SSDs - State of the Product?

[Santilli]

Except for that one program, even playing DVD's, or Bluray run at around 40%, and thats pretty much all of them.

I'll do a bit more cpu watching, but, most of my tasks just don't stress the Xeons.

 

[ddrueding]

What I'm saying is that that single number doesn't really tell you anything. You need to watch the graphs to the right. Are some cores pegged while others are idle?

 

[Santilli]

Not for most of my useage. I may get spikes in the first couple cores to 60%, but, they go back down below 50% quickly.
I was just testing a Bluray disk, and the processors spiked to about 80% briefly, then settled down to about 20-40% while the DVD was playing.

 

[LunarMist]

There are many reasons for a creaky old machine to be slow. Is it really worth figuring out why?

​

 

[Santilli]

Who ever said anything about slow? The only thing it doesn't do really well is play games.

Other then that, for what I do with it, it's perfect.

 

[LunarMist]

The CPUs are slow and the memory is slow, among others.

 

[Santilli]

For what? How do those factors affect the things I do?
My functions seem internet speed limited, and, hard disk limited.

 

[LunarMist]

Man I could really use a cheap SSD.

 

[ddrueding]

I have 4 30GB Vertex, 2 160GB G2, and 2 32GB -Es on my shelf right now...

 

[LunarMist]

It looks like prices are generally holding or increasing.

 

[ddrueding]

The thought of computer parts as an investment is just silly.

 

[LunarMist]

Sure, but who said it wasn't? I'm just annoyed that

prices are not falling yet.

 

[ddrueding]

You want some more -Es?

 

[Handruin]

You guys probably read the news, but there was an announcement of the Intel/micron 25nm NAND chip for Q2 this year. Basically doubling the capacity of the recent 34nm chips in roughly the same amount of physical space. They're discussing 8GB (64 Gb) monolithic chips capable of 200MB/s throughput. Micron is already in progress of sub 25nm NAND chips, so they will be getting smaller and larger in the near future. This should hopefully bring up sizes and drop prices of older drives.

 

[jtr1962]

That's definitely good news. It seems like SSD prices haven't moved down in quite a while. I'd like to take the plunge on SSDs myself, but for now they're too pricey for my tastes. I'm hoping they get to ~$100 for 250 GB in the not too distant future. Last I checked on Newegg, they were about 6-7 times that.

 

[LunarMist]

I want either a faster, 6Gbps X-25E series for around $500 or something cheap, but decent like the X-25M series for around $150. I wish they made a 40GB X-25M.

 

[Santilli]

Two 30 gig Vertex Turbos, 309 bucks, Tiger.direct.

I want to max out my card, and, the bus.

That will give me 120 gigs for a c drive, at 255 mb/sec or better.

See what that does.

So far, they only confirm what I found with SCSI, access time RULES. I wish Eugene was still around...

I ran the recent version of ATTO while ripping a Bluray with AnyDVD, and was still getting writes around 100 MB/SEC, and reads around 200 MB/sec, with 20% cpu usage...

 

[Santilli]

With the SSDs, watching a bluray movie, ripping, AnyDVD, another, doing this, and, cpu usage is mainly 25-30%, with high spikes on P1 at 50%, and, 1.15 gb pf usage.

 

[Santilli]

http://www.newegg.com/Product/Produ...GNEFL021610-_-SolidStateDisk-_-L0C-_-20227393

LOOKS PRETTY GOOD. Really 119 dollars

 

[Handruin]

That drive is a bit too small for anything I'd want to do. I'd need 80GB or preferably 160GB which is considerably more expensive.

That drive is $3.96/gig
80 GB Intel X25M = $3.375/gig
160 GB Intel X25M = $2.98/gig

 

[LunarMist]

http://www.newegg.com/Product/Produ...GNEFL021610-_-SolidStateDisk-_-L0C-_-20227393

LOOKS PRETTY GOOD. Really 119 dollars

Look at the reviews. There are plenty of bad ones and numerous similar defects. The manufacturer seems to have poor customer service and a useless rebate policy.

 

[ddrueding]

I've never had one fail, so I guess their RMA dept. is out of practice.

 

[LunarMist]

Are you are using them personally in heavy use or elsewhere?

 

[ddrueding]

I only have about 10 in use here, but have deployed at least another 50, probably more. I have also stuck a pair of 120GB Vertex and 160GB X-25M in a very heavily used VM Server for several months and they are holding out nicely.

 

[Santilli]

Has anyone any information on changing block sizes, and how that might affect performance? I wonder if I screwed the pooch by using 64k as the block size in the raid array...

 

[Handruin]

Block size changes are best when you analyze the average data pattern you will be using with the drive. For example, there is information that suggests a SQL Server drive should be formatted with 64K blocks because of the way SQL writes to the disk. You should also take into account your stripe size in the raid array in conjunction with the block size.

At a bare minimum, I would suggest using the default 4K block size until you can figure out your usage pattern. Also keep in mind if you go beyond 4K, I believe the standard windows defrag will no longer work on an NTFS formatted volume.

 

[udaman]

OCZ Vertex 2 Pro...Sandforce controller, just like the Runcore...but

Is everyone who uses laptops going to now upgrade to a SATA 6Gbps model...whenever those ship (only seen USB3.0 on some new Dell or was it HP?) ...just to use Crucial C300 SSD?

Supposed to ship/go on sale next week (less expensive than Intel, on GB/dollar basis):

http://www.dailytech.com/Crucials+6Gbps+C300+Hits+Stores+February+22+Targets+Intel/article17599.htm

It uses the same 34nm NAND flash memory as Intel from their joint venture IM Flash Technologies, but the C300 uses a new controller from Marvell along with custom firmware developed by Micron.

http://www.tweaktown.com/reviews/31...56gb_sata_6gbps_solid_state_disk/index11.html

Wonder if there are SF 6Gbps controllers in testing with the likes of OCZ or Runcore, etc?

 

[Santilli]

Block size changes are best when you analyze the average data pattern you will be using with the drive. For example, there is information that suggests a SQL Server drive should be formatted with 64K blocks because of the way SQL writes to the disk. You should also take into account your stripe size in the raid array in conjunction with the block size.

At a bare minimum, I would suggest using the default 4K block size until you can figure out your usage pattern. Also keep in mind if you go beyond 4K, I believe the standard windows defrag will no longer work on an NTFS formatted volume.

HMMMM.
Defrag is not an issue, since the drives are two Vertex Turbos, 30 gigs, in Raid 0, for about 59
gigs of boot drive. Reads are about 200 mb/sec.

Others have been getting faster results...

It's kind of too late, since it's already done...

 

[Handruin]

Did you go with 64K? Did you rerun ATTO with 4K and beyond? I'm curious if you noticed any differences.

Hopefully you don't need to store lots of smaller files. You'll chew through space pretty quick if you do. If I'm not mistaken, even a 1K text file will now take 64K to store on your drive.

 

[Santilli]

Did you go with 64K? Did you rerun ATTO with 4K and beyond? I'm curious if you noticed any differences.

Hopefully you don't need to store lots of smaller files. You'll chew through space pretty quick if you do. If I'm not mistaken, even a 1K text file will now take 64K to store on your drive.

I looked at the transfer rates below 64k, and they are really terrible.
current ATTO results below:

4k was 25 mb/se
8 36mb/sec
16 64 mb/sec
32 was 95 mb/sec
64 was 106 mb/sec
from there,
128 is 132 mb/sec
256 is 173 mb/sec

 

[Handruin]

It might be worth researching how your SSD structures data during reads and writes via its controller. If you fit the block size of the file system to the block size of the SSD and align the format, you might end up getting better performance.

 

[Handruin]

You're doing RAID 0 over the two, correct? What stripe size did you pick? There are mixed reviews that suggest a 4K stripe is faster than the normal suggestion of 128K.

If you formatted the file system as 64K and your stripe doesn't match, that could be greatly affecting your performance numbers.

 

[Handruin]

Newegg has the Intel 80GB X25M (SSDSA2M080G2XXX - OEM) with 3 year warranty for $219 with free shipping. Seems like a decent price considering it was $299 only a couple days ago.

I think the drive is just missing the installation kit from the retail package, but I doubt it's worth $80.

 

[Santilli]

It appears the smaller the stripe, the better the HDTach results, and, the worse the ATTO
results.

System feels snappier with the 16k stripe.

I'm going to assume the atto numbers have problems with the SSD's, or, at least the version I'm using.

More after install...

 

[Handruin]

Check out HD Tune. It seems to have a good variety of tests, some similar to ATTO. It's a lot more modern than ATTO also. Try the demo.

 

[ddrueding]

Anyone else drooling over the ioDrive Octal?

The ioDrive Octal has the following specifications:
• 800,000 IOPS (4k packet size)
• 6 GB/s bandwidth
• 5 TB maximum capacity
• x16 Gen-2 double-wide PCI Express form factor

 

[Stereodude]

I know I'm not...

 

[ddrueding]

Party pooper

C'mon, they kept building until they could saturate a 16x PCIe Gen2 slot. That is pretty awesome.

 

[LunarMist]

I have no interest either, but I'm not in IT.

 

[LunarMist]

http://www.newegg.com/Product/Produ...GNEFL021610-_-SolidStateDisk-_-L0C-_-20227393

LOOKS PRETTY GOOD. Really 119 dollars

I received that OCZ (what a disgusting name). It is just so-so in performance, being slower than the X25-M and slower yet than the X25-E. Of course that is fine in absolute terms. I have not found any defects yet, so it will be utilized in the backup computer as the boot drive to replace an old hard drive from 2007.