Vertex Solid State Disks - Finally up and live

JasonSTL739

Weaksauce
Joined
May 5, 2008
Messages
72
Figured I would share my new setup - I don't post here much but have gleaned lots of information from lurking around in the past. I previously had a pretty decent rig - but this new build is cheaper than the old one and much much faster. I'm really quite impressed with it for the $.

This new one is all about SSD's. 4X30GBrs on RAID0 with OS installed and a MFT partition to be specific.

Overall system:
Asus P6T6 WS Revolution Motherboard
Intel i7 920 2.66 running at 3.61 (up to 3.8 stable in turbo mode)
Noctua cooler
12GB OCZ Platium DDR3 CAS7 running at 1443Mhz
XFX GTX 260 Core 216 896MB

Storage controllers:
Onboard ICH10R for SSD's OS/programs/data
Areca 1220 with battery backup for standard disks/data

The OS - Windows 7 Build 7057 - is running on a 30GB aligned partition against the 4X30GB Vertex SSD's in a RAID0 array. I'm running firmware 1275 on them, 128 align with diskpart, and a 128k stripe in the ICH10R.

The rest of that same array is formatted as a MFT volume, D:\, which addresses any write issues that might exist with the SSD array (essentially creates a huge RAMdisk for writes as all writes just go into memory). I'm not sure that on the RAID array if this is really necessary actually, because of their on drive cache the benchmark differences aren't as drastic with the typically little windows writes as you would see with other SSD's. HOWEVER - if you get something like photoshop or lightroom working hard, it is another story entirely.

Data is served from a standard RAID10 array of 1.5TB disks.

Results are impressive for four little 2.5" disks that are about the same cost today as two WD Raptors!

I'll let the bench do the rest of the talking. It is clearly limited by the ICH10R, which is OK with me at 600MB/sec. :D

rgvxnl.jpg


The MFT volume is very difficult to bench, as it the cache in RAM will cause results from most standard benches to be all over the place. Short bursts - which is typical in actual use, easily see 1GB/sec in write bandwidth.

2niabmw.jpg
 
That looks like quite a system, and I am thinking of something along similar lines.

Of worry, what's the long-term use of 30GB drives? I think we all know that eventually we're going to get that capacity and performance in single 2.5" enclosures, in fact it may come quickly, and I'm not sure what happens to drives such as these. A 500GB or 1TB magnetic disk I can hand off to a friend; even a 120GB SSD would make a fine boot drive for a laptop or netbook, but once I outgrow these I can't see myself doing anything with them but tossing them in the closet.

Are there any cheap enclosures with eSATA that would work with these sorts of drives? Actually, a couple of years from now, if wrapped in something cheap and solid, these could be handy portable USB or eSATA drives for moving movies around :) So maybe they would see life past 2009.
 
Now, if you can, run that Crystal DiskMark with default settings on it, would love to see what that shows.

Damned impressive speeds, utterly ridiculous for the cost... very nice.
 
Now, if you can, run that Crystal DiskMark with default settings on it, would love to see what that shows.

Damned impressive speeds, utterly ridiculous for the cost... very nice.

On C and D, respectively. Probably not the cleanest benches. I don't care that much about benches really. Hell I didn't even close outlook. :p

29ncjno.png

2py43dw.png
 
I wonder why the 4K read speed seems to always be capped at around 25 MB/s?
 
Jason - awesome setup man. Can't wait for FW1275, although not a single issue with FW1199 here.

For kicks, here are my 2 Vertex120's (FW1199) on my ICH10R. Rock solid and hella fast.

64129144.png

vtxr01199128dmewce.png

vtxr01199128dmewceatto.png
 
Jason - awesome setup man. Can't wait for FW1275, although not a single issue with FW1199 here.

For kicks, here are my 2 Vertex120's (FW1199) on my ICH10R. Rock solid and hella fast.

64129144.png

vtxr01199128dmewce.png

vtxr01199128dmewceatto.png

Fantastic example of how limited my 4 30GBers are by the SATA controller! :)
 
Nice, I am really more and more impressed with the Vertex drives. I want to get one for my laptop.
 
Now, if you can, run that Crystal DiskMark with default settings on it, would love to see what that shows.

Damned impressive speeds, utterly ridiculous for the cost... very nice.

Can you tell me why everyone is so in love with that particular test?

As a 30GB Vertex owner, I should note that mine has dropped from 170/90 seq r/w to 70/15 in the span of 2 days. I'm assuming I just got a defective unit, but I would be doing potential buyers a disservice if I didn't mention that here. And to be honest, the real-world "wow" factor wasn't all that impressive even when it was (briefly) working at peak performance. The lack of seek noise was by far more noticeable than the minor increase in responsiveness.
 
The OS - Windows 7 Build 7057 - is running on a 30GB aligned partition against the 4X30GB Vertex SSD's in a RAID0 array. I'm running firmware 1275 on them, 128 align with diskpart, and a 128k stripe in the ICH10R.

Might I ask where you came by firmware 1275?
 
Might I ask where you came by firmware 1275?

I had called into support at OCZ before they stopped allowing it out. The version I have is not to be redistributed.

A real public 1275 will be out shortly - I think today or tomorrow in fact; so it isn't that big of a deal time-wise.
 
Anyone have any suggestions on a PCIe RAID controller (or two controllers) that could take on the following:

-4 30GB Vertexes (obviously the ICH10R is a limiting factor) for data/programs/VMs
-Two X25-E's on RAID0 for OS
 
Anyone have any suggestions on a PCIe RAID controller (or two controllers) that could take on the following:

-4 30GB Vertexes (obviously the ICH10R is a limiting factor) for data/programs/VMs
-Two X25-E's on RAID0 for OS

The ICH10R is no slouch... For RAID0 it shouldn't really matter.
 
So, FW1275 is roughly the same performance as 1199 on my freshly flashed 120's. However, I and another user (everyone I'm assuming) show a significant performance hit in 512k random reads n writes. CDM shows this. Pretty substantial hit. I was rock stable on FW1199 - shoulda stuck with it.
 
So, FW1275 is roughly the same performance as 1199 on my freshly flashed 120's. However, I and another user (everyone I'm assuming) show a significant performance hit in 512k random reads n writes. CDM shows this. Pretty substantial hit. I was rock stable on FW1199 - shoulda stuck with it.

You can always flash back to 1199....

I have some serious corruption issues with 1199 in both systems (laptop and desktop) when pushed hard - so I'll pass on 1199.

If you can't find it/don't have it PM me and I can email you the 1199 firmware.
 
Why is everyone so hung up on CrystalDiskMark? I'd never even heard of this bench prior to the Vertex threads on this forum. Seems like more of way to benchmark the NTFS file system than the actual disk performance.

FWIW I got a slight increase in performance with 1275 on my Vertex 30GB. I did not show a "significant" performance hit in 512 random anything. I hope those of you making such claims are aware that different host controller modes and file systems can affect the drive's performance in canned benchmarks significantly. I can get 220 MB/s sequential reads if I run my board in legacy IDE mode, but it only comes out to about 175 if I run in Intel RAID mode (which is what I normally use).
 
Why is everyone so hung up on CrystalDiskMark? I'd never even heard of this bench prior to the Vertex threads on this forum. Seems like more of way to benchmark the NTFS file system than the actual disk performance.

FWIW I got a slight increase in performance with 1275 on my Vertex 30GB. I did not show a "significant" performance hit in 512 random anything. I hope those of you making such claims are aware that different host controller modes and file systems can affect the drive's performance in canned benchmarks significantly. I can get 220 MB/s sequential reads if I run my board in legacy IDE mode, but it only comes out to about 175 if I run in Intel RAID mode (which is what I normally use).

My testing shows the same - and the stability differences in my setup was quite pronounced. I could easily corrupt 1199 as a single disk or an array by pounding it with heavy I/O.
 
My testing shows the same - and the stability differences in my setup was quite pronounced. I could easily corrupt 1199 as a single disk or an array by pounding it with heavy I/O.

Yeah, same thing that happened to me... 1199 (shipping for my unit) just up and died one day in the middle of copying a 1,600 file game to it :p

Flashed to 0112 and got abysmal performance, set the drive aside and waited for 1275... flashed that last night and now I'm screaming along again... time will tell if the lockup issue is solved.
 
Back
Top