Micron C400 Benchmarks!!!!

[Burner27]

I owned a Mushkin Callisto Deluxe Enhanced 60GB and 90GB SSD (SF-based drive - about the same as an OCZ V2) and my Intel 120GB SSD and my current 128GB C300 feel much 'snappier' than the Mushkin's did. Now I know that is all relative to each person but I could feel the difference.

 

[Bauss]

IMO, anyone talking about an OS doing that consistently is not really understanding an OS's data flow in the first place. Windows is an operating system and the very nature of an operating system requires it to do much in the background at the same time as a user requests a file.

I don't think anyone is actually denying that the OS does lots of little random reads and writes in the background--especially modern OSs. All they do is process lots of small data without the users knowledge. That said, there's a difference between accessing a few dozen files every second, and accessing tens of thousands of files every second. Remember, the OS might do lots of small random writes, but in the end, it's still designed for HDD storage, as 99% of users still use this. So, an OS will NEVER demand more than what a HDD can do. Not while HDDs are the most popular non-volatile internal storage media. So let's just close the book on needing an SSD for OS background tasks. Very helpful, but not crucial.

With SSD random reads are key and IOPS are king. Most don't seem to understand that.

I think what the recent news in SSDs suggest is that what vendors understand is that, to a user, there's no tangible difference between a drive that does 50MB/s random 4k writes and a drive that does 150MB/s random 4k writes when all software is designed to run on HDDs that only manage a 1-2 MB/s in 4k writes. That's why every SSD vendor has decided to make the big gains in sequential transfers. Not that it's more important, but that it all they can do to make SSDs feel faster.

With my current 6 drive Vertex2 R0 setup, I hear the typical "SSD raids like that aren't going to help speeds for a typical user"... all the time. Well.. I guess seeing is believing then, because everyone who currently owns SSD and has sat in my chair immediately see's otherwise.

Probably, but the difference between a RAID SSD setup and a single SSD setup is not random performance--It's strictly sequential. So If they start copying video files and such, they'll surely notice a difference. But they won't notice one in common file loading. Not one that is tangible, anyways.

Again, I'd never argue the merits of getting an SSD in a system, whether it be a single SSD, or a RAID config. But, all it takes is one SSD to get the relatively instantaneous access times that make them awesome. Any additional SSDs are only padding your sequential throughput.

 

[Burner27]

SuperBiiz (the only retailer that supposedly has the drive for sale) has raised the price $5 for the 128GB C400 for drives that are currently 'out of stock'.

 

[carnage10]

http://thessdreview.com/our-reviews...-–-unexpected-performance-in-a-small-package/

Flamenko, im confused by some of the results on your Vantage HDD suite scores..It shows the 64gb c300 beating the 256gb c300 by over 13,000 points?? Also according to your chart the 64gb c300 and two 64gb c300's in raid 0 score EXACTLY the same? (55880)

 

[Edrick]

Flamenko, im confused by some of the results on your Vantage HDD suite scores..It shows the 64gb c300 beating the 256gb c300 by over 13,000 points?? Also according to your chart the 64gb c300 and two 64gb c300's in raid 0 score EXACTLY the same? (55880)

Yea, as soon as I saw that, I determined that the results in this chart are unreliable at best.

 

[Edrick]

I think what the recent news in SSDs suggest is that what vendors understand is that, to a user, there's no tangible difference between a drive that does 50MB/s random 4k writes and a drive that does 150MB/s random 4k writes when all software is designed to run on HDDs that only manage a 1-2 MB/s in 4k writes. That's why every SSD vendor has decided to make the big gains in sequential transfers. Not that it's more important, but that it all they can do to make SSDs feel faster.

No, vendors understand that large sequential numbers printed on the outside of their box sell much better. It is all marketing. And most people are buying into that.

 

[groberts101]

I don't think anyone is actually denying that the OS does lots of little random reads and writes in the background--especially modern OSs. All they do is process lots of small data without the users knowledge. That said, there's a difference between accessing a few dozen files every second, and accessing tens of thousands of files every second. Remember, the OS might do lots of small random writes, but in the end, it's still designed for HDD storage, as 99% of users still use this. So, an OS will NEVER demand more than what a HDD can do. Not while HDDs are the most popular non-volatile internal storage media. So let's just close the book on needing an SSD for OS background tasks. Very helpful, but not crucial.



I think what the recent news in SSDs suggest is that what vendors understand is that, to a user, there's no tangible difference between a drive that does 50MB/s random 4k writes and a drive that does 150MB/s random 4k writes when all software is designed to run on HDDs that only manage a 1-2 MB/s in 4k writes. That's why every SSD vendor has decided to make the big gains in sequential transfers. Not that it's more important, but that it all they can do to make SSDs feel faster.



Probably, but the difference between a RAID SSD setup and a single SSD setup is not random performance--It's strictly sequential. So If they start copying video files and such, they'll surely notice a difference. But they won't notice one in common file loading. Not one that is tangible, anyways.

Again, I'd never argue the merits of getting an SSD in a system, whether it be a single SSD, or a RAID config. But, all it takes is one SSD to get the relatively instantaneous access times that make them awesome. Any additional SSDs are only padding your sequential throughput.

some of what you say isa true and I've heard the logic many times. But there are many errors in that response.

First off, I've had nearly evry raid concievable both onboard and off(I also use a Adaptec 5805 wih 6 drive HDD in R0) of riad configuartion. Some add latency, and some not so much. Bandwidth will only get you so far otherwise my 6 drive HDD setup would trump the SSD evry time. It comes no where close for an OS volume due to the SSD IOPS and random small file performance. This is pretty well known by most.

What others who have no firsthand knowledge with raiding Sandforce drives do not undrstand is that with wider arrays comes additional headroom for those extra heavy worksessions. Kinda like a 1000 horsepower street engine.. sure most don't need it.. but your sure glad when it's there.

And as far as 1 SSD being no different than 6 for the typical user? Was what I was trying to convey earlier in that it IS quite noticable to everyone who uses this system even when doing simple tasks. I absolutely hate the horrid performance of a single SSD drive on my system. I don't even like booting to my backup 2 drive Vertex 1(Indilinx) array since the difference is night and day. That's how spoiled this current setup has me and no stopwatch is required to see it.

It should be noted that I run premium hardware and also run OC'd at 4.2ghz 24/7 so it may have much to do with the fact that my other hardware imposes little restriction in other areas which may very well let me "see" these gains much moreso than other may on lessor systems. Dunno,.. but the 2 systems I have here with raided SSD are noticably faster with 2 versus 1. Speeds go up for just about everything the wider the array gets. With HDD we get penalized due to cumulative latency as the array gets wider. Not so much with SSD and my 6 drive Sandforce array actually has less latency than 1 single Indilinx. Ultra low latency is one of the reasons Sandforce is so strong as an OS drive.

Here's a 6 drive Sandforce bench collage to show how R0 scales everything with a good chip. If you don't think that can be seen within the OS?.. then you are simply needing to experience firsthand, the power of a premium low latency SSD setup. Many around me had the same opinion as many others here. Until they saw the benefits firsthand. They all use raided SSD now. Now the funny thing is.. all the non-raiders who said this type of performance couldn't be leveraged with typical usage are wanting to step up to SATA3 drives and effectively end up with near raided speeds. lol


Now back to the C400 subject at hand here. Seems to be a lot of single drive users on this forum and I'm wondering who the known raiders are around here? I know many others over at the other big forums using C300's and was wondering if you guys could point me towards any here? Specifically the ones using C300's who may be moving to the C400's anytime soon.

thanks for the time.

 

[groberts101]

Yea, as soon as I saw that, I determined that the results in this chart are unreliable at best.

wouldn't it be logical that comparing raided 64's on SATA2 would be similar to a single 64 on SATA3?

You'd think the score would have gone up a bit anyways but that would help explain it. They should have specified a bit better there for sure.

 

[Bauss]

No, vendors understand that large sequential numbers printed on the outside of their box sell much better. It is all marketing. And most people are buying into that.

That may be true as well, but consider this: Who goes out and buys a $300-$500 SSD and bases their choice on the numbers on the box? Nobody. EVERYONE who buys an SSD at this point knows what they're getting into. Whether their reasoning is misguided or not can be debated, but nobody looks at claimed sequential speeds and chooses based solely on that. Nobody that I know, at least.

That said, It would be equally irresponsible to buy an SSD based solely on random performance, especially considering this:

http://www.anandtech.com/show/4205/the-macbook-pro-review-13-and-15-inch-2011-brings-sandy-bridge/6

The line graph shows 4 SSDs with fairly different designs following what I'd consider to be one of many desktop use patterns--loading tons of apps at once. We see that despite the different designs, all the SSDs seem to perform the same, but all of them are MUCH faster than a HDD.

 

[carnage10]

wouldn't it be logical that comparing raided 64's on SATA2 would be similar to a single 64 on SATA3?

You'd think the score would have gone up a bit anyways but that would help explain it. They should have specified a bit better there for sure.

I'm not sure but i would think dual 64s on sata 2 would still outscore a single 64 on sata 3. Besides that still doesnt explain why the single 64gb outscores the 256gb by so much when they are both tested at sata 3.

 

[Edrick]

EVERYONE who buys an SSD at this point knows what they're getting into. Whether their reasoning is misguided or not can be debated, but nobody looks at claimed sequential speeds and chooses based solely on that. Nobody that I know, at least.

Not everyone. And based on a lot of posts on this and other forums, some reasoning is misguided.

And there are a lot of people who quote sequential speeds as their primary arguement on which SSD to buy. Just look around on various forums.

To be fair, for some people's workloads and uses for such SSD, sequential reads/write may be more important as opposed to people just using an SSD for a OS drive.

 

[Bauss]

Not everyone. And based on a lot of posts on this and other forums, some reasoning is misguided.

And there are a lot of people who quote sequential speeds as their primary arguement on which SSD to buy. Just look around on various forums.

To be fair, for some people's workloads and uses for such SSD, sequential reads/write may be more important as opposed to people just using an SSD for a OS drive.

In these forums, I'd wager that most people make their decisions based on random 4k performance. Hell, that's how this thread got started.

In fact, the whole reason this thread exists is because the OP saw that 4k performance on the C400 was less than the C300. Not that he didn't have a legitimate concern. It's just that in the scope of a Desktop usage model, there are bigger things to worry about.

That's why I initially posted. Because I'm tired of people weighing the merits of an SSD based on one largely moot aspect of their performance.

 

[groberts101]

I'm not sure but i would think dual 64s on sata 2 would still outscore a single 64 on sata 3. Besides that still doesnt explain why the single 64gb outscores the 256gb by so much when they are both tested at sata 3.

good point. I didn't even notice that 256 way down the chart there. lol

If it was grouped up near the top with the others I would think it was a config or hardware bottleneck but obviously something up with the tests there.

 

[Edrick]

That's why I initially posted. Because I'm tired of people weighing the merits of an SSD based on one largely moot aspect of their performance.

Agree with you there!

And I recently upgraded from an Intel G2 to a C300 and I do notice my machine a bit snappier. I gave up on the C400 based on early reviews.

 

[groberts101]

That may be true as well, but consider this: Who goes out and buys a $300-$500 SSD and bases their choice on the numbers on the box? Nobody.

LOL. You've never been to the Sandforce frums across the different mfgrs sites then. The "why my drive slow?" posts are funny sometimes and really points to the general publics lack of understanding and need to research before they buy an SSD.

People buy the drive based on ATTO scores and then get mad or want their money back when the drive shows slower results when using incompressible data streams(AS SSD/CDM3). Is the law of averages with Sandforce, that's for sure.

Are you consdiering a 6G upgrade Bauss? If so will it be the C400?

 

[Edrick]

I almost bit on the 550MB/s speeds of the Vertex 3.......almost. Then I realized I wanted a drive that would last more than a month and not run like a slug when using incompressible data.

 

[Bauss]

I almost bit on the 550MB/s speeds of the Vertex 3.......almost. Then I realized I wanted a drive that would last more than a month and not run like a slug when using incompressible data.

Solid choice IMO. My daily workflow includes copy large HD videos (large incompressible copies). The V3 seems to not be at top form there. Also, it's so untested at this point that you never know that kinda crap will turn up once you've had it for a few months.

 

[groberts101]

first off.. these drives don't "die". They panic lock due to incompatible bios implementations(most often with laptops), bad flashes, or sleep transition issues(much of which seems to be fixed with new firmware). Other mfgrs have the same issue as well and has nothing to do with the vendor since from a hardware standpoint much is the same. It's a known controller problem and blame needs to put squarly on Sandforces shoulders where it belongs.

And then.. wanna bet?

the throttling has been changed quite a bit now. I know for a fact as I have one to beta test. looky see



Little background to give some perspective. I had to literally keep pounding the drive with back to back tests WELL beyond it's 240GB capacity with CDM3 using 9 x 4000MB tests(about 140 gigs per run) to get it to throttle. Was very tough to get these screens due to the quick lifting of the throttle that the new controller/firmware allows. These are back to back runs and you can clearly see where the controller has lifted the throttle suddenly and without logic. The SF2281 has more of an "on demand" algorithm now. Good to finally see as that was my biggest complaint with these drives.

As I always tell people trying to judge Sandforce on numbers... the drive operates on sort of a "floating point of performance" due to compression capability. Use one type of data and it writes at xx speed. Change back to another and it changes just as quickly. OS usage is more towards ATTO numbers than CDM3 or AS SSD. Although I have yet to find the real data twin to ATTO since it appears to be compressible like no other data I have found to data. AS/CDM is closer to video writes. Either way you judge it, OS data is in between the best case and worst.

Oh yeah, SATA 2 there. here's sata3

Uploaded with ImageShack.ushttp://img16.imageshack.us/i/smaximusboardv3asssd.jpg/

 

[Bauss]

LOL. You've never been to the Sandforce frums across the different mfgrs sites then. The "why my drive slow?" posts are funny sometimes and really points to the general publics lack of understanding and need to research before they buy an SSD.

People buy the drive based on ATTO scores and then get mad or want their money back when the drive shows slower results when using incompressible data streams(AS SSD/CDM3). Is the law of averages with Sandforce, that's for sure.

Are you consdiering a 6G upgrade Bauss? If so will it be the C400?

Hah I guess not. I didn't know there were so many people out there with more money than brains..

As for my 6G upgrade, already done! I have a 2k11 MBP 15 with an Intel 510 256GB. I saw the huge write speeds of the 510 (especially with incompressible data) and took the plunge. Also, Intel's reliability with it's former SSDs is second to none.

So far I haven't been able to slow it down at all. Most of what I really push the SSD for is 're-containing' my HD movies from MKV to M4V for iTunes. It's an operation that essentially requires pulling out the video stream, transcoding audio for AC3 (from DTS), then re-packaging the file in M4V. It used to take my old HDD 3-4 minutes to copy the H.264 video streams, but now it takes about 1, and I'm stuck waiting for the audio to finish transcoding.. Amazing to watch.

Sometimes, I even load all the apps on my Mac up and watch them all spring up within 2-3 seconds or so. That's about the only time I really see this thing touch 3-4k IOps.

 

[groberts101]

sounds sweet.

I run mine for vid/gfx/website editing with Adobe's stuff. Even use it for the kids PSP rips/encodes since it's such a beast.

I'm a hotrodder for years and all this stuff is right along the same lines as that. "Speed costs money.. how fast do you want to go?"

 

[taltamir]

Honestly ... so many people convinced SSD's are SO good and who actually bought one.

Every single one of the SSD advocates in this thread owns one.

You people should really consider THIS:

most if not all of your io activity IS sequential
most if not all of your io activity IS "big" -> as in not fetching just one 4k block

This is false.

 

[Dadofamunky]

.Dunno,.. but the 2 systems I have here with raided SSD are noticably faster with 2 versus 1.

Hmmmmm... (*Glances quickly at NewEgg to see how much another 128GB C300 costs*)

 

[LamTek]

$200 @ Superbizz...

Debating jumping on 2 due to all this proclaiming love for RAID. Do you lose trim support if you go RAID on SSD's? Any other feature losts?

You probably would want the coupon code as well:
http://bensbargains.net/deal/187889/

 

[Old Hippie]

Do you lose trim support if you go RAID on SSD's?

Yes.

 

[Bauss]

$200 @ Superbizz...

Debating jumping on 2 due to all this proclaiming love for RAID. Do you lose trim support if you go RAID on SSD's? Any other feature losts?

You probably would want the coupon code as well:
http://bensbargains.net/deal/187889/

Keep in mind, the only benefits you'll see from a RAID array is the massive amount of sequential performance you'll have available. Also, larger random writes will probably see a boost. That said, if your workflow doesn't REALLY take advantage of all that, you won't notice a difference at all in common operations.

I'd recommend going with 1 SSD for now. If you're coming from a HDD, you'll already see a huge improvement. But if you feel you need more for specific operations, then go for a 2nd.