I have good news to report, I have just spent 15 minutes on the phone to a Samsung tech guy (HDD department), they are now looking into the problems I have experienced and will get back to me ASAP (a few days).

I explained all of my problems to the tech guy from Samsung that Samsungs test tool and SMARTMonTools show faults and many end users on forums have also had problems. All of my own testing and other apps dont show faults, this does not correlate with the results from Samsungs test tool. The drive apears to be working, other apps other than SMARTMonTools can read the SMART data and say everything is OK, I have copied 264GB of data to the drive and I am currently running a drive "error scan" via HD Tune (20% so far with no errors).

I am hoping that my drive is perfectly fine as it appears to be and that the 2 programs that find faults are in the wrong.


Andy

For what it's worth, that's precisely what I've been thinking since the beginning. I just think that current test software isn't able to correctly monitor those new F1 drives, and that it's the software that needs updating, not the drives.

So until all this becomes crystal clear, people, please STOP pretending that F1 drives are faulty when you don't have evidence of it (obviously user comments from Newegg can NOT be considered as evidence).

@andyb: Thank you for your information That's really good news

I've also considered that possibility (to be honest, I'm really hoping it is that, then I can buy a 1TB drive finally), however the newest version was released after the release of the F1, and the website lists it supporting F1s.

Either way, I'm holding off until Samsung either fixes their drives, or fixes their program.

Hi all,

I'm one of the people who have submitted negative reviews on Newegg. I ran across this post while doing some searching. Let me update you guys a bit since it may help you.

I'm running on an EVGA Nvidia 680i motherboard and bought two drives and found errors on both drives when scanning with Hutil 2.10. On one drive I found both ECC errors and "Check M.C" errors and the second drive just the "Check M.C" error. I have RMA'ed the first for replacement, but have been using the second drive with just the "Check M.C." error since the first of the year and without any problems. My thought is that the v 2.10 of the Hutil application is incompatible with the HD103UJs even though Samsung's site says otherwise. I'll be interested to see if andyb can come up with any answers.

I've also noticed that Newegg has had my RMA'ed Samsung drive for 4 days now and has not shipped out a new one. I don't know whether to read anything into this, but they're usually quicker on the turn around. I wonder if they've flagged the product for quality control issues? Purely speculation, though.

I'll let you know how things turn out with the new drive.

I should also say that there may be precedence to the theory that there is some sort of incompatibility between the Nvidia chipsets and the HD103UJs or maybe just with Hutil. I'll drag out my Via-based chipset computer this weekend and see what Hutil says.

I'm also a newegg customer... though I haven't reviewed the product yet, I've gotten 2 drives in a row (rma'd one for the second and they waited 2-3 days before approving after receiving the drive) that have this same m.c./ecc error. The previous drive did find an error when doing a full read scan but after zeroing w/ hutil it didn't do it again. The second is still being zeroed and then will be tested fully with the surface scan.

This drive is being tested in another computer but both boards are nvidia chipsets (different chipsets though)... I called techsupport and they just told me to take the drive back. I asked about nvidia chipsets being the problem and he said maybe since I got 2 "bad" ones in a row with same error and otherwise working fine... apparently (there was that bad block...). He also noted this being the first call he received about this type of problem. Either way he told me to return it for another (again >_>) and this was after him talking to another person for 5 minutes. It also took him a while to report back to me on what the error meant which was that it was a problem with the drive's memory.

On a side note, on the first one, I was copying over many many gigabytes of data onto it and the filesystem (xfs) crashed on it once during the operation... recovery proceeded fine and I didn't get any errors after that but it's weird and possibly one more point towards me not liking samsung hds.

Re: the mc error on only nvidia systems. I think that is a false rumor. My friend who got the mc error did not have an nvidia system.

Can't even run HDUtil here, it doesn't seem to like my SATA DVD drive.

OTOH, my 1 TB F1 installed Vista x64 flawlessly. I can barely hear seeks with it screwed into the grommets on my P182 (stretch magic awaiting installation). 5.9 on the Vista score (yay! /sarcasm). Acronis has repartitioned a bit with no probs as well.

Very happy with the drive so far.

Keto, no, Hutil won't run off of a SATA CD\DVD drive. I had to go and find a IDE drive to load it off of.

I have an IDE DVD drive sitting around in my old computer, but given the information above, until I get errors in my installation, I'm not going to bother dragging it out.

Just FYI, after a full weekend of work, everything is working as expected here. 3 Samsung HD103UJ in a Debian 4.0 software-controlled RAID-5 array :



Hardware specs :
- Motherboard : Gigabyte
- Chipsets : Intel 945P and ICH7
- Processor : Core 2 Duo E4300
- Drive : SATA DVD-R from LG

All the disks first went through a full RAID-5 synchronization that took nearly three hours, with no errors. After that, we did a pretty extensive R/W test during nine more hours (based on lots of randomly-generated 1GB files), with a md5 check at the end on all the files. Not a single error. Finally, we copied 466 GB of real data into the RAID-5 array, with no errors either.

The disks are recognized by the Debian SMART monitor (lenny version - be sure to update, and be sure to check the SMART option in your BIOS menu of course), AAM has been successfully activated (AAM 128 - quietest mode - very impressive ! ), all the temperatures are OK. See the pictures : disks outside the PC, with no aircooling at all, their temperatures were 37°C to 40°C during all the intensive testing process (just 3°C increase after all the hours of R/W).

FYI, we are using a 256-bit encryption method (luks) all over the array (every single byte of data is encrypted, a passphrase is required to access the info). That means everything has to be encoded/decoded in real time by the processor, thus dramatically slowing the performance of the array (even with a Core 2 Duo). But like I previously said, performance is not an issue here, this array is only for storage purposes. The R/W array performance with real-time encryption activated is 40MB/s, which is good for us. The R/W array performance without encryption is 180MB/s. Finally, the R/W performance of a single drive is about 100MB/s.

These disks are everything you can expect. They are good, they are reasonably cheap (in the US), they are fast, and we haven't encountered any problem with all the nine we have bought from NewEgg. I thought it was worth the report, so people can stop whining and complaining. So if you are getting any software problems with these drives (drives not recognized by a software or a particular chipset, errors indicated by a software...), be sure to update your software or your BIOS first, or contact the developers and wait for fixed software/BIOS versions, before incriminating your drives. Generally speaking, most computer problems are not caused by hardware products, but by people who haven't succeeded in using them with the right software. I'm not accusing anybody here, I'm just saying that we need to be careful before making strong assertions like "the new F1's are not ready", etc. Thanks in advance.

So far so good...

Do you guys think that 750 GB F1 is better choice for performance than 500 GB P166?

Jose, thanks for the info.

If your raid array permits running HUTIL somehow, I would be very grateful if you did, and reported your results here. should they wind up positive, I will be ordering the F1 soon.

Yes, absolute.

De 750GB has 250GB platters, and the 500GB has 166GB platters. The more space on the platters, the faster it is.

I've just ordered one of these drives. Which software is good to scan it for errors? I want to scan it before putting any data on it. Thanks

To be honest, I don't trust the current version of HUTIL. And to be more specific, the Debian-based tests we've been through are more exhaustive than HUTIL will ever be. We've R/W every single byte of these drives with a command-line UNIX system that tells us absolutely everything that happens. I don't think any DOS-based utility, written by Samsung or not, can do better than that. Besides, I don't want HUTIL to accidentally erase my data (yes, it can happen according to Samsung - what a reliable utility ! ).

I wouldn't worry about testing at least a single drive (at a time) since you're on a raid system.. killing one can't hurt right? Besides in testing mode it won't write to the drive, only read (just make sure you don't select format/erase disk). And the results of your tests could increase many people's faith in these disks. At 750/1024 Gb, would you really want to have a disk that can prematurely fail? Talk about a major pain in the ass if it does - salt on the wound if it gave you warning. Especially for those of us not running with a raid.

Also, I believe that since they are the manufacturer they will know of things only a manufacturer can about their disk and their utility might exploit this in its diagnostics of the drive. This can very easily be outside of the standard disk testing procedures/smart.

My personal believe so far is that it's some weird incompatibility but I cannot see why such such a strange incompatibility exists and this has me a bit worried. The fact remains that this is their drive diagnostic utility, alot of people have the error, and not all of them are on nvidia chipsets, tech support from samsung doesn't seem to be in a rush to clarify nor does samsung seem to care to update hutil.

If you would be so kind though, could you run the utility on at least one drive (with full surface read scan - should take about 3-4 hrs) and report your chipset to help ease at least my mind a bit more.

I guess I've been misunderstood. I was kidding about the "losing data" part. Of course I'm in a RAID 5 array, so I'm not afraid of that.

This being said, I'm sorry but I won't waste my time in using that stupid - and probably buggy at this time - Samsung HUTIL thing. I don't know if you really are aware of what a RAID-5 UNIX system can do, but unlike DOS/Windows crap, with a UNIX system in command-line mode you can absolutely master and monitor every single tiny bit of information you desire, with 100.00% confidence. So when I say that we've already tested the drives, I mean we really *have* tested them, more seriously that any other utility will.


I really hope people's faith doesn't depend on that.


I was secretly waiting for this one. I understand what you mean, but honestly, I don't buy it. The myth about a HD manufacturer knowing secret ways into testing his own devices is... well, just that : a myth. For complex devices (mother boards, etc.), it could be true. But not for the particular case of hard drive testing.

A HD is a very simple R/W device. You can test it in R mode, W mode, or both modes, with or without SMART. We have already been through all that, and I can assure you that every byte in the disks has been tested. Had there been the tiniest little error, the RAID 5 array would already have reported it. Are you trying to compare a perfectly understandable UNIX command-line script with an obscure DOS utility on a 1,44" disk ? Come on... Please, let's be rational and stop thinking that a simple Samsung utility will report magic errors that wouldn't be found by a professional UNIX RAID 5 system, just because the utility is written by Samsung. Do you imagine the consequences if it were true ? All major companies would stop using UNIX RAID systems and would check their servers with HUTIL twice a week !

If HUTIL exists, it's because most people run Windows and not all people want to install a UNIX system to test their drives. For UNIX users, HUTIL is simply useless. That's all.


I have already reported that.

Look, in short raid5 and unixish systems do a very good job of notifying of everything that goes wrong - that is, of what it can notice, but a hard drive is a black box and generalizing doesn't work well for those. Further more, there's more to a drive than the platters, there's the full blown embedded system on the bottom which is a computer in itself. I do want to state that I have complete faith that your system is running fine overall and every byte on them is correct. Encrypted right? I mean nothing would work if a single bit is bad, correct? But that just means the black box is performing input and output as expected, it does not completely reflect on the internal state. That is what the manufacturer's diagnostic is for. And their diagnostic may have some weird incompatibility/be buggy... but overall it doesn't seem like it with the possible exception of the m.c. error.

As for why enterprise doesn't use hutil... its too costly in many ways to screw around with a potentially bad disk and makes no sense in a high availability environments, so use it unless the management system says its bad in which case the black box is not performing correctly and to get another drive or who cares it works. There's also the downtime of using hutil which would be waaay outside what a business would go for. But most of us are in a desktop world, at least wrt these drives and where they will be used.

In the end though, I'm highly curious whether you get the error or not as you have a number drives and probably different controller to plug them into. I'm basically asking you to try out the test sometime , at least on a single disk with no full read scan, but you could do that too to help prove that raid/kernel sees right inside the blackbox if there are absolutely no ecc errors/etc.

Btw I used the cd version of hutil

Sorry I ment T166. Does it have also those 166GB platters?

Yep. That's why it's called T166.

I've got to agree with nevion. Jose, your analysis and experience with your drives helps the rest immensely, so thank you for your posts. However, I disagree with your assessment, Jose, that your Unix command line tools can monitor all aspects of these drives. I'd use the analogy that your Unix tools can tell you that the plane's engines are running and that you're flying straight, but they won't give you access to the internal circuitry of the airplane (or in this case, the hard drive) which could predict problems in the future.

I do agree with you, Jose, that Hutil is a poorly designed and buggy piece of software. I think what the rest of us are curious about is simply what sort of results in Hutil you're seeing on your large batch of drives. If, for instance, you have managed to get a batch that Hutil gives the "all clear" to, that would tell us one thing and if, for instance, your drives "fail" every Hutil test, but are still managing to work for you in your RAID array, that would tell us another. With this information, the rest of us could extrapolate what to do with our drives where we have limited information to go on. That is, should we trust our own drives or not?

To add to the pool of knowledge on this, I received another HD103UJ last night from Newegg (shipped 1/24/08). It failed the Media Check test in Hutil, but passed all others. I currently have it in its 12th hour of running another hard disk test without so much as a single error. I'll keep you guys posted on how it goes. If it passes, I'm going to put it in a RAID 1 array with another HD103UJ that had similar results in Hutil.

UPDATE from Samsung

They have acknowledged that there is a problem, and are investigating this as we speak.

I have been promised a call between 2-5 PM tomorrow with an update.


Andy

Well judging from Joses report, the problem seems to mainly target european drives, as the newegg ones are fine. Oh well, I will be getting a GP now, since the price drop, a real pity as I'd like samsung's performance.

Looking forward to hearing from samsung, andyb.

A problem where ? With the drives or with HDUTIL ?

As requested by lm, here's the way we have used to test our drives with Debian (after the RAID sync itself had succeeded with no errors) :

1°) Generate a random file 1 GB size named "1go" in /home




2°) Small script : copy that file XXX times (replace XXX with what you want according to your real storage capacity in GB) and write all the md5 results in a file called "md5sum"



Once all is done, you only have to view the "md5sum" file and filter all the md5 that are equal to the first one. If there are no md5 left, then everything's perfect. If there's even a single md5 left, then you have R/W errors.

That's all !

(pretty powerful isn't it ? try to do the same thing with windows )

Debian (GNU/Linux) isn't Unix, and a read/write test does not check all areas of the disk. The master cylinder, for instance, is not checked by that test.

Damn, where is the 640GB version? I'm running out of space on my 321KJ... Any news on when those are going to be available?

UPDATE from samsung.


Tomorrow I am going to install my PC with my new 1TB HDD (and only use 64GB of it) on the basis that there is nothing wrong with the drive at all.

Its a real shame that Samsung's tech support is shyte, yes they call you back, but none of the HDD techs have a direct number - so you can never get in contact with the same person.

When you call them (and sometimes when they call you) you get to a generic call centre for Samsung HDD's (and other stuff), you get some random twat that cant find their own arse with both hands and a map.......... They then treat the call as a new call even when you give them your reference No. something is really phuqed up there.

Fortunatley the only tech guy I have had contact with gives me confidence, I am not sure why... he fails to return my calls.

All of this is not inspiring my confidence in Samsung, but I will persist as I feel that their HDD's are not as likely to fail as WD's or sHitachi's but in this case their Diagnostic tool is at fault.

My total 100% confidence will be restored when they phone me to tell me that they admit their fault, release a new test tool and send me a new 1TB HDD as a gift for saving them hundreds of returns

My most recent contact with my (usual) Tech from Samsung, it that he has escalated the problem up many, many levels and he has also forwarded many hyperlinks from several forums pointing out these mysterious errors to the people who make the test tool, and is awaiting a response. Unfortunatley he has not contacted me back, and their utter shyte and vaugue team of arseholes cant forward the message's to him that I want him to call me with ANY info he has.

Today I got a very vaugue call asking to speak to DAVE (name made up blah blah), they thought that I was Dave and that I had a HDD problem.......... WTF I thought. What more needs to be said about their "general" team, anyway I pointed out that DAVE (name made up blah blah) "from Samsung's HDD Team" was supposed to be calling me and not te other way round.

As you can expect I am dissatisfied...... did I mention VERY...... please contact me with anyone's e-mail address who works for Samsung (who is relevant to HDD's). Ideally post it straight to this forum as I am sure many people will e-mail that address and ask questions etc etc, you never know Samsung might just take one of those thumbs out of its arsehole and its call centre might work much better not being plugged up

Andy

PS: The next time I speak to DAVE (name made up blah blah) I will send him to this thread