Sign in to follow this  
GPett

Anandtech review of Super Talent SSD

Recommended Posts

Anandtech.com Supertalent SSD review has beat Storage Review to the punch. I have been waiting for reviews like this from SR for a long time. Solid State Drive (SSD) are here. Sure they are not feasable for most solutions. But at least 3 manufacturers have them available. SanDisk, Samsung, and now Supertalent all have SSD out there.

I am really interested in the consumer versions of SSD coming down the pipe. How many of us need 1TB of storage anyways? I am perfectly happy with 30GB.

Anywho, good job Anandtech and lets get these reviews going SR!

Share this post


Link to post
Share on other sites

Now we just need to figure out how to turn off all caching and logfiles that windows generates to get rid of extraneous writes to the SSD to prolong its life. Hmm.. I am liking the thought of this: No log files, no inernet caching, 4-8GB of main memory so you can disable the page file. I am sure you could have a nand flash drive for main storage last a long time if you can eliminate unneccesary writes from the OS.

Share this post


Link to post
Share on other sites
Now we just need to figure out how to turn off all caching and logfiles that windows generates to get rid of extraneous writes to the SSD to prolong its life. Hmm.. I am liking the thought of this: No log files, no inernet caching, 4-8GB of main memory so you can disable the page file. I am sure you could have a nand flash drive for main storage last a long time if you can eliminate unneccesary writes from the OS.

Yes, rewrite Windows code. The OS wont properly work without pagefile and cache.

Share this post


Link to post
Share on other sites

Sure, I have turned off pagefile and internet cacheing many times. Its all of those log files I want to know how to turn off.

Share this post


Link to post
Share on other sites

Windows as an OS is not designed to run without a pagefile. If you're running Windows, leave the pagefile on.

Just run enough memory that you're not hitting the pagefile very often. Again, Windows is designed to run with a pagefile. Leave it on!

Share this post


Link to post
Share on other sites

The Windows pagefile thing was over and done with since 2Ksp1. No need for one anymore. Discussion has been had on these very forums with supporting MSDN links. Page files are valuable in scenarios where the performance gained by the system cache is of greater benefit than holding stale applications in RAM. As system cache size increases, the performance returns diminish. For the typical desktop profile, 2GB of RAM usually justifies disabling the page file. For gaming systems, the sweet spot is around 3GB (unless you play Oblivion). Like all other tweakable kernel parameters, pagefile size and availability (enabled vs, disabled) should depend on the memory profile of the applications that you choose to run.

Summary... Unless you know what your typical usage profile looks like in perfmon, leave the pagefile settings close to default. If your system cache is > 50% your available memory size (after taking into account how much of the page file is being used), disable the pagefile.

Thank you for your time,

Frank Russo

Share this post


Link to post
Share on other sites

thats why you buy yourself an iram - put all the caching stuff that you can on there, i've got 2gb of ram lying around so will be buying one for that reason alone.

Share this post


Link to post
Share on other sites

The stupid "writing to flash kills it" myth yet again.

Even my Sansa MP3 has wear leveling and ECC. Every name brand does too.

The FAT and root dir get rewritten thousands of times. These things last years.

Share this post


Link to post
Share on other sites

I've been using Vista X64 without a pagefile since the day I installed it without any issues at all. I'm not a 'power user', but mainly surf porn, play CS:Source & re-encode movies. The system boots to the desktop using about 950MB of memory, leaving me 3GB to play with. I noticed a local shop is selling my exact Corsair 1GB sticks of DDR667 for $CAN45. I'd buy another 4 sticks if my board had the sots!

Share this post


Link to post
Share on other sites

And just to touch on what Frank said, I do believe there maybe be a few programs (like Photoshop) that insist on finding a Windows pagefile, but I'm not totally sure on this. Someone else can chime in correct me on this.

Share this post


Link to post
Share on other sites

And for those apps that require one, assuming you have plenty of free RAM, you can always put the pagefile on a RAMdisk. Faster than an IRAM (since it's not constrained by the SATA interface), and the pagefile doesn't have to be non-volatile (though most good RAMdisk programs will write the data back to the hard drive when you shut down).

This was also discussed years ago, and the conclusion then was that you're better off just having the extra RAM. But I think we're getting to the stage, especially with 4 GB + desktop systems becoming more viable, where a RAMdisk is a valid use of excess RAM, especially when we don't expect to use the pagefile much anyway.

Share this post


Link to post
Share on other sites
But I think we're getting to the stage, especially with 4 GB + desktop systems becoming more viable, where a RAMdisk is a valid use of excess RAM, especially when we don't expect to use the pagefile much anyway.

I think a using a RAM disk is always a sign something else is wrong or could be done better.

Now we just need to figure out how to turn off all caching and logfiles that windows generates to get rid of extraneous writes to the SSD to prolong its life. Hmm.. I am liking the thought of this: No log files, no inernet caching, 4-8GB of main memory so you can disable the page file. I am sure you could have a nand flash drive for main storage last a long time if you can eliminate unneccesary writes from the OS.

Disabling 'internet caching' is very bad for browsing performance, so why do it?

I am really interested in the consumer versions of SSD coming down the pipe. How many of us need 1TB of storage anyways? I am perfectly happy with 30GB.

I'm not. Anyway, 30 gbyte is still way too expensive compared to a normal HDD.

Share this post


Link to post
Share on other sites
I think a using a RAM disk is always a sign something else is wrong or could be done better.

True. In this case, it's the longevity of flash media under frequent rewrites that we're working around.

I would imagine temp files are less of a problem than pagefile, and even then with modern (wear levelling), large flash devices, the rate at which individual locations are rewritten would be pretty low. Inversely proportional to the free space on the drive, in fact. And it would be mitigated by the amount of physical RAM in the system reducing the activity hitting the pagefile.

I wonder what the typical rate is at which data is written to the pagefile? Assuming the system isn't idle, it could measure in the low gigabytes per hour. Call it 2 GB/hour, and assume you have 10 GB of capacity (including the existing pagefile) for wear leveling purposes. So every 5 hours of use, each spare bit of space gets written once. Suppose each bit can handle 100,000 rewrites, on average. So with 24/7 active pagefile usage, you'd kill your flash drive in 500,000 hours. I make that about 57 years. So even if I'm an order of magnitude out on my estimates, it'll still last 5 years of constant pagefile use.

Most likely, it's not a problem at all, but for the paranoid or those that need to maximise the longevity of their SSD, RAM disks are there as a fallback.

Share this post


Link to post
Share on other sites

I pointed out way back here that you only need some files from XP Embedded to stop this silly Windows behavior once and for all. BartPE for USB flash devices works similarly, but of course the Preinstallation Environment OS times out. Embedded doesn't. :)

Share this post


Link to post
Share on other sites
I would imagine temp files are less of a problem than pagefile, and even then with modern (wear levelling), large flash devices, the rate at which individual locations are rewritten would be pretty low. Inversely proportional to the free space on the drive, in fact.

I don't think the free space bit is correct.

Share this post


Link to post
Share on other sites
Guest 888
The FAT and root dir get rewritten thousands of times.

I have also wondered about how does the SSD handle that FAT which is almost constantly under rewriting. Does the SSD move the physical location of FAT all the time over the SSD? Probably yes. Otherwise it can't live long.

Share this post


Link to post
Share on other sites
The FAT and root dir get rewritten thousands of times.

I have also wondered about how does the SSD handle that FAT which is almost constantly under rewriting. Does the SSD move the physical location of FAT all the time over the SSD? Probably yes. Otherwise it can't live long.

Probably the same way as any other block that gets rewritten a lot. That's what wear levelling is all about, right?

Share this post


Link to post
Share on other sites
The FAT and root dir get rewritten thousands of times.

I have also wondered about how does the SSD handle that FAT which is almost constantly under rewriting. Does the SSD move the physical location of FAT all the time over the SSD? Probably yes. Otherwise it can't live long.

You seem to have the odd notion that the logical data position on a flash drive has any physical meaning at all.

The drive has its internal, low level filesystem with wear leveling/ect. _Everything_ that gets exposed to ATA is just payload, whether its FAT or anything else.

Share this post


Link to post
Share on other sites
The FAT and root dir get rewritten thousands of times.

I have also wondered about how does the SSD handle that FAT which is almost constantly under rewriting. Does the SSD move the physical location of FAT all the time over the SSD? Probably yes. Otherwise it can't live long.

You seem to have the odd notion that the logical data position on a flash drive has any physical meaning at all.

The drive has its internal, low level filesystem with wear leveling/ect. _Everything_ that gets exposed to ATA is just payload, whether its FAT or anything else.

Please remember that logical fragmentation is 'almost' as important as physical fragmentation. Granted, it is most important in respect to memory where applications want to commit large contiguous blocks. I'd really have to think about whether this applies to SSDs or not.

Frank

Share this post


Link to post
Share on other sites
Please remember that logical fragmentation is 'almost' as important as physical fragmentation. Granted, it is most important in respect to memory where applications want to commit large contiguous blocks. I'd really have to think about whether this applies to SSDs or not.

Frank

Memory fragmentation is only an issue with virtual memory in 32-bit systems. In 64-bit systems the address space is so large that I think it's not an issue.

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
Sign in to follow this