A Whole Bunch of Solid State Disk Testing

Status
Not open for further replies.
JasonSTL739 said:
Single 7200RPM drive to a Vertex is a gigantic difference to me in real world usage. Worth every dollar. To the point where I swapped every machine I have over to SSD for OS/programs/swaps, laptops, two systems, and an entire ESX server (6X 64GB X25-E's at the moment, running vSphere 4.0 with a set of Win2008 VMs)

There is a "crispness" that the benchmarks cannot show. You have to feel it.
Click to expand...

I am currently using Raptor 150GB 10K rpm. I have read benchmarks, and agree that SSD is faster. However, in real world, is it really faster, that you can see the difference? Been wanting to get SSD, but still unsure...

Chris
 
sequoia said:
I am currently using Raptor 150GB 10K rpm. I have read benchmarks, and agree that SSD is faster. However, in real world, is it really faster, that you can see the difference? Been wanting to get SSD, but still unsure...

Chris
Click to expand...

It's REALLY faster. :)
 
AppaYipYip said:
Until it starts to fill up, at which point it suffers greatly.
Click to expand...

What brand SSD are you experiencing this on?

I run the 120gb vertex in my laptop with 100GB or so on it all the time and while benches will show a difference actuall usage does feel much different.
 
sequoia said:
I am currently using Raptor 150GB 10K rpm. I have read benchmarks, and agree that SSD is faster. However, in real world, is it really faster, that you can see the difference? Been wanting to get SSD, but still unsure...

Chris
Click to expand...

I came from four 15k SAS disks for my OS disk, and I feel the ssd's roast those. Single raptor to single SSD there is no contest. SSD will bury the raptor (assuming you get a vertex or intel based disk)
 
JasonSTL739 said:
I came from four 15k SAS disks for my OS disk, and I feel the ssd's roast those. Single raptor to single SSD there is no contest. SSD will bury the raptor (assuming you get a vertex or intel based disk)
Click to expand...

With 4k writes, it would take something like 10-15 15k SAS drives to compete with the speed, and probably 100+ to catch up with IOPs. :p
 
aznx said:
With 4k writes, it would take something like 10-15 15k SAS drives to compete with the speed, and probably 100+ to catch up with IOPs. :p
Click to expand...

Yup.

1 SSD = 12 15k SAS disks, roughly.

I benched 48 Seagate 15k 36GB SAS Disks in external enclosures on PERC6's with a TPC-C OLTP-style benchmark.... compared to four X25-E's on internal RAID.... and the four little drives beat up on the SAS disks. Fun stuff.
 
aznx said:
Nope. You're giving quite the misinformation..or you've been ill informed.
Click to expand...

Does the SLC/MLC have anything to do with this possibly? I think MLC's suffer more from capacity speed issues than SLCs do... Might be the difference of opinions in this area...
 
hmchk said:
Does the SLC/MLC have anything to do with this possibly? I think MLC's suffer more from capacity speed issues than SLCs do... Might be the difference of opinions in this area...
Click to expand...

I have both in different machines, at capacity; and the only way you'd think that is by reading reviews that aren't testing correctly.... or are using some of the really crappy SSDs or old firmware....

If you buy Vertex's, or a version of the X25-M or X25-E you will be fine.
 
JasonSTL739 said:
I came from four 15k SAS disks for my OS disk, and I feel the ssd's roast those. Single raptor to single SSD there is no contest. SSD will bury the raptor (assuming you get a vertex or intel based disk)
Click to expand...

WOW...
Now if the price would go down so I can afford one of these... I probably end up with one soon though....

[H] is making me broke :(
 
For normal desktop use you probably won't notice a thing. It gets even more marginal with Windows 7 since it aggressively caches/pre-fetches.
 
Obviously he has not used a good SSD or he wouldn't have said that. Having your desktop and apps respond instantly all the time feels like a 600% speed boost over platters.
 
Yes I actually have 4 SSDs; in my sig: X25-E 64GB, X18-M 80GB, OCZ Vertex 120GB, and OCZ Vertex EX 60GB.

I have 5 OSs in my boot loader (for redundancy, benchmarking, re-installation, etc) so I have a fairly good experience with SSD performance. Many of the SSD users on Anandtech have had similar experiences during SSD testing.

That said, I definitely do not mean Windows XP. My test platforms are Vista and primarily 7 RC.

EDIT: I actually have 5 SSDs, with my Dell M1530 (Samsung 128GB) although I can't use it for any useful and/or direct comparisons because a laptop is a closed platform.
 
Last edited:
I was using 1 x single core Opteron at 2.4GHz, when I uncompress a 8GB ISO file using Win RAR 3.4, it takes about 7 to 10 min. approx.

Then to uncompress that ISO file into all of its files, it takes about 22 min.

Upgraded to 2 x dual core Opteron 2.8GHz, the above 2 operation doesn't change any speed

If the uncompressing is done at a SSD drive, would it make a difference, and if so, by how much?
 
No. Decompression is more or less a sequential write operation. You aren't going to extract any faster on an SSD than on a 4200rpm hard drive. I just tested unzipping a 1GB file on my SSD vs my old a$$ Thinkpad. It was more or less identical. My guess is the unrar algorithm its a single-threaded application.
 
WinRAR is dual-core aware, and depending on what you are extracting you'll either peg the CPU or your drive. You can use perfmon to figure out which.
 
astrallite said:
Yes I actually have 4 SSDs; in my sig: X25-E 64GB, X18-M 80GB, OCZ Vertex 120GB, and OCZ Vertex EX 60GB.

I have 5 OSs in my boot loader (for redundancy, benchmarking, re-installation, etc) so I have a fairly good experience with SSD performance. Many of the SSD users on Anandtech have had similar experiences during SSD testing.

That said, I definitely do not mean Windows XP. My test platforms are Vista and primarily 7 RC.

EDIT: I actually have 5 SSDs, with my Dell M1530 (Samsung 128GB) although I can't use it for any useful and/or direct comparisons because a laptop is a closed platform.
Click to expand...

Maybe since you use SSDs so often that you've forgotten how "slow" using a single Raptor can be?

I'm not saying this to be glib, it's all about perception. It's like silencing your PC, if you're used to a really quiet PC, another person who thinks they have a quiet PC you might think is really loud because your PC is whisper quiet. Now that I've gotten used to whisper quiet, anything that's louder than that but still fairly quiet I perceive to be quite loud.

I've been mulling over this, too, using a 300GB VR in daily use and thinking about getting the 160GB X25-M but would want my perceived gains to be noticable to justify the expense.
 
I went from a Raptor to an intel x-25m and the difference is night and day, let alone the heat and noise reduction in the case. (150gig, can't get the dam P/N to show up right)

I don't know what more you want to know, but there's two classes of PC's out there now in terms of end user experience. Those with high quality SSD's and those without. The problem is once you get used to the performance of an OS on SSD, going back to the traditional drives sucks (as you can see by people saying 'oh god I just had to put a ssd in every machine after using one on the main pc' etc).
 
soekev said:
Maybe since you use SSDs so often that you've forgotten how "slow" using a single Raptor can be?

I'm not saying this to be glib, it's all about perception. It's like silencing your PC, if you're used to a really quiet PC, another person who thinks they have a quiet PC you might think is really loud because your PC is whisper quiet. Now that I've gotten used to whisper quiet, anything that's louder than that but still fairly quiet I perceive to be quite loud.

I've been mulling over this, too, using a 300GB VR in daily use and thinking about getting the 160GB X25-M but would want my perceived gains to be noticable to justify the expense.
Click to expand...

A VR is not slow. And I use it regularly while I am doing OS upgrades on the drives, or flashing. "Generation 2" SSDs are IOPs monsters and are in another world with multithreaded (high queue depth) reads. I'm talking about basic personal desktop use. It's not going to suddenly load Word in 0.1 seconds when it takes 2-3 seconds before. Paint brush is not going to suddenly go from 2 seconds to 0.1 seconds. Opening folder explore is not going to get faster. Windows folder gallery is not going to load any faster. AIM not going to load faster. The reason for this is simple. Windows 7 (and I assume Vista SP2) is designed with extremely aggressive pre-caching and indexing.

For business, graphical, production oriented, and any useage that requires multithreaded use (multiple application load) no doubt it will be faster typically. A lot of tasks are CPU limited. Take level load in Far Cry 2 for example. The difference between an X25-E and a VR300 is a fraction of a second, well within error range.

For a laptop, the difference between an SSD and a typically 7200rpm hard drive is a big gulf however. You are far more likely take your work with you than sit on your desk playing games or surfing. When I come out of hibernate I'll have multiple instances of Word, Frontpage, Excel, etc. Loading all of these simultaneously is bleeding fast.
 
Last edited:
astrallite said:
A VR is not slow. And I use it regularly while I am doing OS upgrades on the drives, or flashing. "Generation 2" SSDs are IOPs monsters and are in another world with multithreaded (high queue depth) reads. I'm talking about basic personal desktop use. It's not going to suddenly load Word in 0.1 seconds when it takes 2-3 seconds before. Paint brush is not going to suddenly go from 2 seconds to 0.1 seconds. Opening folder explore is not going to get faster. Windows folder gallery is not going to load any faster. AIM not going to load faster. The reason for this is simple. Windows 7 (and I assume Vista SP2) is designed with extremely aggressive pre-caching and indexing.

For business, graphical, production oriented, and any useage that requires multithreaded use (multiple application load) no doubt it will be faster typically. A lot of tasks are CPU limited. Take level load in Far Cry 2 for example. The difference between an X25-E and a VR300 is a fraction of a second, well within error range.

For a laptop, the difference between an SSD and a typically 7200rpm hard drive is a big gulf however. You are far more likely take your work with you than sit on your desk playing games or surfing. When I come out of hibernate I'll have multiple instances of Word, Frontpage, Excel, etc. Loading all of these simultaneously is bleeding fast.
Click to expand...

Folders with a large number of files (20,000+) do open quicker for me..in OS X
 
astrallite said:
No. Decompression is more or less a sequential write operation. You aren't going to extract any faster on an SSD than on a 4200rpm hard drive. I just tested unzipping a 1GB file on my SSD vs my old a$$ Thinkpad. It was more or less identical. My guess is the unrar algorithm its a single-threaded application.
Click to expand...

so besides a faster boot up, I take it that I don't gain anything by buying a SSD drive then? Because I use Excel, Wordperfect, and Firefox. Most of the time, I just cut & paste text from firefox to wordperfect.

My WP data files are saved at a Flash USB drive.
 
JasonSTL739 said:
I'd buy a 120GB Vertex if a single drive - the extra cache does make a difference.
Click to expand...

I know this post is several months old, but those new to SSDs should note that cache isn't the primary determining factor of performance, though additional cache certainly helps. The controller should be considered first and foremost (and the Vertex has a good one!), with other factors such as amount of cache and memory speed/type coming second.

The only reason I mention this is because drives such as the OCZ Summit have been released with larger amounts of cache, yet it doesn't necessarily make them faster than "premium" drives such as the Vertex and X25-M. I know you didn't say otherwise and your post might have been before the Summit was released. I just read this thread in its entirety and think that it was an excellent read. No reason to let it die! Let the testing continue with newer, different drives...I'm sort of interested in finding out more about the new Samsung controller.
 
This thread should have stayed dead. Benchmarks from April 2009 are no longer relevant. The SSD market has been evolving at a breakneck pace.
 
Status
Not open for further replies.
Back
Top