New Samsung SSDs?

[FishAk]

A few corrections here, Fish:
- The C300 drives are using a Marvell controller, not JMicron.

Ya, you're absolutely right. I get these two mixed up. It's probably due to my bias against them. I don't think either controller is all that great.

- The SF drives *do* take a performance hit if you fill them up, look at Anand's graphs in his articles.
- The ability of the SF controller to use free space beyond the over-allocated amount has nothing to do with compression. The Intel controller also uses all available free space just like the SF controller does.

I believe you're referring to the graph on this page.


If you read the actual text that accompanies the graph, you will note that the SF drive was filled with completely random bits with a custom build of Iometer in order to defeat the compression algorithm of the SF controller. I'm sure you would agree that under normal use, data written to the drive is much more compressible than the truly random bits used for this test.


I have said this before, for which I am now called a SandForce fanboy, but I believe compression has a huge impact on the performance of a full SF drive. Since you have read the AnandTech articals on this subject, you know that when the SF controller saves disk space through compression, it doesn't report any savings to the operating system. The controller is the only entity that knows about this extra, saved space. Because, like Intel, the controller uses all available empty space as spare area, I don't see how this saved space would go unused by the controller. I simply can't imagine the guys at SF being so slow-witted that they would allow this saved nand to sit on the drive, serving no purpose, when it can be used to improve performance.


I believe the ability of the SF controller to use empty space beyond the over-alocated amount is very much influenced by compression. The Intel drive also uses all unused space, but it doesn't compress, so it has only the default reserve space when the user space is filled.

 

[poohbear]

I would just refer you to these graphs and encourage you to read Anand's explanations of each of the tests in his Storage Bench. You can't look at random reads in isolation and expect that to carry the day when other stats are SO WEAK. Please note that the 64GB Crucial drives are not on these charts, only the >= 128GB ones, so the numbers for the 64GB would be even lower.

http://www.anandtech.com/show/3812/the-ssd-diaries-crucials-realssd-c300/6

Sometimes I wonder how many people on these forums actually read Anand's articles...

other stats are so weak??? lol u use hyperbole to try and put the crucial SSD down. There's nothing "so" weak about this drive.

and fyi i already read the article on the c300 (the c300 128gb and vertex 2 are neck to neck, not sure why u even referenced that page?), infact i've read every review on the c300 on the net before purchasing it, i did my research thoroughly and im very happy w/ it.

THE AVERAGE USER WILL NOT NOTICE A DIFF BETWEEN THE C300 AND THE SANDFORCE DRIVES ON SATA2. in Sata3 it surpasses the sandforce drives. you go for the best price/performance ratio, and since the c300 is cheaper, its a no brainer. If u have win7 this drive is blazing.

 

[redlinez33]

Thanks all, bought an SSD. Hoping it works out for me

 

[LokutusofBorg]

I believe you're referring to the graph on this page.

If you read the actual text that accompanies the graph, you will note that the SF drive was filled with completely random bits with a custom build of Iometer in order to defeat the compression algorithm of the SF controller. I'm sure you would agree that under normal use, data written to the drive is much more compressible than the truly random bits used for this test.

Yes those are the graphs I was talking about. You make a good point about Anand having to use artificially cryptic random data to fill the drive up, and the unused space compression frees up. I'll give you that.

I believe the ability of the SF controller to use empty space beyond the over-alocated amount is very much influenced by compression. The Intel drive also uses all unused space, but it doesn't compress, so it has only the default reserve space when the user space is filled.

You just made it sound like compression is what enables the ability to use all free space for wear leveling, etc. which is not the case. It's all about the controller logic, which the Intel controller has as well without compression.

 

[LokutusofBorg]

other stats are so weak??? lol u use hyperbole to try and put the crucial SSD down. There's nothing "so" weak about this drive.

and fyi i already read the article on the c300 (the c300 128gb and vertex 2 are neck to neck, not sure why u even referenced that page?), infact i've read every review on the c300 on the net before purchasing it, i did my research thoroughly and im very happy w/ it.

THE AVERAGE USER WILL NOT NOTICE A DIFF BETWEEN THE C300 AND THE SANDFORCE DRIVES ON SATA2. in Sata3 it surpasses the sandforce drives. you go for the best price/performance ratio, and since the c300 is cheaper, its a no brainer. If u have win7 this drive is blazing.

I doubt you realize it but you come off very strongly as someone trying to justify their purchase. You've made it clear you own a C300 and you love it. Blah blah blah. I own a Vertex 2 and love it. Yet I recommend different drives under different circumstances. To say that the C300 is the be-all-end-all of drives is naive at best.

I have a friend that works for Micron. I live just over the mountain from the IMFT plant here in Utah. I buy Crucial RAM exclusively for all the computers I build. If the C300 drives deserved it I would trumpet them. They don't, so I don't.

Give it a rest.

 

[LokutusofBorg]

Thanks all, bought an SSD. Hoping it works out for me

So which one did you buy?

 

[FishAk]

You just made it sound like compression is what enables the ability to use all free space for wear leveling, etc. which is not the case. It's all about the controller logic, which the Intel controller has as well without compression.

The reason I think compression plays a part is due to the empty space provided by the compression.

Intel doesn't compress, so when user space is filled to 100 percent, only the default over-provisioning is available for the controller logic to work with.

SandForce compresses, so when the user space is filled to 100 percent, both the default over-provisioning, and the space saved due to compression is available for the controller logic.

Not only that, but the compression is substantial. Close to 50% in some instances, and probably averaging 20 or 30 percent in most. That's a lot of empty nand.

Of coarse, this is extreme example, as- if a drive is filled to 100 percent, no other data can be written to the drive- hence no need for wear leveling at the 100 percent level. Back this off to 98 or 99 percent, and now you have a working drive, that needs to counter write amplification.

So which one did you buy?

+1 I'd like to know to.

 

[flamenko]

I like that thought Fish. perhaps that can explain why the OWC drives do not drop- in performance when completely full and,even better yet, why the pro (7%OP) and the RE (28%OP) still have the same benchmarks when full. I have asked around and it seems to be suggested that the additional overprovisioning will benefit only in the life of the drive.

 

[poohbear]

I doubt you realize it but you come off very strongly as someone trying to justify their purchase. You've made it clear you own a C300 and you love it. Blah blah blah. I own a Vertex 2 and love it. Yet I recommend different drives under different circumstances. To say that the C300 is the be-all-end-all of drives is naive at best.

I have a friend that works for Micron. I live just over the mountain from the IMFT plant here in Utah. I buy Crucial RAM exclusively for all the computers I build. If the C300 drives deserved it I would trumpet them. They don't, so I don't.

Give it a rest.

when did i ever say it was the be all end all of all drives? dude, u love hyperbole. read all my posts and quote me once saying that. I said THE AVERAGE USER WONT NOTICE A DIFFERENCE BETWEEN THEM (and u even quoted me saying that!! lol). how on earth is that saying its the be all end all? all my posts in this thread can be reread by anyone, including u, and i've never suggested that. drop the hyperbole already. i just corrected u saying its "performance hit" in the smaller drive is only in write speeds and that wouldnt be noticeable by the average user (especially on a small drive not meant for frequent writing), pointed out some exaggeration u use when speaking of its "lower" performance, and u all of a sudden got defensive. just relax man.

 

[redlinez33]

I ordered one I didnt even talk about, the OWC Mercury Extreme Pro. Paid tad extra for what I have read is good customer service, and hopefully a great drive.

Should have by Friday....

 

[redlinez33]

Got my drive, so far I am loving it. Fast boot, fast app loading. Only use 32GB of a boot drive so 60GB was fine, even though it was really only 55.8GB..


Thanks to everyone with opinions

 

[flamenko]

Nice choice!!!!

 

[StinkyPinky]

SSD was a massive performance increase for me. I'm amazed people have top of the line hardware but still use ye olde 7200rpm drives to host their OS.