Sign in to follow this  
Brian

Crucial C300 Discussion

Recommended Posts

It doesn't run fine on SATA II
Which board/controller is in SR's testbed again? Dell XPS something with an Intel ICH9? Does SR address that fact in the review (differing driver revisions are also significant factors, as both TR and Anandtech have exhaustively addressed)? Other results around the 'net beg to differ...

Techreport uses an Gigabyte GA-P55A-UD7 with the P55 PCH (for 3Gbps) and the Marvell 9128 (for 6Gbps), and they show varying results-- but the RealSSD C300 is consistently fast on both.

Per Techreport, HDTune shows ~265MB/sec reads on 6Gbps and 217MB/sec on 3Gbps (Intel), while ~151MB/sec writes (6Gbps) and ~187MB/sec wries (3Gbps). Yes, 3Gbps writes are actually faster.

Results bounce back and forth, at least in TR's review, between which interface is faster. Some differences are huge (190MB/sec in File Copy Test - Create - MP3 - on 3Gbps, vs. 95MB/sec on 6Gbps).

Also... if SR feels that some of the lead is due to the 512MB cache on the MegaRAID 9260 used, why not hook up another SSD to it to see how it benchmarks, so that relative comparisons can be done on the same controller? As it is right now SR's 6.0Gbps benchmarks are not exactly fair nor directly comparable. Most other sites aren't quite either, but SR has by far the largest difference as the 6Gbps controller used is a 512MB-backed card rather than a more conventional implementation.

Share this post


Link to post
Share on other sites

Our testbed currently uses the ICH10R. In a followup piece we are centering around the LSI 9260 we will be addressing the increase it gives most drives. Most SSDs and HDDs see about a 10-15% bump within reason when routed through the RAID card. In the tests we ran on the SF SSDs we did see an increase in some areas, with the SF gaining on the productivity test and a flip flop on the HTPC test (close between the C300 and SF).

In a perfect world once we find a plain 6gbps interface we would love to readdress the benchmarks with all drives on equal terms (no beefy cache) but at the time of the review the only card that worked for us was the RAID card.

Share this post


Link to post
Share on other sites

Hmm. Well, the point of a standard testbed is a standard testbed... you could just standardize on the MegaRAID 9260. :P Although all the cache is indeed an issue to address.

Lots of others have chosen to standardize on whatever's onboard typical systems today to make sure they're more representative of what typical enthusiasts are deploying (such as the Marvell 9128..), as plunking down $500 for a separate RAID card is a little beyond what most would do.

Share this post


Link to post
Share on other sites

In this case, the alternative would have been to say the drive isn't compatible with SATA II, or at least, performs poorly in that case. While that's partially true, it's not entirely accurate. This does underscore however how the SSD space still has some room for maturity.

Share this post


Link to post
Share on other sites

Hmm. Well, the point of a standard testbed is a standard testbed... you could just standardize on the MegaRAID 9260. :P Although all the cache is indeed an issue to address.

Lots of others have chosen to standardize on whatever's onboard typical systems today to make sure they're more representative of what typical enthusiasts are deploying (such as the Marvell 9128..), as plunking down $500 for a separate RAID card is a little beyond what most would do.

Yea well that's the rub, a lot of the cheap options didn't work (connected but super slow). The 9260 is pricy, but it does start at about 300 for the 4-port version.

Our next test bed will be centered around consumer Sata 6.0. Right now to keep things consistent across the board for all our drives we are testing first on the Intel ICH10R native to our Dell XPS 9000 and for drives that need it we bring out the nicer card. This wasn't the first time we used it since you can see its mentioning in the VelociRaptor reviews.

Share this post


Link to post
Share on other sites
In this case, the alternative would have been to say the drive isn't compatible with SATA II, or at least, performs poorly in that case. While that's partially true, it's not entirely accurate.
but look at others-- Xbitlabs, Techreport, Anandtech.. the drive is still class-leading performance even on SATA 3Gbps... how "poorly" do you mean here? The point is a standardized testbed for results to be comparable. If you want to offer 6Gbps results too, then you should (!), but you need to offer comparable drives tested on it to make it useful. (naturally that will come with time, but a kick-start to make it useful would be good...)

TR and others listed RealSSD C300 benchmarks with both 3Gbps and 6Gbps in their charts (not just in the text), which was very useful. Why not do at least that?

r drives that need it we bring out the nicer card.
Ok, but that still doesn't mean other drives (even if they do not explicitly need SATA 6Gbps) won't benefit from the other features of your 6Gbps card if it has such dramatic differences...
a lot of the cheap options didn't work (connected but super slow).
So you guys tried.. 2 6Gbps cards? Why not more? I understand it's not SR's necessary intent to determine which cards work, but why not at least throw around a few common suggestions/cards to avoid? The Marvell 9128 wasn't mentioned at all, or was it? or who else? Does Crucial have a hardware compatibility list for the card at least?

This is a good issue to be aware of, but the lack of information is frustrating... the RealSSD C300 is staring to be recommended everywhere, and if controller issues are such a big thing why is awareness so low? (SR's forum is the first I've seen to remotely mention it as being something drastic, I didn't even notice it in the review itself, in Techreport, Xbitlabs, Anandtech, or any other review...)

Share this post


Link to post
Share on other sites

cont -

If you're happier reading reviews elsewhere, feel free to skip ours ;)

We included both SATA II and SATA III in almost every chart, including all three real world benchmarks.

As far as comparing other SATA III SSDs, would love to do it. If you send me links to all the ones you know about I'm happy to contact them. Unfortunately to this point we haven't been sent a SATA III SSD by anyone other than Crucial.

As you say, it's not really our job to make up the list for Crucial, we suggest checking with them for a compatibility list as it's frequently changing. We don't do this for any other drive either. We assume if it works in our system that it generally works. But we don't try half a dozer other cards to be sure it works. We could have tested more, but honestly, with the review load we have, we have to make decisions about how best to spend our time.

Share this post


Link to post
Share on other sites

Later this week I will try to compile a mini uncharted list of some of the 6.0 stats on the other drives for our traces. Like Brian mentioned, we don't have unlimited amounts of time, so a few items do get skipped over (not including all stats in the custom charts).

If someone wants to ship in lots of consumer 6.0 controllers for us to connect to the test bench and post results from that would be awesome, but since we have recieved none for review to date (the current one was bought at MicroCenter)... we just don't have the resources to get em all. It seems some review sites had luck with compatible hardware or the perfect balance of drivers to get it to function. In our case it was just *bad*... like in our traces it was half as fast as the Intel ICH10R.

Share this post


Link to post
Share on other sites

Well, that could be a useful new article to write... no one else has attempted it.

As far as comparing other SATA III SSDs,
I meant comparing other existing SSD's (be they SATA 3Gbps or SATA 6Gbps) on the same controller (the LSI Logic 9260) to give a fully equal test platform-- at least enough of them to get a reasonable database of results. That is the essence of 'standardized' no? :)
e don't do this for any other drive either. We assume if it works in our system that it generally works. But we don't try half a dozer other cards to be sure it works.
hence why I'm kinda surprised the LSI Loigic 9260 came out in the first place. Plug it in, test it, report the results, if it seems odd, note reasons why... no need to modify the standard, no? ;)

Share this post


Link to post
Share on other sites

Hahaha, well if you can stand some unaverage scores (posting snapshot values, not the usual average of three) here are some 6.0 stats on each.

First up is our Productivity Test:

IO MB/s response time

C300 raid0 single 128k

9061.93 266.92 0.868

intel 160gb raid0 single 128k

6252.35 182.78 1.268

owc 120gb raid0 single 128k

8316.87 242.82 0.930

intel 40gb raid0 single 128k

3281.67 95.80 2.411

HTPC Trace:

C300 raid0 single 128k

5457.00 252.90 1.427

intel 160gb raid0 single 128k

2999.24 139.86 2.625

owc 120gb raid0 single 128k

5598.39 261.16 1.395

intel 40gb raid0 single 128k

1458.80 68.00 5.467

The HTPC test the SF-based SSD won that round, with the C300 taking lead in the productivity test. Don't have the gaming trace stuff on hand though. These were already recorded from a night I had tons of freetime and just so happened to have all of those drives connected.

Share this post


Link to post
Share on other sites
It seems some review sites had luck with compatible hardware or the perfect balance of drivers to get it to function.
Well, I think some sites note performance is slower than expected, or whatnot... but given the wide variance in SSD performance in today's SSDs, performance is still indeed in the ballpark.
In our case it was just *bad*... like in our traces it was half as fast as the Intel ICH10R.
Half as fast as the X25-M G2 (or did you mean when used on the ICH10R?) So long as specifics on both controller firmware and driver revisions are noted then the data is still useful...

Without that information then the results, however good or bad, are somewhat less useful. I think that is a key component (making existing, standard hardware in the testbed work, and the steps taken), rather than glossing over it.

Reasonable measures of course. Heck even other sites admit to issues, but they usually don't pick up new hardware to do it either... hell Techreport and others have known issues with differing versions of Microsoft and Intel's AHCI drivers, either with entire drives or with specific benchmarks, or both. The steps they take to make it work and the specific drivers/firmware used may be much more helpful to readers, no?

http://www.techreport.com/articles.x/19049/6

DriveBench doesn't produce reliable results with Microsoft's AHCI driver, forcing us to obtain the following performance results with Intel's 9.6.0.1014 RST drivers. We couldn't get DriveBench to play nicely with our the X25-V RAID config, either, which is why it's not listed in the graphs below. The app will only run on unpartitioned drives, so we tested drives after they'd completed the rest of the suite.

Anandtech went ahead and published results with the same controller you guys used and rejected.. heck... it's still valid data no? :)

http://www.anandtech.com/show/3812/the-ssd-diaries-crucials-realssd-c300/2

The 6Gbps performance is an odd story to tell here. The 128GB drive actually performs the same regardless of whether or not it's on our 6Gbps controller, however the 256GB drive takes a performance hit going to the 6Gbps controller (HighPoint Rocket 620, using the Marvell 88SE9128 controller similar to many motherboards). This is purely a fault of the controller and unfortunately one of the tradeoffs you'll have to make with the move to 6Gbps. It's very difficult to beat Intel's own SATA controller.

Share this post


Link to post
Share on other sites
Hahaha, well if you can stand some unaverage scores (posting snapshot values, not the usual average of three) here are some 6.0 stats on each.
Crap! Should have refreshed before I made that last post...

The numbers are very useful, actually. Much more apples-to-apples rather than apples-to-oranges or worse, apples-to-steak. ;)

Seeing what looks like a 50% boost in some of the competition (X25-M) and 20% in others (OCZ Vertex 2)in the Productivity mark, and even if other drives made less or no gain in the other benchmarks (such as HTPC), it is all still very relevant data...

Share this post


Link to post
Share on other sites

Half as fast as the X25-M G2 (or did you mean when used on the ICH10R?) So long as specifics on both controller firmware and driver revisions are noted then the data is still useful...

Anandtech went ahead and published results with the same controller you guys used and rejected.. heck... it's still valid data no? :)

Our first SATA 6.0 results were coming up half as fast as the Intel ICH10R (chipset, not the Intel SSD) 3.0 results in our traces. In the preliminary IOMeter tests the results weren't too bad.. some were in line with the 3.0 stats or just below. As soon as the interface was loaded with our trace the scores plummeted. I don't think I recorded them at the time, but instead of the C300 being in the 9000 IO range that we found with the LSI RAID card it was closer to 3-5K.

We didn't want to take that approach Anandtech did since the results were pretty far skewed. Something was wrong so we consulted with Crucial in it before we did anything with the review. We verified the card being used (I think BenchmarkReviews had luck with the same card). In the end we ended up settling on the LSI RAID card since it *worked* even if it was expensive. We werent in a rush since we were already late to the game so to speak, but we wanted to make sure we came out with the correct results that were in line with what Crucial advertised.

Now some might bring up the OCZ situation where they didn't like the results we published, but in that case it was a difference of really really fast minus 2-3% compared to the other drives. Not 50%+ slower.

Share this post


Link to post
Share on other sites

apples-to-steak. ;)

mmmm apple sauce and steak :D

We have a huge catalog of unpublished stats waiting for future reviews or just oddball things I decided to do at odd hours. If you want something in particular or another drive included I will see if I can work some magic behind the scenes. Just don't count on everything being in the final review. Our art guy would probably have a heart attack if we doubled or tripled the stats we sent him. :ph34r:

Share this post


Link to post
Share on other sites
, but we wanted to make sure we came out with the correct results that were in line with what Crucial advertised.
Well I think in fairness to your readers, you have a standard testbed and it should be tested as such, even if the results don't seem right. Fairness to Crucial's advertising shouldn't take priority over... ;)

Now if you feel that additional benchmarks should be done then that's all well and good. :P

As soon as the interface was loaded with our trace the scores plummeted.
If driver revisions and whatnot could be added to the testbed information section in the review, or at least which ones were tried and whatnot, that could be very useful as far a datapoints goes... more information that I think would be useful to the consumer. :)

Share this post


Link to post
Share on other sites

Well I think in fairness to your readers, you have a standard testbed and it should be tested as such, even if the results don't seem right. Fairness to Crucial's advertising shouldn't take priority over... ;)

Well thats the rub. We changed the testbed for that review by adding the HighPoint card to make it support the SATA 6.0 interface. That interface was flakey so we looked for an option that worked for us in the past.

Share this post


Link to post
Share on other sites

I wonder if different drivers or a different card with the same Marvell 9128 chip would have worked better, since it looks like most reviewers around are using the Marvell 9128 for their 6Gbps testing (usually integrated into whatever motherboard they're using). As other sites have found though, drivers and firmware definitely are part of the mix too.

Share this post


Link to post
Share on other sites

Hmm. Well, the point of a standard testbed is a standard testbed... you could just standardize on the MegaRAID 9260. :P Although all the cache is indeed an issue to address.

Lots of others have chosen to standardize on whatever's onboard typical systems today to make sure they're more representative of what typical enthusiasts are deploying (such as the Marvell 9128..), as plunking down $500 for a separate RAID card is a little beyond what most would do.

going assbackwards, the 4port 9260 is competitively price ~$330. How does one define 'consumer' 'enthusiasts'? Gamers will spend huge $$$ to tweak their systems, Raid cards are part of their arsenal, no?

Wondering Y SR didn't go for an old industry standard in ATTO, same price range, perhaps better performance...which until this month was less of a 'cache'/standard HBA card??? the ATTO H644 SAS/SATA 6GBs 4port Raid card can be had for as little as $307:

Can't really understand the announcement, what the cache in question is vs the cache contained in the LSI card:

ATTO’s 6Gb/s SAS/SATA RAID Adapters Now Available with Battery-less Cache Memory Data Protection

Also, ATTO uses something they claim improves performance:

http://www.attotech.com/ads.html

* Bus Mastering eliminates the need for communicating with the host CPU by transferring data directly between the ATTO host adapter and system memory. This relieves the CPU from the burden of managing protocol activities and increases system bandwidth for other applications.

* Tagged Command Queuing allows for multiple I/O commands to be processed in any order, improving overall performance of read and write requests to hard drives.

* Disconnect/Reconnect increases bus performance by eliminating wait time between an ATTO host adapter and other devices. This enables dynamic sharing of the bus bandwidth in sophisticated environments.

* Advanced algorithms are implemented to provide optimized performance for workgroup applications such as digital content creation, rich media, prepress and databases.

Then we have the enthusiast which uses the 'standard' same chipset that SR is using the Dell, ie ICH10...which viola has it's own cache implementation! And so, not only is the SDD controller & firmware import, we though in another difficult item to 'standardize' on.

Marvell SATA 6G SSD Performance vs Intel ICH10

Results in bandwidth speed tests for the Agility SSD were very similar to the IOPS benchmarks. The X58 and P55 platforms produced roughly 222 MBps reads and 165 MBps writes, all while connected to the Intel ICH10 Southbridge. In comparison, the Marvell SATA 6G controllers produced approximately 154 MBps read speed, and 106 MBps write. This results in a 42% advantage for the ICH10 SATA 3G controller over Marvell's 9123/9128 SATA 6G controller in read speed, and a 55% advantage in write-to performance. Why, you might ask? It's all in the driver.

You see, Intel's ICH9 and ICH10 chips utilize a driver architecture that creates a virtual RAM-disk buffer for cached transactions. The amount of RAM-disk created depends on the chip and available system memory (which we detail in our spin-off article), but the combined driver refinement and memory cache capability lead to very high performance. This is where Marvell needs to grow, and do more than a simple reverse engineering Intel's work in creating their own driver.

Just to make sure we aren't seeing an isolated issue, Benchmark Reviews went on to test several more SSDs on the ICH10/Marvell 9128 and 9123 controllers.

Kingston SSDNow V+ SSD: SATA 3G vs 6G

Edited by udaman

Share this post


Link to post
Share on other sites
ow does one define 'consumer' 'enthusiasts'? Gamers will spend huge $$$ to tweak their systems, Raid cards are part of their arsenal, no?
No, hardware RAID cards with onboard cache generally are not part of such systems.

See the system guides put out by Tom's Hardware, Anandtech, Ars Technica, Tech Report, etc... almost nothing below $2000 is going to include a RAID card as such. Heck, stroll on over to Dell, HP, or even your more local system builder such as Puget Systems, and again, in typical gaming box price range you're not going to see a hardware RAID card with significant cache.

ie ICH10...which viola has it's own cache implementation!
Yes, but Windows has a file cache too! OH NOES! ;) The main concern here is the dedicated hardware cache, which is a significant difference in a real hardware RAID card vs. a standard SATA controller, and hence is a variable we can worry about and more important, exercise some degree of control over. :P
! And so, not only is the SDD controller & firmware import, we though in another difficult item to 'standardize' on.
heh... true there, definitely another item we have to pay attention to. Fortunately since ICH10R or the later P55 PCH is pretty common now those are "easy" things to standardize on (or AMD's 750 or 850 southbridge...).

And going along the same path, hence, listing details of the testbed including driver revisions is something that might not have been so critical before, but is now likely to be of interest to many more people.

FWIW, just about everything you quoted from ATTO is standard in any SAS controller. Hell bus mastering devices have been around for what, 15 or 20 years? Actually nothing even uses TCQ (tagged command queuing) anymore, as that has been long superseded by Native Command Queuing (NCQ), which like connect/disconnect is a native part of the SAS spec.

Share this post


Link to post
Share on other sites

Seconded...reviews up for my C300 if it halps at all... I threw all my tests in.

Share this post


Link to post
Share on other sites

We worked with their technical team pretty extensively. They have recommended hardware to get optimal performance. Unfortunately the list is still somewhat in flux.

Share this post


Link to post
Share on other sites

A lot of other reviewers have had issues with the C300. Lloyd Chambers does tests on the Mac platform targeted at photographers and found its performance degraded and was unreliable.

The SandForce drives seem to perform excellently in any conditions.

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