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

at work i don't have an ssd, so i put my tempfiles for ie on it. it's a 5400rpm platter, terrible.. (up to ten minutes tilll the system is actually "smooth running". and if anyone believes stuttering is an ssd only thing he can visit me and my work notebook )

 

well, as a chip costs around the same, unimportant how much bits per cell it can store, those 4bit/cell chips would reduce cost to half.

and yep, right now i realize it's not 4bits / cell but actually 3bits?

slc has 2 states (1bit), mlc right now has 4 states (2bits) and the new ones have 8 states (3bits)? or do they have 16 states (4bits) ??

if so, a factory could suddenly create 2x or 4x the storage per $. so that would be a massive drop of cost for ssds. (and usb sticks, and mp3 players, and phone storage, etc.. )

 

None of the commercially produced flash is more than 2 bits right now. 3 bits and 4 bits is at research, I think they will both release in the future.

The problem right now is that with proprietary controllers the pricing depends greatly on the controller. Bring on the SSD controller on chipset!

 

I don't know if people have posted this already (this thread is getting way, way too big), but Intel plans to release a 320GB version of its SSD in Q4 2009. It'll almost certainly cost some completely absurd amount of money, but it is equally certain that it will drive down the price of Intel's 160GB SSD and probably most other SSDs on the market.

 

one facturer got wavers with 3 bits recently, there was an anouncement.

and yeah i know your chipset idea..

 

I'm actually surprised Intel is waiting so long to release new SSDs, by that time other companies will be at 1TB, maybe more.

 

Current MLC implementations are 2 bits/cell. According to Sandforce, it can be used for MLC or SLC implementation.

It doesn't need seperate cache on chip or DRAM. You can just use system memory like FusionIO does(FusionIO relies on both the system CPU and the RAM to do its work).

The most common MLC implementations use 512KB as a block and 4KB as a size. SLCs like X25-E uses 128KB blocks. You don't have to use 8-channels all the time. It may be even impractical to do that. Just using one channel to write if necessary is a smarter idea.

 

Getting some stuttering with the new Intel firmware... at the end of copying over a number of files from my external drive to the X25-M, programs freeze for just a second. A video will start playing very slow, or you won't be able to scroll up or down or interact with a web site in your browser. This also occurs occasionally while downloading usenet or torrent straight to the drive.

Any one else experiencing this, or able to recreate it on their machine?

 

What exactly is DRAM?

 

Dynamic random access memory?

 

Type of memory used for computers. Like when people say I got 2GB of RAM they use type of DRAM.

 

More info can be found via Google or Wikipedia...

 

Anyone know where the current/best place to get the Samsung 64GB SLC drive?

 

Yea sort of. If you haven't noticed on the review by now let me tell you something interesting. On the graph for new firmware test, the dip is drastic. You'll see on the "slow" firmware that while the overall speed might be low, the lowest speed isn't that much lower.

You'll never see that on hard drives or DRAM solid state devices. SSDs have inherent bottleneck that can only be mitigated. The thing you want to make it 0 stutter is not only have decent IOPS, but to maintain it.

I'd love to see "Braidwood" controller kill the current X25-M devices so we can get the benefit without the prohibitive pricing.

EDIT: Yea, who the new firmware helps is people who experience slowdowns to a point that becomes unbearable. Those who were getting catastrophic slowdowns(write speeds in 10-20MB/s range) will see a benefit.

 

$180-200 from www.geek.com, last time I heard.

 

I remember seeing that also but they are sold out. Direct link is: http://www.geeks.com/details.asp?invtid=MCCOE64G5MPP-0VA&cat=HDD&cpc=JAD

 

I think Newegg has the 1.8" version for $175, if you want to get a 1.8 to 2.5" adapter.

 

Yes, Newegg but you'll probably need an adaptor unless your hardware "natively" supports 1.8" drives.

Unfortunately it looks like they don't list it anymore though... only the 32GB version now:
http://www.newegg.com/Product/Product.aspx?Item=N82E16820147055

Maybe they will get more.

 

Last i checked ebay, a lot of the auctions were going for under 200$

 

Yes, their enterprise controller supports SLC and MLC. (The datasheet for their mobile controller only lists MLC support.)

Your two points contradict each other: using system RAM and using fewer than 8 channels, it would be impossible to maintain a 250MB/sec transfer rate. And if the cache were in system RAM then there would be no point in mounting a super-capacitor on the drive itself. Therefore these points of your post are clearly wrong.

The controller obviously has on-chip DRAM, that's the only way to maintain high transfer rates, and that's the only reason why the drive would need backup power.

 

Regarding my earlier report of stuttering on the Intel drive, I got in touch with Allyn who wrote the two X25-M articles at PCPer. Here's what she had to say: There are some obscure cases where the new firmware may stutter while writing to new areas of the drive. This should only happen the first time those areas are written. If mine was doing that I would HDDErase it to start from a clean slate.

I'll HDDErase the drive and move back my OS and files onto it over the weekend, and I'll let you guys know if that eliminates the stutter.

 

You are confusing with sequential and random throughput like rest of the people do. 30K IOPS in random writes is 120MB/s. Plus, they can maintain that by caching into faster memory so they don't need all the channels active all the time.

You still would want super capacitor even if using system RAM because the system RAM's content would disappear and would result in all the data being lost otherwise.

http://www.sandforce.com/index.php?id=145

So they call adding DRAM a "band-aid" solution but you still need a buffer which if you look on the datasheet it says it has a buffer.

Let's see what they REALLY do. But in the meantime their numbers are all we have.

 

Quick question here: Size aside, which is better: Samsung 256GB or Intel X25-M?

Thanks

 

I would go with sammy...
why you ask ?

AFAIK, sammy has no problems with performance degradation...
new fw from intel should resolve that issue, but, look few posts back...

 

256 GB Solid State Drive Ultra Performance

i just purchased a dell studio 16 with 256gb ssd,
i think im gonna sell the ssd to fund the purchase. i heard that all dell ssd are
from samsung. so how much i can sell samsung 256gb ssd on ebay?

 

I think I love ruining people having fun:

http://pc.watch.impress.co.jp/docs/2009/0212/hirasawa014.htm

http://pc.watch.impress.co.jp/docs/2009/0212/hira_sam8mbw.gif

Do you not love the 4.8MB/s sequential write minimum speeds?

With the same 8MB size Intel's minimum sequential write is 14.8MB/s in that site, before firmware update.

Ashura: Is that a personal e-mail? Because I can't find that claim by her on the forums. If it is, I guess its a good thing.

 

Yes, I got in touch with her via e-mail. And completely agree with you, QuadAllegory I would still go with the Intel.

 

I had no idea that PcPer article was by a chick Stereotypes at work. More power to her and good for her!

Back on topic, I guess with the SSD tech nowadays if you were to push any drive to inordinate limits, you would literally find a flaw. So, we should be looking at normal world usage, I am not sure if a lot of people got the slowdown on the X25M until pcper got their hands on the drive or we would have never known. The same is true for the Sammy. I am sure they will cut new firmware to deal with this soon enough.

 

The samsung 256 GB drive is now on newegg: http://www.newegg.com/Product/Product.aspx?Item=N82E16820147058

 

Samsung going straight to retail now, hmm? Neato, hopefully we'll start to see their newer drive in smaller capacities. Of course with OCZ (and others) set to release their Samsung controlled SSDs soonish, it won't matter I supposed.''

Also, the 256GB 1.8" Samsung dropped in price by $1,000 to $960.95 USD if anybody is interested >_>
http://www.memory4less.com/m4l_itemdetail.asp?rid=fd_10&itemid=1442178602

I'm guessing it was an error before when it was priced at close to $2,000

 

yeah, here in switzerland samsung now starts to appear in (online) shops, too.

 

Though it is a 100GB less, almost, I would much rather get the X25M-160 for a bit more. 'Course if you get the Dell variant for a tad over $500 with a coupon, it's a different matter..

 

Incase no one saw it yet the Samsung 256GB 2.5" is now at Newegg for $669.
Strange considering I thought it was an OEM drive, but I guess they bought them in bulk.
http://www.newegg.com/Product/Product.aspx?Item=N82E16820147058

Good for people looking for the Corsair, and it's almost $100 less than the Vertex.

I still want to see the Vertex and the Sammy go head to head. With sammy's latest firmware it should be close I would think.

 

This is becoming reasonable. Except for Intel, the price for non-JMicron drives of decent size is now $2-3/GB. Let's hope the prices continue their rapid decline.

 

Isn't the Sammy better due to no performance degradation?

 

All SSDs degrade in performance over time; it's a function of how they write data. Read the AnandTech SSD anthology article if you want the details.

 

If you're looking for a big SSD, I suggest waiting till the end of 2009. Samsung is supposed to come up with its new 32nm chip that can also do 3bitMLC. That means that it will cost them 1/2 the price to make a chip, and each chip will hold 50% more data. If we take 64GB as a reference point(150 bucks), we can see that 200GB drives might very well only cost 150 bucks by the end of 2009!

 

its not necessary to wait till the end of 2009.
say someone waits till the end of 2009, and another guy will tell him to wait till mid of 2010 coz new product will come out.

 

I don't know where you heard this. The last I heard from Dell tech support rep is that they use Sammy/OCZ and one other brand, but don't remember it. Take it with a grain of salt however. Dell's great at keeping things confusing and frustrating.

 

I was always under the impression that dells 256gb ssd were sammys. I haven't heard of them selling any ocz drives.

 

I'm fairly certain Dell doesn't use OCZ. The vast majority of their drives should still be Samsung, but I think there's actually an X25-M option for the E6400 now There isn't any other 80GB MLC on the market, is there?

 

yep, only heard about sammies.. hm..

 

Yea, I think i'll be waiting til the end of the year. I think it'll definitely be worth the wait. The price will need to come down though!

 

what is known for the new 256GB Samsung SSDs ? Do those work with no tering ?

they poped at newegg recently

 

yep, all current ssds work without bigger problems. intel, vertex and other indilinx based ones, samsung, mtron. those are the names that work fine. any others i forgot?

 

Hmmm... at that rate it might not be too long until hard drives become pretty much obsolete! I thought they would last a while (for cheap storage reasons) since they're so much cheaper on a per/GB basis but that may quickly change in the next couple of years... faster than I thought.

 

i like that. i like it when markets push forward at fast pace. i don't care about obsolenting stuff that still works great

i prefer this over stagnating markets like the cellphone market where no one want to introduce amazing new tech, or drive prices down as they all want to make as much money as long as they can..

and believe me, no ssd, no matter how "old", gets obsolent as long as it still works and there are hdd's out there

 

Guys, stop worrying about the Intel stuttering. That issue has been already dealt with through the new firmware update. In any event, most of us would have probably never seen the issue of performance degradation to begin with.

 

Anyone with a Dell + SSD notice a performance difference between the BIOS accoustic mode set to either "Bypass" or "Performance"?

 

Regarding the Samsung products - Anyone seen any IOPS data for any of them?