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

Ouch. There are a couple of reviews that said it has higher power consumption. Didn't know it would be that much higher.

 

so which SSDs are known to be power efficient?

 

Sandforce, Kingston V+, Samsung.

 

Hmmm, looks great but beware -- no pricing estimates shown there. And when will we get SATA III (6GB/s) controllers in laptops?

 

Hey guys, sorry for the n00b question. I have a gen 1 x-25m 160gb which i'm pretty sure does not have TRIM (its 50nm). From what i have gathered it seems on SSD's with no TRIM the degradation of the drive is a lot quicker but you can help your SSD by not filling it up 100%. I have 60GB free and was just wondering how much roughly i need to leave free? TY

 

The free space has to be unallocated space.... NOT in any partition and NEVER written to. So, you may be out of luck. Shrinking your current partition won't help. Free space within a partition will not be used for wear-leveling nor will any cell that has ever been written to. You could secure-erase the drive and re-partition to leave free space, but the benefits of that are reported to be considerably diminished compared to a new drive.

You might want to watch the Intel IDF video (2009) -- it explains all this in great detail. I don't have the link handy -- (google is your friend).

 

This is total nonsense!

If that is true, and it isn't, than all of us who use all available space don't have any benefit form wear-leveling.

 

Watch the video. Bottom line from what I gathered from it is:

1) wear leveling is effective with factory preset over-provisioning
2) allowing any amount of free unpartitioned space greatly prolongs the life of the drive.
3) 17% free space is optimal from a consumer useage standpoint
4) You can secure erase a drive, then partition for freespace and you will still prolong the life of your drive, just not as much as a fresh drive.

 

You are welcome to go get the information where I got it -- INTEL.

They should know, they make the drive. I never said this applied to any other drives. I only studied the Intels, not the others.

 

Here's my Vertex 2 60GB score:



Is this good, bad, indifferent?

I thought the sequential writes were supposed to be 200+???

 

Run it on 0 fill data. You can set it on 3x 50MB, that's enough.

 

This can be put to the test using Intel's SSD Toolbox. Check SMART Attributes and look at Host Writes for the running total of writes to the SSD. Then, go into hibernation, come back out and check the hostwrites again. That should confirm that it's only the amount RAM used that's writing to the SSD. That can still add up to a lot daily, unless default behaviors to enter hibernation are changed and hibernation only happens by user choice.

UPDATE: I just realized that I replied to a post that was made days ago. Well, in case it was never resolved. That's my $.02

 

There's another way to check it, just check the size of the file hiberfil.sys

 

That too!

 

You can also run an ATTO score to assist.

 

That is not a good check: the size of the file should equal the total RAM that is available to Windows.

Just because it allows that maximum to be written, does not mean it will actually write it though.

 

Thanks, what good does zero fill data do? Is this realistic for regular use? Is random data like worse case scenario and zero fill best case then?

Here it is with zero fill



Posted Random fill again for reference:

 

When you write only 1s or 0s, you basically check how fast can the controller process the data and compress them, with random fill you check how fast the controller can process and write/read to/from the NANDs as well. The real world performance will be somewhere in the middle depending on the kind of data you write to your SSD (how compressible they are)

 

That is correct.

The question here is this:

Who actually fills his RAM to 100% ???

Windows would actually crash before; therefore, hiberfil.sys will never hold as much data as 100% of the RAM one system has.

 

Are You talking about this video?

If yes, can You tell in which part are they talking about wear leveling?

Can You tell me where did You get that information?

Do You even know what wear leveling is for and how it does it's "magic"?

 

Which is exactly my point that the size of the hibernation file does not indicate how much an SSD is written to when the system goes into hibernation.


And, I do use very close to 100% of my 8GB RAM - I just don't hibernate when I'm doing so (nor does Win7x64u crash).

 

So if my hiberfil.sys is 2,8 GB it means there wasn't 2,8 GB written?

 

hiberfil.sys is bit smaller then actual RAM capacity and it takes, in Your case, 2.8 GB of drive but it doesn't need to hold any actual data, but when going to hibernation it writes content from RAM to hiberfil.sys file.

I think it's taking that much of space just to make sure it'll be able to write all content from RAM to drive.

 

How much RAM you got Phil ?

 

I have 4GB RAM. If the file is 2.8GB large, I would assume 2.8GB is written.

 

Well that makes sense.

Hiberfil.sys is never equal to the amount of RAM; it's more like 70 ~80 % of it.

You got 2.8 (hiberfil) out of 4 (RAM) which is 70%;

I got 6.2 (hiberfil) out of 8 (RAM) which is 77%


I guess windows assume we won't fill our RAM for more than 80% of it's capacity...

 

Ok concerning hibernation,

"when the system moves to S4, the power management saves the compressed contents of memory to a hibernation file named Hiberfil.sys, which is large enough to hold the uncompressed contents of memory, in the root directory of the system volume. (Compression is used to minimize disk I/O and to improve hibernation and resume-from-hibernation performance.)
After it finishes saving memory, the power manager shuts off the computer..."

page 637, windows internals 5th edition

 

That was already clear. The question was: if the hiberfil.sys is 2.8GB, did Windows write 2.8 GB?

It would seem logical. Some people say it's not true but didn't give a clear explanation.

 

intel IRST 10.0.0.1026:link removed
be cautious, this driver seems not compatible with PM55 chipset. HM55 compatibility is unknown.

 

For more info concerning the size of the Hiberfil.sys and how to adjust it check here:
http://download.microsoft.com/downl...EA-470B-A97E-CE7CE0D98DC2/HiberFootprint.docx

According to this, it's not necessary for Windows to write as many data as is the size of the file, but also there are cases where Windows cannot compress the data enough for them to fit into that size, so there will be cases where you will write that many data and other cases that you will write less.

 

stamatisx, if I hibernate and resume, and the hiberfil.sys is 2.8GB large on my HDD, did Windows write 2.8 GB?

 

According to the document I attached on my previous post, it's not necessary for Windows to write as many data as is the size of the file, but also there are cases where Windows cannot compress the data enough for them to fit into that size, so there will be cases where you will write that many data (or have to write more than that) and other cases that you will write less.

 

Yes that is the video. If you look at and understand the charts, and listen to what he is saying, my statements hold true. I would not post bad information here. And I will not respond to any more posts which question my intentions or knowledge about the subject at hand. If I am not sure of my information, I will say so.

There is more info about Intel SSDs on Intel's website and on Microsoft's websites. Look around, you will find a lot of info. The part about wear-leveling (TRIM) needing UNALLOCATED space came from either the IDF video or directly from Intel's website -- it was not a questionable source. I had no reason to doubt the information because they make the drives and write the controller firmware.

 

Simply put, Windows will write to hiberfil.sys what (the data) it had loaded into RAM before going on hibernation.

If you had 1.6GB worth of data in RAM, that's what's gonna be written to hiberfil.sys, and so on.

Unless I'm missing something...

 

I wonder if that driver is safe to use with the HM55 Chipset & Intel SSD (non-RAID). I assume it supports TRIM.

 

I will give an example

We have 4GB RAM and a Hiberfil.sys file which is 3GB (75% of 4GB)

If we have 3.5GB of data loaded on the RAM when we press the power button and put the laptop to hibernation there are two cases:
1. either those data are compressible enough to fit into the 3GB hiberfil.sys file (they can be 2.8 or 2.6 GB or 3GB in size) or
2. they cannot be compressed enough and the hibernation will through an error (and we have to increase the size of that file)

I hope that's more clear now

 

just installed the drive, now my windows wont start


edit: running last known working configuration worked.

 

it seems this driver is not compatible with PM55 chip set, i have to reinstall OS for PM55. i will not try it in HM55.

 

I removed the links to non official Intel RST drivers.

Please be careful with installing and posting drivers from non official sources.

 

So uh, I am using 1.6GB RAM right now with my OS running and some tabs open. I have a 5400 RPM 320GB hard drive which... at BEST has maybe 40 MB/s write (sequential!!!), but I can hibernate my computer in about 10 seconds, and resume it again in less. I am pretty damn sure that it's not writing everything I have in memory to HDD as is.

 

Have you timed how long it takes to hibernate your system? 10 seconds is unusually low for a 5400rpm hdd.

 

If your data are compressible enough to occupy maybe 400MB and your sequential speed is about 40MB and it takes you about 10s to hibernate, maybe it does.
On the other hand I have never used hibernation and never checked what is written, how much and how fast. I prefer the sleep mode.

 

Guys...

What's wrong with this one ?

I got an M17X-r2, which indeed is powered by the Intel PM55 chipset, and I did not notice any problem..

Thanks for your lights !

 

Any difference concerning performance?
Did the installation files contain any info regarding the changes on this version?

 

Nope !

Nope !

Even the pdf (on the site) is not the right one; it's regarding previous version

 

i tried in HP which has PM55 chipset, it rebooted into windows but stuck there for a while then got BSOD. then i tried for a USB boot repair but found OS is gone (in x-25M 160G).

 

That is most interesting as I've never ran into a situation where windows did not have enought space on hiberfil.sys to fully write what's loaded into RAM...

May I ask you if this is only theorical (kind of readings from Microsoft) or did you really indeed face that situation, no matter why (too small hiberfil size or unable to compress data...)

Interesting indeed....

 

Okay, I guess I need some clarification here about the "alleged" reserved space for wear-leveling on Intel SSDs.

Please see my post HERE in another thread.

Thanks.

 

Wow, almost interesting (no joke...)

Can't speak for HP, but I know Dellianware's M17X is a piece of crap; it happens so many times with no apparent reason that if I restart, I sometimes get no video at all, have to force-shut-down, than force-power-drain, all those funny little things that you would just love to do for the rest of your life if only you didn't have to eat, sleep and others...

After my first reboot upon installing this one, got that, so did what's said above and now everything's OK...

I always took my IRST from this site (driverStation) and never had any problem with them... ...but maybe it's just me...

Anyways, I did not loose my O/S, I am clueless here...

Sorry not to be more helpful...