WD 2TB drives not advanced format

So, i got the WD2001FASS and after plenty of investigation, found out that even if they do support AF, they are not by default advanced format, and have decided to put this information online for anyone who is trying to install one on LINUX.

The problem is that ONLINE, there are many answers, some saying it is, some saying it is not, and some saying it is compatible (Whatever that means),

The short answer is, NO, Western Digital 2TB BLACK WD2001FASS IS NOT ADVANCED FORMAT, i am sure about that but don’t hold me liable for anything (In case some crazy American gets too excited).

Proof

1- the results of hdparm are.

hdparm -i /dev/sdd

Model=WDC WD2001FASS-00W2B0, FwRev=05.01D05, SerialNo=WD-WMAY00589096
Config={ HardSect NotMFM HdSw>15uSec SpinMotCtl Fixed DTR>5Mbs FmtGapReq }
RawCHS=16383/16/63, TrkSize=0, SectSize=0, ECCbytes=50
BuffType=unknown, BuffSize=unknown, MaxMultSect=16, MultSect=off
CurCHS=16383/16/63, CurSects=16514064, LBA=yes, LBAsects=3907029168
IORDY=on/off, tPIO={min:120,w/IORDY:120}, tDMA={min:120,rec:120}
PIO modes:  pio0 pio3 pio4
DMA modes:  mdma0 mdma1 mdma2
UDMA modes: udma0 udma1 udma2 udma3 udma4 udma5 *udma6
AdvancedPM=yes: unknown setting WriteCache=enabled
Drive conforms to: Unspecified:  ATA/ATAPI-1,2,3,4,5,6,7

* signifies the current active mode

————————————-

and

hdparm -t /dev/sdd

gives us

Timing buffered disk reads: 418 MB in  3.01 seconds = 138.72 MB/sec

Which is the number on the Western Digital Website

———————————–

In any case, i have formatted it once assuming it is AF and once assuming it is 512b (Not AF) and the results were the same (And that is why i say it might be compatible but i don’t know how that can be done anyway)

Disk read and write measuring and monitoring

The tools i need are in
apt-get install sysstat

this will give me the command

iostat -x 5

Where the 5 means tell mne about disk usage for all partitions and disks every 5 seconds…

When you stop the command, running it again will display data since the last run, there is much more that has to do with IOSTAT that i will come to in a bit, this one here is simply the basic command and usage for my referance

IO stat will display something like the following lines for the second disk that has 4 partitions, the first line is the totals

sdb 0.20 0.20 33.20 16.00 728.00 1392.00 43.09 3.87 41.01 3.04 14.96
sdb1 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
sdb2 0.20 0.00 7.00 0.00 502.40 0.00 71.77 0.06 8.46 7.54 5.28
sdb3 0.00 0.20 25.60 15.40 204.80 1384.00 38.75 3.79 47.41 2.85 11.68

Just got my ATOM D525MW motherboard

I got this motherboard because it is fanless, and all reviews say that this Dual core (4 thread) ATOM processor is faster than a Pentium 4 for something like a database application and a web server, and that is exactly what this PC will be doing. Reviews also say that this PC also puts some decent performance for an average home PC used for browsing and similar little tasks (Did photoshop CS5 fine for me, but surely, no GPU optimization, did the job but probably too slow for a professional designer).

In any case, this fanless PC turned out as cheap as $137 ($79 For motherboard and processor, $18 for 4GB of RAM, $20 for a case and a small power supply, $20 for a small SATA hard drive). very cheap price wise, and much cheaper on electric with around 20 watts of power compared to the 200 watts the average PC draws ( 1/10 of the power consumption).

And for all i know, the power supply can run fanless when this little power is drawn from it (i did punch a few holes on the top for ventilation and passive cooling), so this is a completely fanless PC that costs $137 (without the monitor).

As soon as the new PC was switched on, i could F7 to update the BIOS, so i did, before we even start,
The motherboard had the bios (MWPNT10N.86A.0083.2011.0524.1600), while the latest bios on the INTEL website dated 10/13/2011 0098 MWPNT10N.86A.0098.EB.EXE so i installed the latest.. everything is working fine up to now.

First of all, i installed a hard drive that already had windows 7 64Bit on the new Motherboard, the drive was on an intel G41 series motherboard (Socket 775), it work right away, windows seems responsive, although it needs activation now.

NOTE: i think, that since this PC uses the Unified Extensible Firmware Interface (UEFI) (The new system replacing Basic Input Output System BIOS, since BIOS is being phased out this year), The Windows 7 Activator by Hazar and Orbit 30 would probably not work, the similar software that would work is probably DAZ loader for Windows 7, and from the feature list i have read, probably Version 2.1 and above.

So, The motherboard booted from a hard drive that was on a G41 motherboard, and it also booted Linux directly from a VIA chipset motherboard with a celeron processor, on that drive i got from the Celeron PC with via chipset, all i needed to do was edit the file in /etc/network/interfaces to use eth3 rather than eth0 (Since the name changes with the new network adapter built in the motherboard) and it seems everything is good to go !

LOAD seems to have gone down,

I installed one 4GB 204 PIN SODIMM DDR3 (The motherboard’s max is 4GB, you can put all 4 in one piece or 2*2).

Anyway, now that this thing replaced the Celeron, i can expect increased performance (Since the Celeron D 331 is based on Pentium four) and some saving in the electric bill (Seriously now, who cares, but it sounds responsible)

For those of you wondering how much those savings could be… here are some estimates since i can never tell for fact (Unless i hook it to my ammeter)

I think the headless / Monitor-less celeron consumed about $10 per month (And this is a very educated guess).
This one will probably consume the same for everything but the processor, which is 13Wat max TDP from a 85W for the celeron
Making the PC consume something around $2 or $3, and therefore i should expect $7 in savings every month, That is, i will get my money back within a year’s time.

The memory puzzle (Solved)

I have edited this article many times, and because you are probably interested in the problem (To see if it is like yours), and the solution, i have added a few paragraphs with the information you are looking for at the top, and if you are interested in more details, you can read the old texts, speculations, and the like.

The problem is that, when memory modules are installed alone, they always pass the memory test, but when together, they sometimes pass, and with the same installed memory, other times fail the test with no apparent reason.

As it turns out, even though the ram has the same model number, CPU-Z unveiled that they had different production dates, and for some very strange reason, every 3 of the 6 DIMMs had similar timings (and production dates)

This article describes my problems with the same brand rams with even the same model number, and a surprising difference in SPD data between ram modules, but in reality, this applies more to different models of ram even when they are not on the same channel (In dual channel or three-channel modes). Even though i recall on older models of intel motherboards they say it does not matter as long as identical sticks are installed for a certain DIMM in every channel, it seems the 5XX and 5X series chipsets want the timings to be the same on all modules.

So, the solution was

1- Boot with the first DIMM (RAM module), start CPU-Z, see the timings in the SPD tab. take a screen shot, and label it 1, also label the ram stick 1, do the same for all other RAM modules and label them 2, 3, 4, etc…

2- Compare the timings of the RAM modules, if they are all the same (on the CPU-Z SPD page), your problem is a different problem.

WARNING: We are about to write data to the SPD area of the RAM stick, DO THIS AT YOUR OWN RISK, AND YES, THERE IS A SIGNIFICANT RISK

3- FIX timings in the RAM SPD: If the problem is that they don’t share the same timings (Like my problem), download SPDTool.exe, and read the SPD data from one of the DIMMs into a file, if you have for example 4 sticks with a certain timing and 2 with the other, i would recommend the using the timings from the 4 on the 2 and not the other way around.

4- Boot with other RAM modules and dump the file onto them, make sure the checksum is correct, if the software warns you that the checksum is incorrect, use the software (SPDTool) to FIX THE CHECKSUM.

NOTE: For some unknown reason i had to fix the checksums (There is such an option in the menu of the software), and for some unknown reason, i had to dump the fixed checksum file back onto the source DIMM as well, but it worked.

Also, from the 6 sticks, 3 had timings exactly like the ones from my Kingston models, so i chose that over the other timing set, as for you, when it comes to memory timings, the lower is better (5-5-5 is better than 8-8-8).

Now, the above is what you need to know, if you need more, i didn’t delete anything from the old text, check it out below.

Initial hardware problem description

Less than a year ago (Aug, 26 2010) , I went and got an I7 processor (i7-930), A (X58A-UD3R) 3 channel memory motherboard (6 sticks), and 6 sticks of 4 GB memory (TwinMOS PC3 10600 4GB DDR3 1333 256*8 16IC CL 9.0 U-DIMM 9DCEBNZB-TATP) to achieve that 24GB of system ram (Yes, i need this much to run my own software, the more ram, the faster it runs).

In any case, things did not go well, the motherboard played fine once every time i shuffled* the 6 sticks around, and got memory errors on any second reboot after the shuffle, Sometimes i change something in the BIOS and things work fine for another single boot, and sometimes, i just let the pc with no electricity (Strange since the CMOS battery is new), and things would work just fine.

NOTE *: Shuffled or sometimes introduced a 2GB stick from another computer to replace one of the 4GB DIMMs.

A pity since the ram looked very nice, with aluminum heat spreaders on both sides, (Although you can see that the spreaders do not make contact with ALL the chips under them, but that is not relevant to the problem it appears), and painted in 2 colors, 3 black sticks read Piano Black, and the other blue 3 read Ocean Blue.

So, i am thinking to myself, there is something special about the first boot that makes everything work fine most times after a shuffle, and something is saved for the second boot that gets things messed up, cached SPD ? i don’t know yet, we will work around this together.

In any case, The TwinMOS DIMMs i have cost me $141 * 6= $846, worked fine if you run any 2 DIMMs together (LATER: Appears the 2 DIMMs needed to be the same color and SPD but i didn’t know that) , threw errors on the Gigabyte mobo if all 6 were present.

Got the Kingston for the Gigabyte – All ok

Some time later (March 2011), i got annoyed from the shuffle before you boot game, and decided to get it over and done with, so I downloaded the compatible memory sheet, and sure enough my TwinMos ram was not on that list (LATER: Turned out to be irrelevant), I decided not to waste time on it, and went and got 6 Kingston DIMMs ( KVR1333D3N9/4G 4GB PC3 – 10600 CL9 240-Pin DIMM 9905471-006.A00LF ) at ($59 * 6 = $354) that worked like a dream on the Gigabyte motherboard (No errors at all). even though this exact kingston model was not on the compatible memory sheet.

New Motherboards Intel DH55HC, MSI G41M-P26, and Asrock G41C-S

So now i had 6 TwinMOS DIMMs, and 2 Kingston 2 GB DIMMs (KVR1333D3N9/2G 2GB PC3 10600 CL9 240 Pin DIMM 9905458-009.A00F) not in use, and for a reason i don’t remember (Even though this a few days ago), rather than selling the DIMMs on Ebay or something (Since they all work just fine if used in pairs of 2), i went and got 3 motherboards.

Two of the 3 motherboards had the G41 chipset that can take 2 DIMMs each, and one INTEL DH55HC motherboard that can take 4 DIMMs, thinking that this motherboard (The H55) would run my Ram just fine.

So i updated the BIOS to the latest on all nice new motherboards.

Both G41 motherboards would take any 2 sticks of ram with no problems at all, But the H55 motherboard is doing the same exact thing our Gigabyte motherboard was playing.

On the intel website, it reads that my DH55HC motherboard can take

  • 16 GB maximum total system memory (with 2 Gb memory technology)

What is 2Gb (Notice, Bit not byte) technology, and also note that it is taking my 16GBs of twinmos RAM, but not happily.

The timings of the TwinMOS RAM (That our Motherboard allows us to change) are

Multiplier = 10 / 9-9-9-24 / 1.5Volt (Provided by the SPD from the RAM)

But wait a minute, the timings on the black sticks is not the same as the timing on the blue ones !

As it turned out, SPD is stored on Electrically Erasable Programmable Read-Only Memory (EEPROM), and i can rewrite that with some software.

Here, Go to top to see more :)

Twinmos 4GB 10600 not compatible with GA-x58A-UD3R

So, when i got the TwinMOS rams, they were $141 each, making the total $846 for 24GBs of system ram, Sometimes the computer would work with no “Apparent” errors, other times, i would have to switch rams and reboot a bunch of times till i get a stable run !

The PC is a linux machine that runs MySQL, the result, i was always worried that i would have to update the database files because when i do, i usually strike that error and loose a day of work.

So the problem shows it’s ugly head again, but this time i am very fed up, i ran memory tests and the tests report errors, i move the TwinMOS ram’s to another computer, and they work like a dream.

So i went out and got some Kingstone rams (KVR1333D3N9/4G) and up to now no errors

The files i used to copy and get back MD5 sums for, now i copy and they arrive at destination with no errors in the copied files, everything seems to be fine up to now

So, the Ocean Blue and Piano Black sticks (With heatsinks) are no longer needed, i should now see if someone would take them off my back (Since i tested and they are working fine with all other motherboads)

how to mount a ram hard disk in Linux

In this post, i will show you how to create a very fast, 0 latency hard drive from the extra gigabyte or 2 on your system

1- You need to change the boot options to allow this…

Since i use debian lenny, my boot options look like this in /boot/grub/menu.lst

title        Debian GNU/Linux, kernel 2.6.26-2-686
root        (hd0,1)
kernel        /boot/vmlinuz-2.6.26-2-686 root=/dev/sda2 ro quiet
initrd        /boot/initrd.img-2.6.26-2-686

We need to append ramdisk_size=1572864 if we wanted a 1.5GB RAM drive, make sure you have 1.5GB EXTRA on your system, let’s say you must have a minimum of 2GB for the system to run using the 512MB you are leaving the system with

title        Debian 1.5GB Ramdisk, kernel 2.6.26-2-686
root        (hd0,1)
kernel        /boot/vmlinuz-2.6.26-2-686 root=/dev/sda2 ro quiet ramdisk_size=1572864
initrd        /boot/initrd.img-2.6.26-2-686

You are done for now, if you want to format and use it do this

/sbin/mkfs.ext2 /dev/ram0

Remember that you need to format as ext2 and not ext3, i have seen people format the hard drive as EXT3, why would you ever need Journaling on a RAM disk that is already volatile (Deleted when you reboot)

mkdir /ramdisk
mount /dev/ram0 /ramdisk

You can now get creative and format or load a disk image at boot time, in any case, the RAM is only allocated to the disk when you use it, but you really don’t need to know that to use your new RAM hard disk

NOTE: I chose to explain RAM disk because later on i will show you what advantages we can get from having a block level device, you can surely do this without modifying boot options by simply using tempfs or RAMFS that give you an instant ram disk, but it is not a block level device in that case

Testing Speed

So, you want to see how fast it is, but for that we need a data source that can push this thing to the maximum, the answer is /dev/zero that responds with a stream of zeros, so let us write a 1.3GB file to our new volatile hard disk

dd if=/dev/zero of=/ramdisk/pathtoimage.img bs=1M count=1300

On my computer, this took about 4.54006 seconds.

USB flash not for database

SSD disks such as the 128GB supertalent SATA drive are good for DB applications no doubt, but USB flash disks are not, i experimented with MySQL with an operation that probably has seek issues (Latency)

So i have a database under processing where the problem is supposed to be “Latency and seek time”, so i went and got the SanDisk 16GB extreme that is supposed to be 20MB/s read and 18MB/s write, and surely, like all flash drives, no latency.

The performance was very very slow, it is probably a problem with USB or the USB controllers that work that flash memory.

Anyway, if you want to work a database from a USB flash disk, save those $85 (What i paid for the flash disk), you are better off with a SATA hard disk or even an ATA 133. but nothing USB.

PNY GeForce 8800 GT

I took it apart to get to the bottom line of this

Some people tell me it has the same processor as the ultra edition but that one has higher clock speeds, Some tell me this one has some shaders dissabled (Maybe they are missing not disabled), The nvidia website tells us about the differences, But i want to get to the bottom of this, So i posted the photos of this one here

Click here for photos of the 8800 GT

asus tusi-m bios update

efore finding out that the motherboard only supports usb 1.1, i was getting the TUSI-M motherboard ready for prime time, i made it fanless (Fanless Air cooling), i put it in a nice little box to fit some furniture that was made by idiots for idiots, and installed Ubuntu on it.

But before installing ubuntu, i didnt want to install a floppy drive, they bring back bad memories, so i had to burn a CD and boot from it to update the motherboard

If you have this old motherboard and want a firmware update to the latest 2003 version, download this tusi-m package, Inside you will find the file DOS-BOOT-WINDOWS-ME.iso, burn it to cd and boot from it.

Wait until you are dropped to a DOS prompt
D:
Aflash
When presented with choices chose 2 and hit enter
When asked for the location of the file enter the following
D:1018tm.005
You are done, you now need to exit by hitting escape then restarting the computer

With this out of the way, i installed Ubuntu, but then found out about the missfortunate 1.1 USB posts, and therefore, it was disqualified, at least i had pleanty of fun making the case and the external heat sinks…..