Page 1 of 1

THDD - Correction - (Test HDD)

Posted: Tue Nov 21, 2006 10:24 pm
by Jim
Prior to its current rebuild Superpuppy 3 had a 20 Gig (7200) Maxtor Diamond Max boot drive, and 2 x WD 40 Gig (7200) storage drives. During the current rebuild the WDs got replaced w/ 2 x Maxtor 80 Gig (7200) Diamond Max storage drives. There is a reason for this.

I have always generally prefered WD drives because I have heard they were faster than others. In my collection of HDDs I have a number of Maxtor slimlines, (single platter drives), (1 x 10 Gig 7200, 1 x 20 Gig 7200, 2x 30 Gig 7200 & 1 x 40 Gig 7200), which I bought because I thought they would run cooler because of enhanced airflow in the cramped confines of ATs; but I have heard since I bought them, that they are not very reliable. I also have 5 x WD 40 Gig (7200)s and 1 x WD 80 Gig (7200). Seagates, on the other hand, I have not had many of. Superpuppy 1 has 3 x 8 Gig Seagates in it; but the only other Seagate I have ever briefly owned was a 60 Gig (7200) Barracuda that I bought used.

That brings me to the point of this post. That Seagate would pass MS Scandisks surface check w/ no problems; but when I checked it w/ THDD another picture arose. THDD is a disk checking utility that I have which can be downloaded from somewhere if you check the net looking for it.

What it can do follows :

1) Perform a surface scan on an 80 Gig drive in about 15 minutes.
2) Allows you to start the scan from any point on the drive.
3) Monitor the drives readspeed as it is checking the drive, giving a continous readout in kb per second of the part of the drive it is currently checking. (The 60 Gig Seagate had wildly fluctuating readspeed)
4) Remap the drive to exclude up to 200 bad sectors so they won't even appear running scandisk.
5) Allows you to manually specify any given sector as being bad. i.e. if there is a bad sector, you can, if you choose, specify that the two sectors on either side of the bad one be excluded from the drive map, thereby avoiding having your data stored in questionable sectors.
6) Show you a graph of the read speed of the entire drive, wherein a good drive will show up as a flat green line on a blue background; and damaged sectors will show as spikes upwards from the flat line, denoting increased readtime for that part of the drive. This graph also has a limit to how high the spikes can go before they run off the blue background. Off the blue background equals bad sector as opposed to just damaged.

EDIT : Forgot to mention THDD also lets you mess with the FAT tables and the MBR, neither of which I have had the nerve to do which is why I forgot that part. One of these days I'll try it on an old 100MB drive to see what you can do.

The 60 Gig Seagate had spikes almost to the top of the graph throughout the whole front third of the drive. I took it back to the store where I bought it along w/ my THDD disk and got them to run the test there in the store. They took the drive back and I got my money back.

Now this brings me to the point , "Why Maxtor Diamond Max"? Superpuppy 3, like I said, has a 20 Gig (7200) Diamond Max boot drive. When I was working out my WPCredit tweaks, (See WPCredit as applied to a DFI K6BV3+/66), Superpuppy 3 was my "Guinea Pig". I must have hard crashed the machine a couple of hundred times, and yet THDD says my 20 Gig Diamond Max is still in "Perfect Condition" The graph is still perfectly flat!! They may not be the fastest drives out there; but they are darn near indestructable. (Which is more than I can say for my WDs.)

Thought I'd pass all that along.

Posted: Wed Nov 22, 2006 5:46 am
by DonPedro
I don't think that "hard crashing" (freezing, blue screen, rebooting)a system has any effect on the electro-mechanical parts durability of any given harddisk. I just can't see why. a system that crashes is something that happens in ram because of software-failures, it can't break the hd's platters or heads. if a system is unable to boot from a disk after a system crash it is because of destroyed data on the disk, destroyed mbr or fat. the only situation I can imagine that might do mechanical harm to any disk is when you pull the power plug or the ide-cable while the hd is in action (absent throwing your hd out of the window).

my preferred hd-manufacturer is samsung. their models are very fast while at the same time they operate at lowest temperatures and are almost inaudible. I 've been using samsung disks in all my machines since many years and had no problems whatsoever.

Posted: Wed Nov 22, 2006 7:37 am
by Jim
Peter that is something I would never do, (pull the IDE cable on a running drive), but the fact is head crashes can and do occur. I know I got one on my castlewood orb when I had to reset to get out of a hard freeze.

Posted: Wed Nov 22, 2006 9:38 am
by DasMan2
I have to agree with Jim on the Maxtor drives as I own about 5 of them in the 10/15/20 gig sizes and they run and continue to run very well for the last 4 years. WD similar sized drives are a step down in performance/noise and reliablity.
I have two or three Seagate Barricudas IV liquid bearing that are very quiet and run fine ,but lack some performance strength.
Seagate do have their 5 year warrenties though :)

Hopefully soon solid-state drives will be coming out at lower cost for the useful sized ones that I hope are IDE compatible in Super 7 systems ..then extra performance can be had and enjoyed 8)

Posted: Wed Nov 22, 2006 8:03 pm
by Jim
@ Peter: I didn't have time to write this this morning; but at one time an HDDs read heads were kept floating just above the surface of the platters by an air cushion generated by the rotational speed of the drive. If for any reason that air cushion collapsed while the drive was spinning, and the heads were not in the parked position, then you got a "head crash". (i.e. physical contact between the read heads and the spinning platters) Abrupt power failures were the most common cause of "head crashes". What improvements have been built into newer drives to prevent "head crashes" I don't know. I do know however that my WD Drives are more susceptible to damage than my one and only, (till very recently), Maxtor Diamond Max. The brand new WD 80 Gig drive that I put in SP5 as an internal storage drive was tested w/ THDD yesterday, and it is showing minor damage scattered over all the drive. (Short spikes on the read time graph). So is the old 40 gig WD that went into SP5 when I got a new WD 40 for SP3. (In this case a couple of isolated medium high spikes combined w/ a scattering of short spikes.) If you like Peter I will send you a copy of THDD; and you can check your Samsung.

EDIT : Actually if anyone wants THDD you can download the "com" file here : http://202.116.43.3/life/download/format/hdd/thdd.rar
but you have to extract the downloaded file and copy it onto a sys-ed floppy, as THDD runs under DOS not Windows.

Or to state it more clearly :
1) Download the THDD file and extract it.
2) Take a blank floppy and insert it in your floppy drive, then single click your "A" drive.
3) Click the File Menu, and click format, then in the format popup select "Full" & "Copy System Files" then click "Start" The format will "Sys" the disk making it bootable.
4) Copy the extracted THDD file (thdd.com) onto that floppy.
5) Set your boot sequence to A/C/SCSI.
6) Insert the floppy, reboot, & at the "A" prompt, type : thdd - then press enter.

That will run the THDD program, which is the best HDD checker I have encountered.
It will display what drives are in your machine and ask you to choose one. Once you have chosen a drive it will display a menu w/ the following options.
Clear FAT
Clear MBR
Surface Test
View Defect List
Choice Drive
Exit.
Use the arrow keys to highlight "Surface Test" and press enter. That will bring up a popup screen that asks "CHS" or "LBA". Use the arrow keys to highlight the appropriate selection, (LBA - usually), and press enter. That will bring up another popup screen that asks you where on the drive you want to start the test. Choices are 0%, 25%, 50%, 75%, and custom, which allows you to set the start and finish locations of the scan in MB. Choose 0% and start the scan.

Note: You will see the read speed of the drive you are checking displayed, and changing, drifting up and down a bit, as the surface scan runs. That is normal. You will also see the read speed slowly getting lower as you get toward the back end of the drive. That also is normal. If your drive is any good it should not show any errors in the screen shown, (Similar to the old DOS Scandisk screen). But, when the check is finished, (even if there are no errors shown), if you click on "View Error List" it will bring up a screen showing any and all defective sectors by number, (if there are any defective sectors), and a screen offering the option of Viewing the Read time Graph.

If you type "M", the screen will go black and after a short pause, you will get to see the graph of read times created during the scan for all the sectors on the drive. Slow read times, (damaged, but still readable sectors), show up as spikes up from the normal time for the drive. The more spikes the more damaged sectors. The higher the spikes the worse the damage is. i.e. the closer the sectors are to being unreadable.

EDIT : NOTE : THDD 's drive size counter, (in Megabytes), only goes up to about 65 Gig. After that the counter resets and starts over if you are checking a larger drive. (At least that is what happened when I checked my 80 Gig WD in SP5 with THDD). The scan will still run to the end of the drive however; and the read speed graph shows the entire drive. I would be cautious however about remapping a drive to exclude bad sectors where the drive is larger than 65 Gig.

EDIT : Just rechecked the link to THDD and found it is no longer operational. SOOooo, if it is permissible, what I am going to do is post it here as an attachment, since it is a rather small file anyway. -- Well I tried; but the file extension "com" is not allowed. Maybe Nohr will see this and allow it, - maybe not, - don't know the legality issues.

Posted: Sun Nov 26, 2006 10:45 pm
by Jim
@ Peter : I would appreciate it if you would run THDD on one or more of your Samsungs. The only Samsung drives that I have are very old ones, (less than 1 Gig). Would like to know how well they hold up in terms of the read speed graph. Will be needing new drives for SPZ when I get around to that project.

Posted: Mon Nov 27, 2006 10:37 am
by DonPedro
I just let thdd scan a 80gb samsung (7200, 2mb cache), connected to the soyo5ssm board (sis chipse). as I have explained in the sis-thread the bios does not report the full size, it says the hd is 32gb in size.

I ran the test anyway. speed range 63000 in the beginning working towards 53000 (end of 32gb). the graph shows 7 spikes that are about a tenth of the height of the blue box. between them are 1 to 3 smaller spikes, about half of their height.

Edit: I now also have the data for an "old" 20gb maxtor drive (5400rpm, ?? cache), attached to the soyo board. speed starts at 25000, ends with 13500. the graph is different: the bottom green "line" is quite "thick" now, 3 to 4 times as thick cmpared with the samsung (which was just 1 or 2 pixels wide, I did not take notice then). it has 3 spikes that are about double the size of that of the samsung and it has also a lot more smaller spikes.

Posted: Mon Nov 27, 2006 5:30 pm
by Jim
Thanks very much Peter. Your Samsung is in pretty good shape, though not perfect. Those spikes are all damaged sectors, irrespective of what causes them. "Q" : Now that you have tried it, what do you think of THDD?

Posted: Mon Nov 27, 2006 7:42 pm
by DonPedro
re thdd I wwould like to know what the thickness of the green line represents, in other words do you have any documentation on this program, written by the author?

I also wonder if there could be an influence by the pci-bus on the spikes in the graph. the samsung spikes showed some kind of pattern that I would not expect to represent problems of the surface.

Posted: Mon Nov 27, 2006 8:59 pm
by Jim
Re documentation, no none whatsoever. Re pattern, that is something I have often seen on older drives, even w/ scandisk when it is reporting bad sectors. My best guess for the reason, is when there is a "Head Crash", the read heads hit the surface of the drive then bounce, back up off the surface before landing again further on, sort of like skipping a stone on water. They go through several bounces before things settle down.

Best guess as to the line thickness factor is the higher the read speed of the drive is the thinner the base line. That would help to keep the graph in scale.

EDIT : If you want to check those thoughts, try testing an old slow drive that is known to have bad sectors. If I am right, the base line should be even thicker and the spikes should go right to the top of the graph in some places, and you will probably see a pattern in the damage.

Posted: Sat Jul 28, 2007 2:00 pm
by Jim
Flash !! Thanks to Nohr, a newer version of THDD is now available in the downloads section. I have not tried it yet, so I don't know what all it can do; but if it is an improvement on the old version, then it is pretty good.

EDIT : K, I downloaded and tested the new version. Some things I noticed have changed.

One is that you can now run the test in graphic mode seeing the read time graph displayed as it is generated. Don't recommend doing that however, because the price is that you don't get to see the read speed readout of all sections of the drive as they are being checked; and you can still see the graphic display of the readspeed for the whole drive after the test is finished.

Another is that the drive size counter in megabytes displayed as the drive is being checked is no longer limited to 65 Gig. (At which point the old version would continue checking the drive with the counter resetting to zero and starting over). This means that it is now safe to remap drives larger than 65 Gig. I checked an 80 Gig drive, which is the largest I have and the counter just kept on going up till the test was finished.

Also new is a readout of a bunch of (over my head) internal HDD settings which it appears you can use THDD to modify.

Thanks very much Nohr for finding the site with it and adding it to the downloads.