web server does.

When looking at real world or "common man" performance, you won't see a big diff from RAID 0 because it's just not that demanding. It's like the example I gave with CS 1.6. Until someone starts playing Crysis, they won't notice a big difference between their old graphics card and a current gen one.

Also, you have to realize that what people post in this forum is either e-peen stuff that doesn't reflect reality, or they're quoting synthetic benchmarks that may not reflect the real world performance BECAUSE you're not stressing your drives enough to see it.

So for example, when you see some guy say it takes him 18 seconds to load 1 billion apps, what he's REALLY saying is that compared to your average 1 SSD setup, he will load his 1 or 2 applications 1/10th of a second faster.

Bottom line, you gotta ask yourself. Am I a professional that needs very high IOPS or very high sequential reads/writes or am I just getting RAID to show off my e-peen and impress some forum nerds?

[Ao1]

I'm open to some suggestions

I think it is worth a few people chipping in to agree what would constitute normal os use and what is more cpu or ram dependent. If it would help I could run the exact same tests on my system between 2xX25-e’s and one X25-M g2. (As long as it does not involve installing 40 apps
I think it would also be really interesting to see if over clocking the cpu & ram makes any difference to read/ write speeds. (I’ve not noticed any difference but I can only get to 3.7 and 1600.)
I don’t do much more than; create ISO’s (rare). Create wave and mp3 files typically ranging from 6mb to 700mb. Zip/ unzip files. Play COD . Create/ save ps files (no bigger than 50mb.) Watch blue ray. Rip dvd’s/ blue ray. Rip cd’s. Create 3D images and render (rare.)
The only time I’ve noticed my drives suffer was when I copied 600mb of 4k files and that was with raid 0.

Edit: It would also help if people that have noticed a real world difference between raid and a single drive give some examples. Cx-ray, I'm not sure why you would see any delay in the items you mentioned. I've just opend a file with 1,600 mp3's with embeded artwork. If I switch to thumbnail view it is instant and the mp3's are on a 7.200 hdd!

[Ao1]

Baron Davis I agree with you. I think it’s fair to say that ssd's scale well, but the benefit of that scaling is limited unless you have a specific high demand use.
In some ways it’s like sli (except the scaling)It makes me smile when the first question that 90% of people asks when they have first set up their cards is “how do I know it is working”. If it was something you could notice you would not ask that question. A case of emperor’s cloths.

[XS Janus]

what do you mean by this?

Just copy a 3gb+ folder of pictures/mp3's. And paste them to the same drive?

Yes, that is exactly what I meant.
Simple but real life scenario R/W test.

Just remember to save the folder, for future testing. So the results be comparable.

[cx-ray]

Cx-ray, I'm not sure why you would see any delay in the items you mentioned. I've just opend a file with 1,600 mp3's with embeded artwork. If I switch to thumbnail view it is instant and the mp3's are on a 7.200 hdd!

I used specific examples. Not just thumbnails anywhere and I have no idea whether your files already had them generated before viewing. Another example of generating thumbnails that take time are the ones from in the Desktop Background, but only when they aren't there yet after a fresh OS install.

[Ao1]

^ fair comment. I've just tried opening new folders with Photo Gallery and the thumb nails come up instantly. Tomorrow I will try that with a single G2.

[cx-ray]

Many or perhaps even the majority of imaging applications save thumbnails together with files. Perhaps if you put a good number of movies in there you might see it generating, or Windows Media Center is another example.