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

I was browsing through this thread and found out information about the OCZ Core V2 series in Europe, but not the U.S. All I can say is that I don't know who's planning to carry them, but they are currently available from major technology distributors, including the one I use: D&H. Estimated retail prices are below:

30GB - $279.99 - available
60GB - $399.99 - available
120GB - $639.99 - available
250GB - no pricing info - unavailable/eta end of september

I hope this helps

 

Do you have any links to this?

The suppliers I see have no stock and are uncertain about when they will get them.

 

http://www.ocztechnologyforum.com/forum/showthread.php?t=42137

theres someone with v2 120gb, same problems (stutters, presumably due to v slow random write performance as per original core drives). keep in mind the SATA1 samsung drive was 'pretty slow' and gets blitzed by the samsung sataII (which ocz also resold/rebadged)

i would guess all they have done with core v2 is add the mini-usb to allow future firmware updates (in the hope their engineers can actually correct some of the major issues with performance and installation/corruption). i actually feel sorry for the ocz tech support because it doesnt look like theyve been told much at all about the drives from the engineers, and often you see them saying 'the drive must be faulty, RMA' for the same performance issues that everyone has (i.e., thats just the bad random write performance of the drive, no RMA will fix it)

so back to waiting on intel or maybe micron drives before we get good performing MLC drives

 

My laptop is pretty old, with a 915PM chipset.
Would there be any compatibility issues if I upgrade to a SSD?

 

The last time I checked, a few days ago, only one or two online retailers were listing the OCZ Core2 drives, and they were listed as "September 29" availability. But today I checked and several places had them in stock, so I ordered a 120GB drive from mwave.com. They're only an hour or so away from me so it should arrive by tomorrow. They're selling at $499.00, which is a terrific price compared to what other places are showing.

 

Hi folks,

ive bought OCZ core series v2 60GB yesterday and install it into my Vaio VGN-SZ61XNC (european version of US SZ650). First tests are here http://rapidshare.com/files/140967115/core_series_60GB_v2.zip (Atto, HDtune, HDtach). Maybe its possible to tweak it in some way but i dont know how.. I hope itll help you.

 

Well those benchmarks look very promising. By the way please use attachments next time. Much easier for us.

Do you notice any hickups/ slowdowns?
Can you do a HD Tune write test?

 

Hi,
ive posted these files on some czech sites thats why its on rapidshare.

Ive tried Write test but it asked me to delete partitions. Does it really need just to delete partitions or completely clean HDD? Cause i dont want to disassemble my notebook again.

Hickup/slowdowns - you mean something like freezing for a second or so? I didnt encounter this behavior. But i have this drive just for 20 hours and i was sleeping tonight

Im using XP with some tweaking (nLite, TuneUp, etc.) so notebook worked really fast even with previous Momentus PSD. Benchmarks ive posted before were done with clean XP instalation. After installing all drivers and office/developing software results are a bit slower. Thats seems weird to me.

First impresions (its my first ssd): Starting windows isnt as fast as i expected, but hibernate and waking up is fine. Starting apps and browsing folders (photo previews, etc.) is really impressive. Copying large files is ok, making partition image was fast enough and i didnt have problems with small files..

Ive read somewhere thats best to turn pagefile off. Is that true?

Sorry for my english.

 

Yes that's what I mean. Like described here:
http://www.ocztechnologyforum.com/forum/showthread.php?t=42137
and here: http://www.alternativerecursion.info/?p=106

I'm not sure what the best way is to tweak your OS for OCZ Core. Turning off page file makes sense though.

 

Problem is Im not a normal user. I use old windows, old office, etc. cause of low resource demands - my notebook runs on "slowest" power management settings.

No freezing for several seconds to me.

Ive tried to compress 4,37 GB mkv file from one partition to another with 7-zip. It says speed: 8500 KB/s, time 9 minutes. Starting apps (Maxthon, Word) is fine.

When compressing was in 50% I started another compression. From second partition to first. File 2,3 GB, time 8,5 minutes, speed 4700 KB/s, first compression slowed down to 7500 KB/s, and expected time (i mean processed + time to finish) changed to 11 minutes. Starting apps is slower but still faster then with my previous drive.

Maybe you are interested in another type of "real usage" test.

 

Can you post also a write test of the new Core v2 ?

 

One more test (h2bench) here http://rapidshare.com/files/141397724/CoreSeriesSSD60v2.txt Maybe itll show you real world performance.

 

Tobias could you please run HD Tune or any other write test?

PS. just read what you said, I'm not sure if you have to delete partitions.

 

Hi,

HD Tune, sisoft sandra etc. really needs completely clean drive. I have only one HDD position in my notebook and as I said - 1) i dont want to disassemble my notebook again, 2) I have my SATA ports on other computer filled with HDDs in Raid so I cant disconnect anyone..

Maybe this picture will help you. Its really small utility so U can compare my results to yours very easily.

 

does not look that promising...

 

ATTO Diskbench needs less than 1GB free space. Also, Les has run ATTO Diskbench on a large number of SSDs so we have some sort of comparison (yes, I know that it is far from perfect to compare two different systems, different installations etc. but still it would give an indication).

 

yeh, all the sequential tests like atto, etc should run fine (well hdtune is too but write test is across entire drive). though thats unfortunate since its all the random write stuff that the drives are so poor with, that cant be tested except with empty drive (with another drive having OS)

 

Gopherblue, could you run and post a CrystalMark Diskbench of your Samsung SLC SSD?

I just ran one on my old MTRON (the first 6000-series) and it does not seem to be that much faster than Tobias OCZ Core 2 when it comes to small random writes. However, I do run AES-256 bit full disk encryption in software (Truecrypt) on my SSD, which probably influences the speed negatively. However, since i do run a core duo 1.8 GHz, and the transfer rate is very low when it comes to random writes of 4k blocks, I do no think that the negative effect on the benchmark outcome should be significant.

--------------------------------------------------
CrystalDiskMark 2.1 (C) 2007-2008 hiyohiyo
Crystal Dew World : http://crystalmark.info/
--------------------------------------------------
Sequential Read : 64.917 MB/s
Sequential Write : 55.731 MB/s
Random Read 512KB : 58.541 MB/s
Random Write 512KB : 40.655 MB/s
Random Read 4KB : 15.307 MB/s
Random Write 4KB : 4.716 MB/s
Test Size : 100 MB
Date : 2008/09/30 10:03:08

 

Hi folks,

Atto results were in first file I uploaded.

 

Here's an Atto result for Core v1.
http://benchmarkreviews.com/index.p...sk=view&id=200&Itemid=47&limit=1&limitstart=7

 

I got a question. Say a cell wears out and the wear leveling circuitry removes it from inventory and reduces the size of the SSD. If i had many partitions on that SSD, which partition will see the shrink, since there is no physical correlation to a logical partition on an SSD.

 

U wont see the shrink at all. When an area goes bad on a hd, the entire block gets shut down. In an ssd, the smallest unit gets shut down which is a single bit (I believe) in an ssd....unless of course you have a mlc ssd rather than a slc ssd at which time the smallest units are paired which means you would lose 2 bits ehehe.

It is believed that even though this occurs, the user would never even know of it as it is not announced and there is no visible reduction in the ssd size.

 

So it appears to be the same size but in fact it is smaller? I hope logically the OS will know that it's out of space and not think it's there to be used.

 

Modern hard drives have a pool of spare sectors, which are used as replacements when bad sectors are encountered. Some SSDs also have a pool of spare memory blocks, which they use as replacement spares.

To my understanding the current SLC-based SSDs have memory blocks of approx 128 - 512 kB size (I believe that MLC-based SSDs have larger memory blocks and that is one of the reasons for their poor random write performance).

When your operating system writes e.g. 4kb of data to the SSD, the SSD needs to read the contents of the entire destination memory block, erase the entire memory block and then write back the entire memory block (now containg the new 4kb of data).

This would imply that the smallest failing unit in an SSD would correspond to the size of the failing memory block.

 

What if there are no replacement spares left?

 

then you start losing data...

All current flash devices (SLC and MLC) use 512 byte sectors for reading and writing. If all you're doing is writing new data, you can perform a series of 512 byte writes without any delay. The only time the SSD is forced to do a re-read is when you're rewriting an existing sector, and then it has to read the entire (2MB or so) block and copy it to a new block, with your new data inserted.

 

2MB seems like a large block size to me. That would put each gig at only 512 blocks. This would mean that you only have 32K blocks for a 64 GB drive. That seems like this would have terrible wear leveling characteristics.

Now if you said 2KB I'd be onboard with you.

Cheers,

 

Okay... but what if you ran out of spare blocks, but still had free space? Will it move the data to a free space and reduce de size of the disk?

 

I think he meant a block of data (like, to copy 2mb of text). Not a 2mb block of storage space.

 

NCIX has Intel X25-M listed. The 80GB costs $724.24 canadian dollars, just to let you know an idea of how much it'll cost . It's not available yet however.

 

What a rip.

 

Yeah, that 2MB size was just something I was quoting from memory. Just looked at another Samsung NAND Flash data sheet http://www.datasheetcatalog.com/datasheets_pdf/K/9/K/8/K9K8G08U0M.shtml

There are 4x512 byte sectors per page, and 64 pages per block. So, only 128KB per erase block.

 

well, as it looks like that device (while mlc) can even beat the mtrons mobiles and memorights, the price is nice.

i can't wait for the first benchmarks at least

 

Ok, that's a bit better -- now we're talking a factor of 16 more blocks. Still, that's only 500K (give or take) allocation units for a 64GB disk.

So, if you're going to have swap, don't put it on the SSD

Cheers,

 

What do you mean it can beat mtrons mobiles and memorights? As far as I can see, transfer rates maybe, but who knows about access time..

 

if the numbers are right that got posted earlier in this very thread, it'll be faster at transfer rates, faster at access times, and espencially faster at random reads. (iops are much higher in general, too)

but.. as always, if not for real, we can't know.

 

Heads up guys, new SSD from Patriot is here with 175MB/sec read, 100MB/sec write:

http://cgi.ebay.com/PATRIOT-PE128G25SSDR-128gb-SATA2-solid-state-disk-SSD_W0QQitemZ250290323525QQcmdZViewItem?hash=item250290323525&_trkparms=72%3A392|39%3A1|66%3A2|65%3A12|240%3A1318&_trksid=p3286.c0.m14

128GB for about $500, it is certainly MLC but these specs are nice.

 

It's one thing to claim a spec, yet another to actually meet it. We've seen that enough with other SSDs to be healthy skeptics about the numbers until they've had some scrutiny.

Cheers,

 

Yes, of course. That's why I was searching for a review and there seems to be one or two review of the older generation Patriot SSD but not this one.

 

If this has already been posted, I apologize for double posting.

As you might now, there is a company “Easy Computing Co” that offers a software product called “Managed Flash”. It is said to convert random-writes into sequential-writes. The performance increase in random-writes is said to be increase by a factor of 10x – 30x, which sounds very promising.

The above software sounds very interesting to owners of SSDs (especially for MLC-based drives).

Besides the information on the product home page www.managedflash.com there is only a few tests I have been able to find on the web. However, the benchmarks look incredible.

Check out the following links:

http://www.bigdbahead.com/?p=44

http://www.forumdeluxx.de/forum/showthread.php?t=523043&referrerid=39044 in German, where the following link is also found:

http://img516.imageshack.us/my.php?image=attomftbenchqg5.jpg
(Impressive ATTO results for a raid setup of two OCZ core drives)

If MFT really works, then getting a large MLC-drive and reserving a portion of it for MFT might result in better performance and more storage space than a really fast SLC-drive without MFT.

I am really interested in seeing some exhaustive tests run under Windows XP/Vista. Maybe a project for our resident SSD-guru Les.

 

meh.. if a store in Canada is willing to sell me a patriot warp v2 ssd for under $220 after rebate, I might just pull the trigger and become the guinea pig >_<...
I really hope patriot ssds perform like how they say they would on paper, and hopefully not have any serious write speed disasters like OCZ.

 

I may consider a Patriot myself. I want speed, plus no major issues. Dave

 

Per Intel's website, the part number listed for around $730 is a 5-pack.
http://www.intel.com/support/flash/sb/CS-029624.htm

Each drive would be around $150. Competition, yay.

NCIX
http://www.nxsource.com/products/286626/INTEL/SSDSA2MH080G1C5/
PC Connection
http://www.pcconnection.com/IPA/Shop/Product/Detail.htm?sku=8944049

 

Unfortunately the windows version (not sure about linux) does not allow for MFT on the boot drive =(

 

I'm almost begging that you are right. These 80GBs would be a hands down winner if you are, though it would be kinda stupid if we were forced to buy 5 packs... Heck, it's even less than $150 if you're right

Maybe ill take back my words and get an intel instead of patriot - that's if intel starts selling them in single packs for the same price.

 

That's likely a typo. $700 for single drive coincides with $8/GB rumor. If it really is $150, all other manufacturers are gonna go bankrupt with the performance they are claiming.

Some of the slower drives still sell for more than what Intel drives will go for.

 

Too bad. Hope they upgrade the functionality in the coming versions.

I guess owners of OCZ Core series can try to play with the File Based Write Filter (FBWF) from Windows XP Embedded.

http://www.aaowiki.de/howto:fbwf

They could "protect" the Windows, Program Files and Documents and Settings folders. That and disabling the page file would likely eliminate most of the system stalls due to random writes.

However, I wonder how long it would take to commit all changes to SSD when shutting down the session.

 

Well, it depends which approach Intel wants to take. There are two options - don't sell a lot, but get huge profit margins, or sell everything they can make with narrow margins. I'm inclined to believe that Intel's intent is to sell everything they can make at first to help create the market and gain a dominant position. From there, it'd be possible to increase prices somewhat, but the real advantage of doing so is that it creates a relatively stable market for Intel NAND flash, a fact that should allow that division to be profitable for a change. And if you think that $150 is too low, current spot pricing for 64Gb NAND flash is at a bit over $12, so ten chips would be $120 - $30 for PCB/other chips/packaging isn't unreasonable.

So yeah, I can see either price being the case. I just wouldn't be surprised in the least if Intel decides to take the market by storm.

 

The link in my previous post now only exists in Google's cache.

And the SSDSA2MH080G1C5 part number is now listed as a single item on another page on Intel's website.
http://support.intel.com/support/flash/pssd/sb/CS-029624.htm

The shipping weight of 2 lbs listed on NCIX seems high for a single drive though. (2lbs = 907 g, a bare drive weighs around 100g) ... a similar Patriot drive has a shipping weight of 0.5 lbs
http://www.ncix.com/products/index.php?sku=4355835&vpn=SSDSA2MH080G1C5&manufacture=INTEL
http://www.ncix.com/products/index.php?sku=32999&vpn=PE128GS25SSDR&manufacture=Others

Based on this, I think the price listed could still be for 5.

 

I hope its a 5 pack! Those prices would be most excellent.