Very good article. It's a must read for anyone interested in SSDs.

This is due to long erase time of NAND Flash used in SSDs, MLC or SLC. A blank storage cell is consider one when in it's initial state. Zero data is written into the cell by placing electrons into the so called floating gate. So cells with one can be written to zero but cell already written to zero cannot be reprogrammed to one again. The programmed zero cells can only be restore to one state again by an erase operation. This is removing electrons from the floating gates and usually done in blocks with multiple rows.

When the SSD is filled, the internal controller start to reclaim available space by erase. So performance will slow eventually. More advanced controllers will do reclaim/erase in the background so there will be available space when write occurs.

While large manufactures with long history such as intel, Samsung, M-Systems (acquired by Sandisk) incorporated these technologies; it is not clear smaller manufacturers are able to acquired controller with background erase/reclaim function.

The faster write time in newer SSD's likely come from data being written in parallel, to two or four chips at the same time. But it won't improve erase time much.

Large data write is fast because all data is written at the same time to one or multiple rows in the chip memory array. Small data write is slow because the data size is smaller than the row size. The controller then do a read row, modify, write row operation; read the entire row including existing data, append new data to the end of existing data, write the entire row with new data back to the memory array again. Most problematic is the FAT area where lots of small read and write are occurring. This can be improved by using NOR flash for FAT. But this adds to cost.

HDD is not affected by this erase problem so read and write time is almost equal.

nice article from Anandtech, as usual...

jellycow's results show a 4KB Write Speed of 2Mb/s. That's an important number...

The question I have is: is that "RANDOM write"???
If that's the case, than the Kingspec drive performance is really good when compared against the '"faulty" JMicron equiped SSDs...

Look:

4KB Random Write Speed New "Used"
Intel X25-E 31.7 MB/s
Intel X25-M 39.3 MB/s 23.1 MB/s
JMicron JMF602B MLC 0.02 MB/s 0.02 MB/s
JMicron JMF602Bx2 MLC 0.03 MB/s 0.03 MB/s
OCZ Summit 12.8 MB/s 0.77 MB/s
OCZ Vertex 8.2 MB/s 2.41 MB/s
Samsung SLC 2.61 MB/s 0.53 MB/s
Seagate Momentus 5400.6 0.81 MB/s -
Western Digital Caviar SE16 1.26 MB/s -
Western Digital VelociRaptor 1.63 MB/s

Both the 512KB and 4KB tests of Crystal Disk Mark is of the "random"-nature according to the author.

http://crystalmark.info/software/Crysta ... dex-e.html

The interesting thing is that, according to what I've read, the Kingspec SSD do use the JMicron controllers, at least for the 2.5" models if not also for the 1.8" models.

However, I don't have one to crack open and check.

It could also be a (positive) side-effect of the SATA<->PATA bridge on the 1.8" SSD. (The JMF602 is a SATA controller AFAIK.)

Random vs sequential in flash will not matter much as they have no seek time. They work by addresses and access time is same over the entire flash chip.

Are you sure about that?

http://www.anandtech.com/cpuchipsets/in ... i=3403&p=8

I think the article you referred to is a drive level, after the controller, not the chip level.

The row size on a NAND chip is may be 256KB. So 4K random or sequential will be smaller than the row and it will be written into the whole row regardless.

The drive level data in the article does point out similar things. For instance, it took the same time for the JMicron to write 4KB and 128KB. May be this is because it does a write to a single row on the chip regardless of data size. In the intel case it looks like the minimum write size is 8KB and not 4KB. So the design philosophies are very different between the two.

At the drive level, intel has same 4KB 100% random and 100% sequential performance. This is almost what I would expect on a chip level.

For the JMicron, it looks like the NAND chip performance is very bad. But it caches multiple 4KB sequential writes to gain speed.

Isn't the drive level what really matters?
We should focus on discussing real world performance considering real usability...
I'm sorry but don't understand your point.

Sure, to understand what the drives are doing and how they differ in performance, it is helpful to understand the function of the chips inside. The controllers take different tacks in getting around problems of the flash chips, resulting differences in drive performance.

I recently purchased a KingSpec 32GB SSD replacement for my X40. Here are the results of benchmarking with CrystalDiskMark 2.2. All Figures are MB/sec. The read numbers are consistent with those reported by HD Tach.

Original Hitachi HTC426040G9AT00

Seq Read 15.12
Seq Write 14.14
512K Read 10.72
512k Write 11.61
4K Read 0.314
4K Write 0.674

Replacement KingSpec PA18.1-032M SSD

Seq Read 64.66
Seq Write 45.04
512K Read 58.09
512k Write 23.55
4K Read 13.52
4K Write 1.718

I've experienced no slowdowns or freezing as discussed in previous posts. The performance gains are impressive. The numbers speak for themselves, and the X40 really flies.

funny I didn't experience such a major improvement

You may try FlashPoint write buffer driver on SSD to improve overall speed (currelntly only under XP):
http://forum.thinkpads.com/viewtopic.php?p=502632#p502632

And this is realy incredible software.
JW

So what is the verdict on the kingspec 1.8 ssd's with an X41?

Would I regret not chosing a fast but lower capacity CF drive? Will there be any hangs while using the computer day to day (and over a decent amount of time)? Im extremely fed up with this *****Expletives removed by Moderator***** 4200 drive...

Another experience: Kingspec 1.8" ZIF on Latitude XT...

http://www.gottabemobile.com/forum/foru ... p?TID=7758

Got my 64GB Kingspec drive today. My benchmarks are higher than most any other Kingspec benchmark out there...

My sequential read/write speeds are on par w/ that of the ASAX disk.

hello. i am new to the forum- i actually joined hoping to get some help installing a kingspec ssd in my x40.

the ssd has been successfully cloned but the thinkpad will not recognize it. I upgraded bios (2.08) and it lists the ssd but it will not boot from it- just a blank screen after the initial thinkpad sign.

Is there a bio setting change required? did i miss a step? any advice would be very much appreciated. thank you in advance!

How did you clone the disk? Did you do it on the X40 itself?

I just went through a rather frustrating experience before I managed to get an X31 disk cloned. And the X40 would likely be very similar ...
viewtopic.php?f=8&t=75853

thank you for the response.

i took into a computer shop and they removed my original harddrive and cloned the new one using some external device. Is the cloning process the problem?

Is the drive set as active? Did the cloning process clone the boot sectors?

I cloned w/ Acronis True Image Home 2009, which failed. Ended up having to use a borrowed copy of Ghost 2008.

In my experience, it was, and my symptoms were very similar (or possibly identical) to yours. This is what worked for me:

• Move original HD to a USB housing (not sure what kind of adapter you'd need for the X40 drive, sorry)
• Install new HD into internal drive bay
• Connect USB drive to X40
• Boot X40 into bootable Acronis or Ghost CD using either a USB CD/DVD drive or an Ultrabase CD/DVD drive
• Do the cloning
• Remove the extra drives, reboot into Windows!

The shop should be able to do this for you as well ... they'll just need to know that the X40 drive layout is sufficiently "special" that the cloning needs to be done on the system itself.

SSD like CF are probably 3.3V/5V compatible. Most USB housings and Ultrabay drive trays are 5V. X40 HDDs are 3.3V and not quite sure how long they will work on 5V.

Quite true -- but given that the shop previously cloned the loose 1.8" HDD for the OP, I am assuming that they have a suitable converter available.

thank you all so much!

i just dropped off the thinkpad again at the shop with your suggestions. hopefully itll work this time

Hi, i'm observing this thread very closely. I have X40 & X41. one of the issue is the cost of the SSD. Anyone can provide some details on the current pricing of the SSD?

I've seen the KingSpec 16Gb version for ~130$ usd and the 32Gb for ~160$ usb, both on ebay.

Hi,

I emailed Kingspec directly for a quote, this is their reply:
(i include the email address of Kindspec representative)

"
Wed, 06 May 2009 11:36:38 +0800
From: Zhou Yu <allen.zhou@kingspec.com>

x40/41 should be with PATA interface.

Quotation as below,
KSD-PA18.1-032MJ USD103/PC
KSD-PA18.1-064MJ USD187/PC
KSD-PA18.1-128MJ Not available
Freight cost for 1 PC from China to Singapore:
EMS:USD16
DHL:USD15

Thanks and best regards!

Allen/KingSpec
"

The pricing is much cheaper than from eBay. The only hiccup is the payment method. Need to pay to their Bank account!!

update: cloning worked!!
love the x40 with the ssd. its SILENT, cooler, and much faster!
thank you all for the help!

Yes, in the end, it took a "Moderator Emeritus" to come to the rescue.

Hi,
did you get the 2010 boot error?

I'm too planing buying a Kingspec 64gb for a x40.
And also for a Sony Vaio that uses a Toshiba PATA 1.8" hard drive.

Any feedback is welcome.

Regards,
Oliver

The boot error happens on the X41. There's no boot error on the X40. I just installed a Kingspec 32GB on my X40 and it works flawlessly.