SSDs - State of the Product?

[ddrueding]

I've got my 42", 27" and 24" monitors at the office, and everyone knows what they are missing

I told them that they, like me, are free to buy what they want and I'll hook it up for them.

 

[DrunkenBastard]

http://www.anandtech.com/storage/showdoc.aspx?i=3778&p=7

AMD’s 890GX delivers a staggering 180MB/s in random write performance, a full 31% faster than Intel’s X58. The random write speed makes me believe that with a bit of driver and/or BIOS work we can get random read performance up there as well.

...

Next page:

"We have to give AMD credit here. Its platform group has clearly done the right thing. By switching to PCIe 2.0 completely and enabling 6Gbps SATA today, its platforms won’t be a bottleneck for any early adopters of fast SSDs. For Intel these issues don't go away until 2011 with the 6-series chipsets (Cougar Point) which will at least enable 6Gbps SATA."

I would recommend reading the whole article starting from page 1. It seems many mobo manufacturers are using workarounds to the default 1 lane of PCI bandwidth to the Intel SATA controllers on the non X58 mobos. But what is going into the Dell econoboxes?

I'm also curious what caused his Crucial C300 died of. And what two of them in RAID 0 do in terms of showing up controller limitations.

*idle rambling* Clearly these limits aren't going to be hit in most use cases today. Joe Average checking his gmail and browsing the web is not going to be impacted. However, who could have imagined an SSD under a $1000 delivering over 300MB/sec and 20,000 IOPS even 3 years ago.

In three years time, that kind of performance could be stock standard. And that's when the limits will be hit. *end rambling*

 

[Buck]

U.S. Navy ships (maybe vessel class namesakes - Seawolf and Kidd are class names).

Oliver - Oliver Hazard Perry class Frigate
Austin - Austin class Amphibious Transport Dock
Seawolf - USS Seawolf (SSN-575) Submarine
Kidd - Kidd class Guided Missile Destroyers also USS Kidd (DDG-993)
Wasp - Wasp class Amphibious Assault Ships also USS Wasp (LHD-1)
Joshua - also known as WOPR

These systems aren't named after my kids

 

[Santilli]

http://www.anandtech.com/storage/showdoc.aspx?i=3778&p=7

AMD’s 890GX delivers a staggering 180MB/s in random write performance, a full 31% faster than Intel’s X58. The random write speed makes me believe that with a bit of driver and/or BIOS work we can get random read performance up there as well.

...

Next page:

"We have to give AMD credit here. Its platform group has clearly done the right thing. By switching to PCIe 2.0 completely and enabling 6Gbps SATA today, its platforms won’t be a bottleneck for any early adopters of fast SSDs. For Intel these issues don't go away until 2011 with the 6-series chipsets (Cougar Point) which will at least enable 6Gbps SATA."

I would recommend reading the whole article starting from page 1. It seems many mobo manufacturers are using workarounds to the default 1 lane of PCI bandwidth to the Intel SATA controllers on the non X58 mobos. But what is going into the Dell econoboxes?

I'm also curious what caused his Crucial C300 died of. And what two of them in RAID 0 do in terms of showing up controller limitations.

*idle rambling* Clearly these limits aren't going to be hit in most use cases today. Joe Average checking his gmail and browsing the web is not going to be impacted. However, who could have imagined an SSD under a $1000 delivering over 300MB/sec and 20,000 IOPS even 3 years ago.

In three years time, that kind of performance could be stock standard. And that's when the limits will be hit. *end rambling*

His choice of drives to test with leave a bit to be desired:
Didn't even come close to what the X-58 Gigabyte will do:



Also, I've seen a PCI-E 2 run nearly 800 mb/sec on that motherboard...

4 Vertex Turbos on the same motherboard:

 

[Pradeep]

Yes, of course with a single C300 he wasn't going to exceed say 320MB/sec. Personally I would have liked to have seen him get 2, 4 and more to test them out.

Here's 16 X-25-Es delivering over 3.4 TB/sec over PCIe 2.0, way back in 2009, I'd never seen it before:

http://www.tomshardware.com/reviews/ssd-6gb-raid,2388-4.html

If you average that out each drive is doing over 200MB/sec. Not too shabby.

 

[Santilli]

I'm getting nearly 600 mb/sec on my Gigabyte Exteme, and, David had that one slot raid doing nearly 800 mb/sec.
No dust on those PCI E channels...

 

[Pradeep]

Ironically at 3.4GB (sorry typo there) sustained per sec their RAID 0 array of 1TB of space would run out of room in less than 5 minutes.

 

[ddrueding]

True, but that would be one hell of a cache for a larger storage system.

 

[Pradeep]

Indeed. Especially with the next gen of E series drives likely going to 256GB, you could be looking at up to 4TB of "fast cache".

 

[Pradeep]

http://www.anandtech.com/storage/showdoc.aspx?i=3779

The resolution to his problem with the C300, looks like a problem after Trimming, firmware tweak should fix it.

 

[LunarMist]

Indeed. Especially with the next gen of E series drives likely going to 256GB, you could be looking at up to 4TB of "fast cache".

I'd be happy with a faster 128GB X25-E drive that did not cost an arm and a leg.

 

[Santilli]

I'd be happy with a faster 128GB X25-E drive that did not cost an arm and a leg.

Well, I played with David's 800 mb/sec Z drive machine.
And, I've used mine, which is now approaching 600 mb/sec.

David has tried them side by side, and, he assures me you can't tell the difference. That is also my experience. I wonder if any human could tell the difference between the X-25E and the X-25M 2.???

I do notice a bit of a lag with the 3 Vertex Turbos on the dual Xeons, but, it isn't much. I think it's the difference between a 940 and dual Xeons...also, maybe, you can tell the difference between 250 mb/sec and 600 mb/sec...

 

[LunarMist]

STR benchmarks are one thing, but actual write performance is another. For example, the Vortex is much slower than the X25-E is in the real world, and even the X25-M G2 is faster. I want a new E series that has higher STR AND higher write I/Os.

 

[ddrueding]

...the Vortex is much slower than the X25-E is in the real world

I would argue that it depends considerably on what world you are from

I'm quite certain that you couldn't tell the difference between a machine with a Vertex and a machine with an X-25E if all I gave you was XP with Firefox and Office, even if they were side-by-side. Even basic Photoshop tasks would be tricky to differentiate without a direct comparison or a stopwatch.

I'm not saying that you (and I) don't do things that make the difference worth the cost, just that calling these things "real world" without qualification is a bit of a stretch.

 

[LunarMist]

Of course it is measured with a stopwatch, unless the softw

 

[ddrueding]

If it requires a stopwatch, it is not significant. Appreciated over time, sure, but not significant.

 

[BingBangBop]

If it requires a stopwatch, it is not significant. Appreciated over time, sure, but not significant.

Are you saying that there is an insignificant difference between a 5 minute mile and a 10 minute mile because a stopwatch was used to measure them?

:wink:

 

[ddrueding]

Are you saying that there is an insignificant difference between a 5 minute mile and a 10 minute mile because a stopwatch was used to measure them?

:wink:

Key word require.

That difference does not require a stopwatch. A stopwatch can be used, but you could tell the difference without.

 

[Santilli]

Just benched the 3 Vertex Turbo "Brick" in the Xeon machine:
Not bad for a 9550 through a PCI-X slot:

 

[Stereodude]

If you say so... Those look pretty crappy to me.

 

[Santilli]

I work with one set going nearly 600 mb/sec, and, the other is the set just posted. The only really big difference is copying files, and a barely noticeable lag with the server, in most of my functions.

I haven't tuned the 7 install on the server to get rid of the built in 'lags'
that the MSFT guys put in. If I feel more energetic today, that can be one of many projects.

The Beast is blinding when copying files from the drives, even from a SATA removeable, to desktop, and back.

In some situations it's faster to copy the large file to the desktop, out of the same folder, then scroll to the folder you want to put it in, copy it back onto the same drive, then it would be if you wait for the scrolling on some of my larger folders...

I should mention that both systems are setup on the 37" Vizio, through different HDMI ports. I am comparing results on the same screen, and, switch between the machines often, using the Server for file storage and downloading, primarily.

The Server is now silent, uses far less energy, and does what it's supposed to...

 

[Stinker]

I got my Vertex SSD running in my Lenovo T7500 C2D Laptop. I used Vista to create the aligned partition and then installed WinXP.

Here's my laptop: http://www-307.ibm.com/pc/support/site.wss/document.do?sitestyle=lenovo&lndocid=MIGR-67707

Is there anything I can do to improve my performance numbers?

 

[Pradeep]

Intel X25-V in RAID-0: Faster than X25-M G2 for $250?

http://www.anandtech.com/show/3618/intel-x25v-in-raid0-faster-than-x25m-g2-for-250

 

[Santilli]

I got my Vertex SSD running in my Lenovo T7500 C2D Laptop. I used Vista to create the aligned partition and then installed WinXP.

Here's my laptop: http://www-307.ibm.com/pc/support/site.wss/document.do?sitestyle=lenovo&lndocid=MIGR-67707

Is there anything I can do to improve my performance numbers?

Try running ATTO on the drive. I don't think HD Tach knows what
SSDs are...or how to rate them.

Also, you should really notice the difference? It's not really about SDTR, but access time.

 

[Stinker]

ATTO does give me numbers I would expect from an SATA I connection.

I just wanted to confirm everything is optimized. Looks like it is.

Thanks!

 

[mubs]

It's not really about SDTR, but access time.

There is a God!

 

[Santilli]

There is a God!

Having both access time and SDTR is STILL better;-)

 

[sechs]

Having both access time and SDTR is STILL better

God is dead!

 

[ddrueding]

He is right, you know...

 

[Santilli]

You know it would kill them to admit it.....

 

[Will Rickards]

Newegg has the 60GB vertex turbo for 170 after rebate (190 before) which is 2.83$/GB. Is this a good deal? I haven't been following the prices too closely.
The 120GB is 320 after rebate (350 before) which is 2.67$/GB.

 

[Santilli]

I paid 309 from Tigerdirect, for 2 30 gig drives, tax 9% and including shipping. The Vertex rebates are a pain, and are approaching scam level, so don't include them in the price.

I have 4 of them, and I'm happy with their data transfer level, and, access time.

 

[Stinker]

ATTO does give me numbers I would expect from an SATA I connection.

I just wanted to confirm everything is optimized. Looks like it is.

Thanks!

After some research, it turns out that Lenovo has limited my R61 (as well as the T61) to SATA I speeds even though the ICH8-M controller will support SATA-II. Lame. Kind of a bummer since it was advertised as SATA II in their literature & I should be able to get the higher transfer rates. No BIOS update is planned to fix it, either. Guess that will be the last Lenovo I buy.

 

[Mercutio]

Kind of a bummer since it was advertised as SATA II in their literature & I should be able to get the higher transfer rates. No BIOS update is planned to fix it, either. Guess that will be the last Lenovo I buy.

I thought it was well-known that T and R61s are 1.5Gbps-only. I might be wrong but that fact was mentioned in a large number of notebook-based reviews for SSDs a couple years back. As I recall Dell and Apple notebooks of that vintage also had a 1.5Gbps limitation.

One of my T61s has a 1st-gen X25-M in it and the benefits are quite significant even if the drive isn't operating at its full capability.

 

[Santilli]

After some research, it turns out that Lenovo has limited my R61 (as well as the T61) to SATA I speeds even though the ICH8-M controller will support SATA-II. Lame. Kind of a bummer since it was advertised as SATA II in their literature & I should be able to get the higher transfer rates. No BIOS update is planned to fix it, either. Guess that will be the last Lenovo I buy.

Sorry to hear that. I had a similar experience with Apple, and never bought another. No excuse.

 

[LunarMist]

SATA 1.5 is like, totally normal on many notebooks.

 

[LunarMist]

Is anyone planning to but a Z-Drive R2 p88?

 

[Stereodude]

Is anyone planning to but a Z-Drive R2 p88?

Right after I win the lottery. :lol:

 

[LunarMist]

I meant buy.

Maybe David needs an upgrade?

 

[Santilli]

Pretty sure DD can budget for that.

By the way, Stinker I started a thread:

http://www.storageforum.net/forum/showthread.php?t=8107

For the same reason. It is better to have both access time AND SDTR, so it doesn't make sense to get another laptop until they catch up with the speeds of the current SSD's.

With the i7 chips now in laptops, that might accelerate the SATA II or 3 adoption...