Intel SSDs RAID 0, A Case Study In Speed, Take 2

This post has 19 Replies | 5 Followers

Top 10 Contributor
Posts 24,850
Points 1,114,970
Joined: Sep 2007
ForumsAdministrator
News Posted: Fri, Oct 17 2008 3:04 PM

We recently showed you how Intel was intent on upping the ante in Solid State Disk performance, with our evaluation and performance analysis on the release of their X25-M series SSDs.  Though offerings from other SSD manufacturers like OCZ and Samsung have come to market with better performance since then, there was no question Intel's SSD flat out smoked the competition in the cost-effective, consumer grade MLC (Multi-Level Cell) SSD market. With an average sustained throughput of ~225MB/sec for reads, around 74MB/sec observed write performance, and blistering fast sub-millisecond random access, we were left thoroughly impressed by Intel's first consumer-ready effort in SSD technology. However, at the time of launch, we only had access to one of these new SSDs from Intel and as such couldn't provide you with RAID performance metrics back then. 


 




Of course, that changed the other day when the local courier delivered another Intel kit to our door. As such, and with a bit of that "Friday on our minds" attitude adjustment going on in the lab, we decided to RAID a pair of these SSDs up to see what they could do.  Blinding speed in RAID 0 mode?  Yes, you could say that...

Test system specifications:  Intel Core 2 Extreme QX6850, Asus Striker II Extreme (790i SLI Ultra chipset) motherboard, 2GB Corsair DDR3-1,333, GeForce 8800 GTX


 








 
Sandra HDD Read - Click for full view



Sandra HDD Write - Click for full view








IOMeter Results - 8K File Size, 80% Reads/20% Writes, 20% Random Access


Though there is a pronounced saw-toothed performance pattern here, you can see that a pair of these drives offers, you guessed it, up to double the IO throughput of a single drive.  For any standard SATA RAID 0 array we've tested to date, these are easily the fastest IOMeter numbers we've seen.  Interestingly, our Sandra tests show the drives offer 396MB/s for read performance and 130MB/s write performance, while HD Tune and HD Tach show peaks and valleys from 200MB/s to 300MB/s.  Regardless, we hope you enjoyed this quick-take performance test of what Intel's new SSDs can do in a performance-targeted RAID 0 setup.  As always, with RAID 0, be sure to back up your data since you're effectively doubling your available failure points.  Regardless, we're sure many of you have run RIAD 0 setups reliably for years now and there's no question a pair of Intel's X25M drives will make for a potent storage subsystem, especially as an OS volume.

Word is Intel's performance-tuned SLC drives are waiting in the wings too. So stay tuned here for our analysis as we get our hands on one of those beasts.




  • | Post Points: 155
Top 25 Contributor
Posts 3,409
Points 46,520
Joined: Nov 2005
Location: Metropolis
ForumsAdministrator
Moderator

Wow! Are you using the motherboard's onboard RAID controller?

 SPAM-posters beware! ®

  • | Post Points: 20
Top 10 Contributor
Posts 6,181
Points 90,135
Joined: Aug 2003
Location: United States, Virginia
Moderator

 

News:

Though there is a pronounced saw-toothed performance pattern


I'll take even the lowest point of that saw tooth. These things are crazy fast!

 

Super Dave:

Wow! Are you using the motherboard's onboard RAID controller?

 

Looks like it. I see Nvidia stripe as the HD name.

  • | Post Points: 20
Not Ranked
Posts 5
Points 55
Joined: Oct 2008

Well from what the system spec says it looks like this was only using the onboard raid. Awsome! I was wondering that exact thing. I asked EVGA tech support if this SSD would work in RAID 0 on thier 790i Ultra SLI board and they said it would work fine as long as the throughput doesn't go over the SATA2 spec. Nice to see that it not only works but it works FAST! Here is another good review of these drives in RAID: http://www.extremetech.com/article2/0,2845,2329595,00.asp

  • | Post Points: 5
Top 75 Contributor
Posts 1,218
Points 18,720
Joined: Jul 2003
Location: United States, Texas
warlord replied on Sat, Oct 18 2008 8:11 AM

Can't wait to have that kinda performance in my box. It's about time we see the bottleneck that was hard drives finally catch up

  • | Post Points: 5
Top 25 Contributor
Posts 3,476
Points 53,895
Joined: Jul 2004
Location: United States, Massachusetts
ForumsAdministrator
MembershipAdministrator
Dave_HH replied on Sat, Oct 18 2008 9:28 AM

Yep, simple on-board NVIDIA SATA off the Southbridge in RAID 0. Not sure exactly what's up with the saw-tooth pattern actually. Have to play with things a bit more...

Editor In Chief
http://hothardware.com


  • | Post Points: 5
Not Ranked
Posts 1
Points 20
Joined: Oct 2008
kris2lee replied on Sat, Oct 18 2008 8:24 PM

Very nice! I just got two Super Talent OX 32 GB drives. Haven't been able to test them in RAID 0 configuration because I can not boot into Vista (64) when I configure RAID from BIOS, even when my system is on separate hard drive.

Anyway, this saw-toothed performance seams to be something unique to SSD's. Tooth in my graph are different still - there is more of them, about 30.

Minimum transfer rate of single drive by HD Tune was 108.1 MB/sec, maximum is 146 MB/sec, average was 132.1 MB/sec.

Access time was 0.2 ms and burst rate was 148.6 MB/sec.

I also did write speed test by coping large file from one SSD to another one and average was around 75-80 MB/sec in Vista. This is my estimate, it started on quiet high level like 100 MB/sec and went down from there to 70 MB/sec for 10 GB file. Not really accurate but will give some idea.

In Europe you can get one for about 100 euro that will make about 3,125 euro for gigabyte. Intel X18-M 80 GB will cost about twice per gigabyte.

  • | Post Points: 20
Top 25 Contributor
Posts 3,476
Points 53,895
Joined: Jul 2004
Location: United States, Massachusetts
ForumsAdministrator
MembershipAdministrator
Dave_HH replied on Sat, Oct 18 2008 10:23 PM

I think the big peaks and valleys are a result of high erase/re-write latency of MLC flash. You don't see that with SLC actually. And welcome to HH kris2lee. Yes

Editor In Chief
http://hothardware.com


  • Filed under: ,
  • | Post Points: 5
Top 25 Contributor
Posts 3,409
Points 46,520
Joined: Nov 2005
Location: Metropolis
ForumsAdministrator
Moderator
Super Dave replied on Sun, Oct 19 2008 12:26 AM

kris2lee:

Anyway, this saw-toothed performance seams to be something unique to SSD's. Tooth in my graph are different still - there is more of them, about 30.

That saw-tooth pattern presented itself when I built an experimental RAID array using a couple of old WD 80Gig IDE hard drives with a Silicon Image RAID card, so I don't believe it is unique to SSD's (perhaps to RAID)! Take a look:

 

 SPAM-posters beware! ®

  • | Post Points: 5
Top 75 Contributor
Posts 1,370
Points 20,925
Joined: Feb 2005
Location: new york city
ice_73 replied on Mon, Oct 20 2008 8:50 AM

just thought id mention to you guys you made engadget.

  • | Post Points: 5
Not Ranked
Posts 1
Points 5
Joined: Oct 2008
zewtastic replied on Mon, Oct 20 2008 10:59 AM

Here are the results from Velociraptor Raid 0.

  • | Post Points: 5
Not Ranked
Posts 1
Points 35
Joined: Oct 2008

Looks fast! And it also unfortunately looks like a great way for intel to sell SSD controllers. What we really need is for Intel and/or AMD to start thinking intelligently about SSDs.

AMD and Intel need to agree on a standard for an integrated SSD controller. And then create a new open standard for a Flash SSD DIMM socket.

Then I could buy a 32 or 64 GB SSD DIMM and plug it into a socket next to my RAM, and have a SUPER-FAST hard drive. Imagine a SSD DIMM that costs $50 and puts out even better numbers than the Fusion-IO! With economy of scale, it would only cost a few dollers per CPU and a few dollars more for the motherboard. But the performance would shatter the current paradigm.

The cost of the DIMMs would be low because there would be no expensive controller on the module, like there is now with flash SSDs. And that is how it should be! There is NO need for a controller on a memory module! How we ended up taking this convoluted route baffles me. It is a fatally flawed design that is always going to be bottlenecked by the SATA interface, no matter how fast it is. The SSD MUST have a direct link to the CPU in order to unleash its true performance potential.

This would increase performance so much that if VIA did this with their Nano CPU, they would have an end product that outperforms even Nehalem in real-world everyday PC usage. If you dont believe me, you need to check out the Fusion-IO. With SSD controller integration, you can have Fusion-IO level performance for dirt cheap.

If you understand what I am talking about here, and can see that this is truly the way to go with SSDs, then you need to help get the word to AMD and Intel. Whoever does it first is going to make a killing. I'd prefer it to be AMD at this point but it just needs to get done.

  • | Post Points: 35
Top 10 Contributor
Posts 6,181
Points 90,135
Joined: Aug 2003
Location: United States, Virginia
Moderator

Yeah it will be intresting to see if the keep the same hard drive format. Its kind a waste of space.

  • | Post Points: 5
Top 25 Contributor
Posts 3,476
Points 53,895
Joined: Jul 2004
Location: United States, Massachusetts
ForumsAdministrator
MembershipAdministrator
Dave_HH replied on Wed, Oct 22 2008 8:26 AM

Shadow,

Interesting commentary here on future architectures for SSDs. So remove the SATA interface all together, integrate the Flash memory controller on the motherboard or CPU and just let 'er rip from there? There's a whole lot of protocol, error checking etc that would get left behind without SATA. What about RAID arrays? Something needs to understand RAID commands too. You see where I'm going.

This is an interesting concept regardless, if only feasible for simple, single drive desktop or workstation applications.

And welcome to HH!

Editor In Chief
http://hothardware.com


  • | Post Points: 5
Not Ranked
Posts 1
Points 20
Joined: Nov 2008

This might be a moronic question...

I count 10 chips on the Drive.

Why can't they raid those chips internally?

Increase the speed maybe 8 times the current speed?

I always wondered why it took them so long to put multiple processors on one cpu.......

  • | Post Points: 20
Top 25 Contributor
Posts 3,476
Points 53,895
Joined: Jul 2004
Location: United States, Massachusetts
ForumsAdministrator
MembershipAdministrator
Dave_HH replied on Thu, Jan 22 2009 8:46 PM

Boss, that's being done by some SSD manufacturers right now actually. :)

Editor In Chief
http://hothardware.com


  • | Post Points: 20
Top 10 Contributor
Posts 6,181
Points 90,135
Joined: Aug 2003
Location: United States, Virginia
Moderator

So I was doing some reading and read that the saw tooth performance is because of the controller on most SSDs and that someone had a controller that fixed that issue. can't remember now were I read that? Something to look into at least. IDK.

  • | Post Points: 20
Top 25 Contributor
Posts 3,409
Points 46,520
Joined: Nov 2005
Location: Metropolis
ForumsAdministrator
Moderator

So in my situation with a RAID-0 array with platter hard drives (no SSD), would it be the Silicon Image controller that induces the saw-tooth patern?

 SPAM-posters beware! ®

  • | Post Points: 20
Top 10 Contributor
Posts 6,371
Points 80,285
Joined: Nov 2004
Location: United States, Arizona
Moderator

I wonder what the life on the SSD's will be? I red a review on how they think to test it but they dont know for sure yet. Anyways I am going to get one for my OS on my next PC

"Never trust a computer you can't throw out a window."

2700K

Z77 GIGABYTE G1.SNIPER

GIGABYTE GTX670

G.Skill Ripjaws X 16gb PC2133

Antec P280

Corsair H100

Asus Blu-ray burner

Seasonic X650 PSU

Patriot Pyro 128gb SSD

  • | Post Points: 5
Not Ranked
Posts 3
Points 30
Joined: Jul 2009
bchiu replied on Thu, Jul 9 2009 5:01 AM

I guess it does help a little with RAID0 eh..

However you can see my graph is not a jig-saw, even tho we are using exact same on-board Nvidia RAID and two Intel X25-M 80G. Not sure why..??

http://forums.ncix.com/forums/index.php?mode=showthread&msg_id=2037124&threadid=2037124&forum=103&product_id=33277&msgcount=0&overclockid=0#msg2037124

  • | Post Points: 5
Page 1 of 1 (20 items) | RSS