The new SSD Thread (Benchmarks, Brands, News and Advice)

Well, AFAIK, there are 2 slowdown issues:
1. The issue caused by fragmentation and/or the intel controller algorithm doing some whacky stuff after a while (PcPer article)
2. The issue that when the drive is quite full the controller has less options to work its magic and spread the random writes causing slowdowns.

I am not sure how relevant #1 is, since a lot of users are reporting that it is only noticeable in benchmarks. not normal usage.
#2 has me concerned though, since I will use it as primary and sole drive in my laptop ... and I might easily use 65GB out of 80GB during normal use.

PS: Didn't intel promise a defrag utility when AT benchmarked the drive and noticed some irregularities? It has been months since then and still nothing from intel. (and for all the fans: OCZ might suck but at least they are more responsive in that area. )

 

With regards to number two, I'll fill my drive up to capacity tomorrow and use it for a few hours, let you know how it runs in both benches and real life use. I wonder if that's an Intel specific problem or a general SSD problem, can someone who uses a different but good SSD (Corsair, Samsung, Mtron, pretty much anything non-JMicron, etc.) do the test with me and report their findings too?

Let me know which specific benches you'd like to see to check random writes.

 

well, i still have no (0, zero) use for MFT on my mtrons. hasn't shown any gain, and instead started hard stuttering on lower end ssd's. i won't use it due the more unpredictable writes. and i wouldn't use it because of the danger in backups from now on, ever again. if you have a good ssd, you should have no need for it.

 

I went straight to the authorised distributor and asked them to indent a piece specially for me. Wasn't a problem as they distributed Corsair products for my country.

I would have gone for the Intel X25-M too, but it's rather sad that pricing for the Intel drive is still stuck at around US$479 here (no idea why), so I didn't have a choice but to take the Corsair/Samsung OEM.

 

I understand that the Intel X25-M 'runs better' with AHCI and the Intel storage driver?

What about the Corsair/Samsung OEM? Does the controller work better with ATA or AHCI?

When I was using mechanical drives, I had them in AHCI but didn't install the Intel storage driver as I found no difference with or without them.

 

With regards to AHCI it depends on the SSD. The Intel definitely gets an improvement but others (I believe certain Mtrons, all of the OCZs) get a speed decrease with it enabled. The only thing I can say is try it out yourself and see which runs better.

And, the 'Intel Storage Driver' most likely refers to the AHCI driver. You should already have it installed since you're running AHCI mode with no problems.

 

AHCI only enables NCQ, which the benefits of that are dubious with mechanical drives and non existant with SSDs, and "hot plugging" so if you don't use eSATA it's actually better to leave AHCI off.

 

No, I'm pretty sure that the Intel SSD uses NCQ for its "write-combing" algorithm to speed up write operations.

 

Where can I buy 1.8" SSD drives? Are they more expensive than 2.5" SSD drives? Some laptops with 1.8" hard drive slots only take these. (SATA)

 

intel has 1.8" ssd.
check ebay..

 

Any 256GB 1.8" drives out? The highest 1.8" capacity I can find on eBay is 128GB. NewEgg has a few 256GB drives at around $500, but they're all 2.5".

 

Intel® X18-M Solid State Drive, 160GB SATA II 1.8in, MLC

 

this is just crazy

 

What is crazy ? the 240GB ? the price ? or the speed ?

VS2005/2008 is flying on this rig

 

The controller itself should have the write-combining algorithm without the NCQ. If it relies on the NCQ that's bad news because that would also affect reliability with it off(making it same as drives without reliability feature when the NCQ is off). I think of the NCQ as the 2nd level WC "buffer".

sitecharts: None of the SSDs really escape slowing down as the drive is full. It should be same on the Intel SSD as any other SSD.

 

Actually my mistake, I meant for OTHER ssds, intel uses NCQ supposedly to queue pending operations while working on one concurrently. Most other SSDs seem to suffer with it on and it's probably because it's designed for mechanical disks to optimize reads and writes based on physical location on the media and I'd assume its implemented just to comply with the design necessity of having it.

 

The speed is crazy. The price is about what you'd expect; the 240GB is also nothing special, but the speed is insane.

It seems more and more controllers are coming out. Here's another one by some company called SiS.

 

^
||

yap, the speed is insane, it's f***** crazy

 

And with 3 OS on them, half-full :


they are realy fast, I will upload a PCMark 05 disk bench too

 

how fast it boot in w7 ?

 

Yep, eBay is probably your best bet. Even there, there really aren't that many options for 1.8" SATA SSDs; there's like a couple Samsungs and the Intel and they're all significantly more expensive than their 2.5" counterparts. Furthermore, AFAIK, 1.8" drives don't come with more than 128GB of space right now.

Companies are realizing that there's a market here? Though, of all companies, SiS is the one making a controller. Somehow, I think that's hilarious, but if they can make a good one, all the more power to them...

 

Wow, yeah, given SiS's history I wouldn't have real high expectations until I saw it in use, lol.

 

Is the lack of a controller that can handle larger capacities the reason we're not seeing more 3.5" SSDs perhaps? An article mentioned that the controller the Vertex uses can only handle 512GB drives. When you consider how many more NAND modules you could stick in a 3.5" drive, perhaps that't not practical yet, due to controller limitations. With bigger 3.5" capacities, SSDs could easily compete with Velociraptors, if price continue to drop as precipitously as they have been.

 

Big drives still cost a lot and extra part #s are extra parts to sit on the shelf. If they can make a 2.5" drive that can be adapted to a 3.5" if the user chooses allows it to be sold for more computers. Until flash memory prices drop substantially more the cost of a higher capacity drive wouldn't produce volume to justify building it in the larger form factor and limiting its market.

 

As the price drops the technology will improve. Hopefully they will just start making desktops 2.5" since there won't be a need for bigger platters and faster RPM... unless space starts to become a real issue again and they need the extra real estate for more flash modules

 

I agree with you. That is my understanding too.
But since I will most likely have 65GB used for the OS, programs, Outlook PSTs, etc, that is only ~50% used on the Corsair and 81% on the intel.
Based on that ... can the intel maintain its lead in speed or would it drop off to perform about the same as the Corsair?

@ashura: what say you? Did you notice any differences in general performance? doing a random write test on an almost full drive is probably the key benchmark.

 

Filled the drive to where I only had 500mb free and crystaldiskmark write scores did suffer a bit, but I noticed absolutely zero difference in real performance. I'm a fairly heavy multitasker so I definitely would've caught any dip in speed compared to what I was getting before.

However, I did this on the 1000H and I don't know if the SATA I chipset influenced the benchmark results at all. It shouldn't since the write speeds are well within SATA I limitations, but just in case I'm going to pop the X25-M in my SATA II capable x200 when I get home tonight and redo the benches.

 

The Intel has a long way to go before performance drops off to the level of the Corsair/Samsung. I'd say even with the drive entirely full, it'll be significantly faster. If I recall correctly, even in that PCPer article, the worst-case scenario for the Intel is faster than the Samsung/Corsair?

EDIT: Whatever happened to the AnandTech SSD article?

 

I've been wondering the same, hopefully it'll be up soon. And you're right about the worst case still being better than others' best, at no point did the drive get as slow as the Corsair 128 or Titan when it comes to zippiness and general performance.

 

Well I tested both my Samsung and G.Skill with and without it. The performance gains are huge. The Samsung does work well without it. But the improvement with it is worth it, and it still comes out cheaper than a single MTRON. I've got no (0, zero, nunca, zilch) use for an MTRON, since they are outrageously expensive and don't have any large enough for normal computing on a notebook.

 

cdfreaks has their review of the Vertex with numerous benchmarks up:

http://www.cdfreaks.com/reviews/OCZ-Technology-Vertex-Series-120GB-SSD-Review/conclusion.html

Wish they had an Intel drive in there testbed too.

 

What is the Ridata 64GB SLC? Is that a Samsung rebrand? I've never heard of the Traxdata either...

Also, IO and random writes are looking pretty good now. I'd still get an Intel, but this doesn't look bad.

 

Very nice... I'll wait though - to see more reviews and see if any problems come up... as well as for the price to drop which I suspect it will pretty fast. Wouldn't mind seeing the upcoming Summit's as well.

 

Here are a collection of Vertex review; it looks very promising imo.

http://futuremark.com/news/55225_ocz_vertex_summit_solid_state_drive_reviews/

 

I just swapped out the corsair s128 on my x200 for the intel x25-m. The intel does feel faster but not substantially, at least for how I use my lappy. I still stand by statement earlier that the corsair s128 is the perfect blend of performance/storage for a laptop.

 

And so it begins... the x25-M death dance.

Ugh.

 

Wow, really? Have you just been using it normally (well, define normally, too)? How long?

Also, does it feel significantly slower?

 

^^^That's messed up, dude. What exactly have you done to the drive

 

Been using it about a month as my boot drive (Vista 64-bit).

Nothing unusual in terms of my usage. Just normal workstation stuff...

 

Give it to Anandtech or PCper and see if it strikes a blow to newegg/amazon prices..

 

Seconding Commander Wolf's Q, are you feeling a big difference in real life performance?

 

Most of the time no, *BUT* every now and then the entire system will freeze solid for about 30-40 seconds and then pick right up where it left off thereafter. It happens every 5-10 minutes and it *IS* annoying.

 

Geez... and that's after a month of use? Intel better release that utility mentioned in PCPer and anandtech soon. I don't what Intel's customer service is like, have you gotten in touch with them to see if they have something to say? (probably not considering they publicly rejected PCPer's article)

 

Wow, that's terrible. So the Intel "stutters" after all, eh? Did the performance degradation happen overnight or did these freezes start happening gradually to the point that you re-benched the drive and found it in the state in which it currently is? I'm really surprised this happened under normal use. Maybe this is why the AnandTech article is being delayed...

 

Can anyone using Diskkeeper's Hyperfast tell me what if any downsides exist for using it? ie. with SteadyState you had super long boot times and couldn't move around files totaling more than the set cache in size. Anything like that with Hyperfast?

 

Darn, 30-40 seconds every TEN minutes??? Sounds like absolute torture to me

Dear companies, please rescue us from STUTTER, PLEASE! I will not buy until there is a drive out that is PROVEN not to stutter.

 

I haven't heard or read any stuttering reports from the Corsair S128.

Guess I'll stay away from the Intel for sure - unless they lower the price to justify this issue and/or release that defragment utility.

 

Here's the deal:

I had a pair of Intel X25-M in RAID 0 on this box (Dell XPS 630i), but needed to temporarily use one of the SSDs in another machine, so I saved off the OS state via Acronis True Image Home 11, removed one of the SSDs, and reimaged the remaining drive back to the same state that was just on the RAID.

2 days later is when the stuttering started.

This afternoon I saved off another system snapshot, put the second SSD back in, recreated a fresh RAID 0 and restored the OS.

Currently:




I realize that I probably should be doing a full erase of the SSDs between each step and/or performing partition alignment, but I haven't been (it's a bit beyond me at the moment).

Still, the system halting under a single SSD was quite unnerving. Needless to say I'll be monitoring the RAID performance closely...

 

Hmm, wonder if others who're experiencing the Intel death crawl also imaged or if this inflicts fresh install drives too.

 

I do recall one or two of the articles linked here getting most of the performance back tho, by writing specific data. Think it was the Intel X25-M PCPer article.

Edit: Yeah, PCPer