6_6_6

Intel X25-M Impressions

Recommended Posts

Well then...

6 months of torment with this SSD crap... and thankfully i got kernel panic in vmware with linux 2.6.30 and subsequently withing few mins KERNEL_DATA_INPAGE_ERROR and BSOD on 2008... Reboot no go. So it was time to put a stop to this mess...

Let's run the usual tests since it is offline...

MHDD read... Hmm... stuck somewhere after few thousand LBAs. Let's change cable and SSD port from 1 to 2 just in case (SSD has always been on Port 1, no changes to machine). Voila! it runs now. 30% decrease in read speed with lots of spikes and too many weak sectors. But workable.

Write... I didn't have so many slow sectors even on dying TB drives!

This drive passed initial checkups when i purchased it, read and write was fine after light stress testing. So condition of the drive detoriated with almost no use!

It has only been used as an OS drive (no superfecth/prefetch/hibernation/pagefile/restore/defrag/indexer/sys rescue), nothing has been written, copied or done to it (it would freeze momentarily anyway on few GB copies). It has 8820 firmware that supposedly did not require any manual garbage collection work.

After 2 passes of zerofill and secure erase, its health improved and it is usable now. SMART shows only 8% of its life gone with 92% remaining. No reallocations. It still has weak sectors.

As I mentioned at the very beginning... SSDs are good only for Linux LiveCDs.

Intel also abandoned Gen 1 SSD users... They are not to provide firmware updates for TRIM command. Thank you Intel for your unintentional kindness, I guess a running SSD without TRIM is better than a bricked one! (Intel pulled the firmware same day after everyone got their Gen 2 drives bricked!). Not to mention data corruption issues when Gen2 was first released in July.

x25m-2.jpg

Share this post


Link to post
Share on other sites
CONCLUSION: No benefits to power user. Or not what you would expect.

I don't care about the IOPs or whatever tests people run at this, but do not fall under the false pretense that 11ms versus 0.1 ms make a worlds of difference. IT DOES NOT. As opposed to my Seagate 7200.11 750GB (more than 1 year old), i see 20-30% improvement which would probably be obtained by using a newer mechanical drive.

<SNIP>

Overall, big freaking disappointment.

I have no doubt people will be giving you a lot of stick in this thread for this post.

I would however like to confirm your thoughts on this.

I recently added 2 of the 120GB OCZ agility disks to my PC, I can read over 400mb/s write over 220mb/s and random is over 20mb/s write.

That being said, in 'real world' usage, as someone who has a pretty powerful PC and knows how to optomise it the performance difference really isn't that great.

(I can load Crysis now in 24 instead of 34 seconds and a few other things, sure)

The thing is it's not half as pronounced as I expected, and while I would no longer EVER recommend a ''performance' platter drive like the raptor I wouldn't say I could recommend an SSD yet.

It IS faster but at the current prices and with the articles out the last 4 weeks and rumours, they are working hard on many many new controllers for the things (ONFI 2.0, SATA 6GBS) I have a strange feeling about June 2010 SSD's will be substantially cheaper and substantially faster.

I realise that's obvious, but as a brand new tech it's evolving much faster than other technologies

I would NOT buy an SSD now, I would buy a decent 7200rpm WD drive, learn how to optomise /temp/ /pagefile/ and use a decent defrag tool - patience is a virtue and 2010 is the year of the SSD, not 2009.

EDIT: Also consider a readyboost disk, I'm surprised to admit it works fairly well.

Edited by AbRASiON

Share this post


Link to post
Share on other sites
CONCLUSION: No benefits to power user. Or not what you would expect.

I don't care about the IOPs or whatever tests people run at this, but do not fall under the false pretense that 11ms versus 0.1 ms make a worlds of difference. IT DOES NOT.

I totally disagree ! Using some SSD on one small hugely loaded database, I can tell you that the difference is just HUGE...and this "random usage" price/performance ratio is really impressive.

Although, I use a X25-M 160GB G2 on my desktop and everything is instantaneous. I was not expecting such an improvement.

==> I won't "troll" which sensation/expectation is the best...and I agree that the "desktop usage" price/performance ratio of good SSD are not impressive enough for a very large audience to invest on it...I did it anyway and it rocks !

Share this post


Link to post
Share on other sites
I totally disagree ! Using some SSD on one small hugely loaded database, I can tell you that the difference is just HUGE...

Nobody disputed that. But we are talking about desktop usage. What is a db doing on desktop computer?

Although, I use a X25-M 160GB G2 on my desktop and everything is instantaneous. I was not expecting such an improvement.

Did you change the cluster size as you suggested before or is it at default?

How long do you have the drive? Are you doing any writes on it? What does SMART attributes 225 and 233 say?

Say you copy a large file ~20GB... Is the system functional throughout the operation?

Share this post


Link to post
Share on other sites
Say you copy a large file ~20GB... Is the system functional throughout the operation?

According to what i've read from reviews, like the ones at Anand's, that is exactly why you should get an SSD (a newer one with TRIM that is). Superior multitasking performance??

I was contemplating getting an Intel 160GB SSD Gen2 for VM testing, but such threads as this make me think twice.

Anyone tried an SSD with older chipsets like nForce 4 as i have myself?

Share this post


Link to post
Share on other sites
According to what i've read from reviews, like the ones at Anand's, that is exactly why you should get an SSD (a newer one with TRIM that is). Superior multitasking performance??

First, I didn't have superior of anything when the drive was new (drive performed at its advertised 250MB/s - 80MB/s read/write speeds). See benchmarks and comparison to Seagate at the very first post.

Second, people were promised TRIM suppord when they shelled out $800 for this crap. Intel didn't do this right. It is not a technical limitation, rather then their greediness to suck more from people by selling more of their crap in improvised versions.

Third, these drives do not need TRIM support to start with. Fine when it is there, but it is not end of the world when it is not. The drive should have performed fine. They have their garbage-collection routines and that is to take care of performance degredation. Furthermore, Intel upped firmware to 8820 and it supposedly fixed any issues with garbage-collection.

Fourth, this drive did not have anything being written to it at all. No pagefile/hibernation/sys restore/indexer/rescue/defrag, etc. It was pretty much used as read-only except hte few log files that 2008 server updates. Its garbage collection routines should have had no trouble.

Share this post


Link to post
Share on other sites

Still... I'm itching to get my hands on one and test for myself :ph34r:

Anyone tried SSD on nForce4/Win7? I'm afraid an SSD upgrade will get me into a mobo/cpu/ram upgrade aswell... wich i really don't need.

Maybe i should just get a couple of Vraptors in Raid0 for my "speedstorage" needs instead. I even have a proper adaptec sas controller for that <_<

Share this post


Link to post
Share on other sites

Don't bother getting one until they come down 50% in price and increase 200% in speed (so 12 months)

Seriously, I have 2 in RAID0 and the difference is negligable if you know how to configure a machine to be speedy and have good specs.

Maybe on a slow, crap laptop with 1gb ram, no worries but a 6gb desktop with 4 cores and Windows 7 with readyboost the difference was very very small.

Share this post


Link to post
Share on other sites

My 150G velociraptor isnt noticeably faster than my 1.5tb seagate (the one with the FW issues) in a desktop environment

But I spent good money on it, so I'm gonna keep using it but never recommending it

I guess the OP is in a similar situation, but with a SSD instead if a 10,000rpm drive

Read speeds are a thing of the past as most drives sequential reads top 100Meg/s easily. Whats more satisfying is getting those speeds across a gigabit LAN

Share this post


Link to post
Share on other sites
Maybe on a slow, crap laptop with 1gb ram,

I would agree. It is godsent for laptops. But then again, i always refrained from using laptops because it felt like a was using something from a decade ago.

One a sidenote... Few rounds of zerofills made noticeable difference. It is no longer hanging or lagging.

So anyone who uses one needs to wipe out his OS and secure erase the drive EVERY WEEK if they want trouble-free operations.

So please add the time to erase the OS, shuffle the drives, secure erase it and reinstall the OS every week to your random IOPs or whatever benches you are running too. Because I didn't have to do any of these when I was running hard drives for decades.

Share this post


Link to post
Share on other sites

I've heared that before about the raptors. But i can tell the difference. It's better when you do lots of i/o. For web surfing and booting your puter, sure... not much of a diff, but you would not expect that either.

What i wan't to use the SSD for is VM's. If i have 4-5 VM's in VMware workstation running on a Intel Gen2 SSD, the snappyness and speed when working with these VM's should be in a different world than what i am experiencing now.

Theoretically anyway.

With VM's on a normal 7200 harddrive, the speed and feel of these systems goes to a crawl when all are working against the disk at once.

Also:

There are several in here that states: "if you know how to configure you drive for ncq, swap etc.."

It would be nice if these persons could explain a little more around this. What exactly do you do to optimize disk performance?

:)

Share this post


Link to post
Share on other sites

Cross-posting this from NCQ thread since it involves this very same Intel X25-M.

---

I am gonna bring this topic back from dead... It has been 2 years since it is started. And I still don't have a replacement for my aging Seagate 750GB.

To illustrate you what I mean, here is a 2 mins video example. Quality is not that good, so much from a webcam... But you can see clearly.

Here are the contentenders:

Seagate 750GB (the same drive that made me start the topic)

Intel X25-M 80GB

WD 1TB (Green/Black/Blue? don't remember)

Test notes:

- All drives are connected as secondary drives. They are not booted from.

- All drives are empty, clean, perform at the speeds they advertise.

- Clean booted system. Task manager shows at all times CPU usage is less than 20% on all cores.

- Same movie is played. Same files are copied.

- First run is done on Seagate (so you wouldnt tell me it is cached, memoried, blah blah). 2nd on Intel, 3rd on WD (WD not shown to make file size smaller, but it is the same as Intel performance)

Here is what I did:

1. Copy 10GB file from the test drive onto test drive (no difference in results if it was copied to another drive)

2. While copy is in progress, start a 5GB 720p movie on WMP11/VLC... And try to seek through it.

RESULT:

Seagate is fluid and instant. When you press the slider, you get immediate response. No stuttering, no pausing, no nothing. They way things are supposed to be.

Intel/WD on the other hand... Sometimes you have 4 second pauses! Movie stutters, then frames rush to catch up... Overall... Impossible to watch it.

Now... Is it to much to ask the capability to watch a movie when the drive is doing anything else like a simple file copy operation?

So folks... Test the same on your hard drive and tell me if there is any other drive that would not stutter doing this. Because I haven't come across one except this Seagate 7200.11 750GB (even 7200.11 1TB wouldn't do it).

Seagate-Intel Video

Edited by 6_6_6

Share this post


Link to post
Share on other sites

Who on earth needs to watch a video and seek through it while copying files??

Try actually USING it as you would normally, saving photoshop files for example, etc. that's when they shine - especially the Intel X25-M.

If you hate SSDs then fine - but don't go on about how bad they are, because they are not.

Share this post


Link to post
Share on other sites

This is one of the most basic things a person can do with his computer system... it implies multitasking capability.

My topics are intended for mature audiences. Please stay away from them.

Who on earth needs to watch a video and seek through it while copying files??

Share this post


Link to post
Share on other sites

This is one of the most basic things a person can do with his computer system... it implies multitasking capability.

My topics are intended for mature audiences. Please stay away from them.

It's basic but:

a ) no one would ever do that in normal use

b ) its pointless

And mutltitasking - NOTHING EVER multitasks in a true sense - it just switching between different tasks.

What you would have to do is use some programme that uses a Cache on the SSD/HDD and run some standard operations with that.

Saving large files in Photoshop, modifying large files is an example.

Because something is basic its by no means accurate or useful.

Edited by DetlevCM

Share this post


Link to post
Share on other sites

The video is not there any more. However, if it's as bad as the author writes then it's a matter of serious concern. How is a drive going to handle simultaneous requests if it can not even handle such a basic task? And don't bother with the definition of "simultaneous" - if I have to wait for 2 tasks to finish it's multitasking / simultaneous.

Edit: I just tested myself on a WD 640 Blue. Not freshly booted, file copy within the same drive and then trying to watch some HD video in Zoom Player 7.. which was OK, but jumping within the file was absolutely horrible.

MrS

Edited by [ETA]MrSpadge

Share this post


Link to post
Share on other sites
MrSpadge' date='19 May 2010 - 08:28 PM' timestamp='1274297337' post='261827']

The video is not there any more. However, if it's as bad as the author writes then it's a matter of serious concern. How is a drive going to handle simultaneous requests if it can not even handle such a basic task? And don't bother with the definition of "simultaneous" - if I have to wait for 2 tasks to finish it's multitasking / simultaneous.

MrS

Multitasking implies its done at the same time - and a computer can't do that.

Except run two threads on 2 cores etc - but if 2 tasks run on one core at the same time it's switching between them - so its not multitasking - even if it runs several processes at the same time.

And your comment gave me an idea - queued reads/writes - Intel SSDs handle them brilliantly, and that's what happens if you use most programmes.

Compare Photoshop CS4 on IDE mode vs. AHCI - saving a 170MB HDR image is quicker on AHCI due to queuing.

Share this post


Link to post
Share on other sites

Dude, stay the hell away from my topic. I requested politely twice -- once here and once at the other topic. Do you want me to spell it for you?

It's basic but:

a ) no one would ever do that in normal use

b ) its pointless

Share this post


Link to post
Share on other sites

Please try now in the original link. I added the host apparently twice there.

Now see the first video and judge yourself how it is seeking through the video. This was an example, in that drive, no matter what I do at the background, it is always responsive with the task I am dealing foreground. There is no other drive that gives me that responsiveness... including the SSDs. Every other drive locks the system when there is any operation that requires drive access.

Especially watch 1:45... when I click on the slider... it takes 5 whole seconds for the new frame to arrive (1:50)... and another 5 seconds till 1:55 for the normal pace of movie to catch up... That is Intel SSD... And that is a whole 10 seconds which is instantenous on Seagate.

MrSpadge' date='19 May 2010 - 02:28 PM' timestamp='1274297337' post='261827']

The video is not there any more. However, if it's as bad as the author writes then it's a matter of serious concern. How is a drive going to handle simultaneous requests if it can not even handle such a basic task? And don't bother with the definition of "simultaneous" - if I have to wait for 2 tasks to finish it's multitasking / simultaneous.

Edit: I just tested myself on a WD 640 Blue. Not freshly booted, file copy within the same drive and then trying to watch some HD video in Zoom Player 7.. which was OK, but jumping within the file was absolutely horrible.

MrS

Edited by 6_6_6

Share this post


Link to post
Share on other sites

@video: the Intel SSD behaviour is exactly how it looks like for me with a different HDD (AHCI enabled), different test files and a different movie player. And I even didn't mention yet the couple of seconds it takes the system to catch up with playback again. Even the sound is screwed for these couple of seconds.

Detlev, don't be ridiculous! You're not helping here. The point is that I, as a user, totally don't care if the machine is doing things really simultaneously or if it is switching between tasks in ns or µs intervals - I couldn't tell the difference anyway. It's enough if both tasks are running at the same time in my perception. I don't want to have to stop/pause my file copy to be able to search a scene in a movie - that would be singletasking.

Since the Intel SSD easily has the access time to make the video searching fluid, even in the presence of the heavy file copy, and the Seagate hardware should be much less suited to the task we've apparently got a software problem. I think the problem is I/O priorization. It's not that the Intel couldn't make the video search more fluent. It's probably that it's trying too hard to get that file copy done. It doesn't know what's important to the user. And how should it ever know this? I have no idea how Seagate is managing to do it right on your 750 GB 7200.11 (and not in later models, as you said).

Perhaps it's trying to balance the achieved transfer rates? I.e. if there are 2 different task streams (copy and video) it tries to give both of them 1/2 of the maximum achievable transfer rate. Which would fill the video buffer almost instantaneously, enable smooth searching and afterwards allocate more bandwidth to the copy operation, since the video doesn't need 50% transfer rate once the buffer is filled?

A different approach would be to optimize for overall transfer rate. That would look better in benchmark, would probably be better for servers and would probably get the file copy done quicker. But at the expense of rendering the system almost useless for the user under heavy I/O load, i.e. trade perceived performance for measureable performance. That would be a sensible approach.. it just doesn't lead to the best user experience.

Brian, care to ask Seagate about this? Maybe get the firmware version of 666s drive. It could be that HDD manufacturers are optimizing their drives for maximum scores in the usual benchmarks (they have to do this in order not to look like total morons), but we as a whole community are simply running the wrong benchmarks.. and thus make them give us products which don't give us the best user experience. Measuring perceived performance is of course more difficult, but I've already seen NCQ tests somewhere where they measured the time it took individual tasks to complete in multitasking environments. Not sure where, though. But the NCQ implementations certainly differed.

MrS

Share this post


Link to post
Share on other sites

That's a good starting point - 666 what are the firmwares of the drives you're testing with?

Share this post


Link to post
Share on other sites
MrSpadge' date='22 May 2010 - 06:15 AM' timestamp='1274526952' post='261894']

@video: the Intel SSD behaviour is exactly how it looks like for me with a different HDD (AHCI enabled)

It is pretty much the same with every hard drive except Seagate 750GB 7200.11 SD15. This is not an Intel thing, Intel is just like any other hard drive.

I will give you some examples below. After this, I think it would be more apt to move this conversation to its appropriate thread for Seagate if you agree:

Let me give you some notes and graphs:

1) The only sureproof way for me to quantify this was to run 2 instances of HDD Speed at positions 0% and 50%. Seagate drive has alternating spikes in the graph hitting 0% and 100% at times... All the other drives have a uniform graph.

Here is a run of 2 instances of HDD Speed. Seagate on the left side (left up and left bottom), Intel at right side (right up and right bottom).

NOTE: I have encryption turned on Intel drive right now, that is why I have 170MB/s total. I just grabbed its graph to show you a comparison. Normally it hits 250MB/s with 2 instances but graph lines are the same. Video and all other tests on this drive was made with no encryption.

seagate-intel.jpg

Here is the same story with a 1TB WD SomeColor (dont remember which). WD 1TB behaves same was as Intel.

seagate-wd.jpg

2) AHCI: Seagate 750GB 7200.11 SD15 behaves exactly like the right side above as with both Intel/WD when AHCI is turned off.

3) Firmware: One user noted that he lost the responsiveness of his system when he flashed Seagate 750GB 7200.11 SD15 with a newer firmware.

4) Video: I choose the video example above for the following reasons: a) It is a very simple and common task (a file copy and watching a movie) B) It is easily reproducible at any system. c) It presents user with quantifiable experience.

You can substitute this with anything you do on your system that would require simultaneous hard disk activity like doing a virus scan, encoding a video, zipping or anything. The results are the same. Seagate is ALWAYS responsive, other drives stall.

5) Model: Same line of Seagate 7200.11 with 1TB capacity is not good either. I have Seagate 1TB ST31000333AS with CC1H firmware. Not good. And I am not even talking about a different model of Seagate.

6) Speed drop: All the drives I tested have sharp speed drops on 2 operations. Most drives fall from 100 MB/s single operation to 20 MB/s in 2 instances of HDD Speed. However, Intel does not drop at all just like Seagate! It is always the same at 250 MB/s no matter how many instances of HDD speed I run. But nonetheless, Intel does not perform even though there is no speed decrease.

7) Whenever I have access to any system, I just do this simple file copy / video operation. I have tried in a whole lotta system and NONE of them are like Seagate. All stall.

Since the Intel SSD easily has the access time to make the video searching fluid, even in the presence of the heavy file copy, and the Seagate hardware should be much less suited to the task we've apparently got a software problem. I think the problem is I/O priorization. It's not that the Intel couldn't make the video search more fluent. It's probably that it's trying too hard to get that file copy done. It doesn't know what's important to the user. And how should it ever know this? I have no idea how Seagate is managing to do it right on your 750 GB 7200.11 (and not in later models, as you said).

See #5 above. It is not even a different model!

I am exactly thinking like you do. I think you are right about I/O prioritization. I think this is a firmware issue... and it works when AHCI is activated on Seagate 750GB. Item #6 above would verify your observation about Intel.

Share this post


Link to post
Share on other sites

That's a good starting point - 666 what are the firmwares of the drives you're testing with?

SEAGATE ST3750330AS 7200.11 (750 GB, 7200 RPM, SATA-II)

SD15

Other drives dont matter since I did not come across one that would run like Seagate above.

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now