This past weekend I bought a second 80GB X25-M from NewEgg.com because they were on sale for a good price. (note: I’m talking about Gen 1 X25-M’s in this article). I liked having an SSD running under Windows 7 for my software dev box so much that I decided it can only get better if I get another and stripe two of them in a RAID-0 set. I assumed this would help me in two ways: first this would give me approximately 150GB of disk space instead of 75GB; and second, I expected read and write performance to scale to almost double what I get in a single drive. Now read performance is excellent on a single drive, Intel claims up to 250MB/s. However, write performance, although good, could be better. Intel claims write performance of up to 70MB/s. So, I was expecting to throw these 2 SSDs in a RAID-0 set served up by the Intel ICH10R on my ASUS P5Q-E motherboard and get around 400MB/s read and 140MB/s writes. Here’s what I actually found:

SSD1-R is the read results from HDTune for reading on my first Intel X25-M. SSD2-R is the read results from HDTune for my second Intel X25-M (both are using AHCI and treated as single drives in the BIOS). Similarly, SSD1-W and SSD2-W are the HDTune results for writing.

summed-R and summed-W are simply the sum of the results I got from both drives in individual Read and Write tests. These are the numbers that ideally I would like to see in the striped set.

STRIPE-R shows the actual numbers I got for reading the striped set of SSDs. STRIPE-W shows the actual numbers I got for writing the striped set of SSDs. STRIPE-W+ is the same as STRIPE-W except that I turned on write-caching for the RAID volume via Intel Matrix Storage Manager.

As you can see, the striped set got me nowhere near my expectations, every metric for STRIPE-R and STRIPE-W was much less than ideal amounts calculated by summing the individual drives metrics.

One thing that really bugged me was the very low MIN write times for the individual SSD test. Take a look at these pictures and you can see ridiculous spikes in the performance graphs from HDTune.

SSD1-W (above)

SSD2-W (above)

Those are not what I call pretty graphs. The thing to remember with SSDs (in layman terms), is that once every page on the disk has been written to, in order to write to that page again, the SSD must first erase (clear out all the electrons) in every bit on the page prior to writing the actual data. This could explain what we’re seeing above.

Now what about reading? Let’s take a look at those graphs:

SSD1-R (above)

SSD2-R (above)

The read graphs above are not too bad but still not very smooth. I have no explanation for this result…

Well, the next thing I decided to do was try to figure out how to restore these SSD’s to factory performance by clearing all the pages on each SSD. After searching for some time, I found this page: http://www.pcper.com/article.php?aid=669&type=expert detailing how to zero out the SSD’s. This page: http://www.pcper.com/article.php?aid=669&type=expert&pid=6 also provides a download link to the correct version of HDDErase that will work with these Intel SSDs.

So I went to work on zeroing out these SSD’s… Afterwards, I re-ran the HDTune tests and here are the resultant graphs:

SSD1-R (above)

SSD1-R (above)

Now those are some nice smooth read graphs!

SSD1-W (above)

SSD2-W (above)

Still a little spikey on writes but not bad at all! I’ll take it!

STRIPE-R+(above) write caching turned on in Intel Matrix Storage Manager for this striped volume

nice and smooth, average 479.3MB/s reads, WOW!

STRIPE-W+(above) write caching turned on in Intel Matrix Storage Manager for this striped volume

not bad at all! average 148MB/s for writes, now we’re talking!

-----------------------------------

Now what does this all mean??? For peak write AND read performance, we’re going to have to image our hard drive, secure erase each SSD, and then put the image back on our stripe-set. This really sux! The good news is that newer SSDs including Gen 2 of Intel X25-Ms will support TRIM which will automatically zero out the pages on the SSD that are no longer in use at more opportune times than when you want to write to that page. Windows 7 has TRIM built in, so you should be all set after some number of firmware and windows updates have been completed down the road when TRIM will be working in all its glory… Until then, image, secure erase, un-image for peak performance. Oh and don’t forget, each time you write or zero out a page on your SSD, that is one less time you will be able to write to it before it fails. They don’t call this the bleeding edge for nuthin, folks!

1. Exit all Windows-based programs.
2. Click Start, type regedit in the Start Search box, and then press ENTER.
3. If you receive the User Account Control dialog box, click Continue.
4. Locate and then click the following registry subkey:
   HKEY_LOCAL_MACHINE\System\CurrentControlSet\Services\Msahci
5. In the right pane, right-click Start in the Name column, and then click Modify.
6. In the Value data box, type 0, and then click OK.
7. On the File menu, click Exit to close Registry Editor.

After doing this, reboot, get into the BIOS, change back to AHCI mode and eventually you will be able to login to Windows (i.e. no BSOD). At this time, Windows will figure out that is needs to install drivers for your fancy new AHCI controller that replaced your old legacy controller (which of course forces you to reboot one more time) and then you are all set.

At least this is how it went down for me with Windows 7 RC (BTW: Windows users can be happy and proud again with Windows 7, Mac ain't got nuthing on you no more...at least until Snow Leopard comes out, then we'll just have to see...)

Today I ran into a computer where windows update would download updates from Windows Update and would start the installation process before finally failing to install any of the updates. No reason was given for the failure. Looking at the update history, where one would expect to find the error, did not even have the update attempt listed.

First, where to find more information about the errors related to Windows Update: c:\windows\WindowsUpdate.txt. You might find some clues regarding the problems there. I Google'd on a couple of errors listed there and found the following solution for the problem I was experiencing.

Solution (re-register these dlls):

regsvr32.exe c:\windows\system32\wuweb.dll
regsvr32.exe c:\windows\system32\wups2.dll
regsvr32.exe c:\windows\system32\wups.dll
regsvr32.exe c:\windows\system32\wucltui.dll
regsvr32.exe c:\windows\system32\wuaueng1.dll
regsvr32.exe c:\windows\system32\wuaueng.dll
regsvr32.exe c:\windows\system32\wuapi.dll

Next, restart the Automatic Updates server (you might have to reboot to get Windows Update working in IE).

This problem was on an XP system for me but I have heard this also affects other Windows OS's.

• QX6700 2.66 GHz extreme overclocked to 2.93 via 11 Muliplier in BIOS
• ASUS P5Q-E BIOS 1406
• OCZ OCZ2VU80016GQ 16GB RAM
• Adaptec 3805 Firmware version 15753 (you'll need this version of the driver as well)
• WESTERN DIGITAL WD10EACS 1TB SATA 7200 RPM 16MB Hard Drive (8 of them)
• Windows 2008 server (also Windows 2003 server)

now make sure you have these versions of the firmware (especially the Adaptec firmware/driver). This caused me many nights of troubleshooting to get the Adaptec RAID controller to work. The symptom was the following messages printed on the screen after the controller completed loading:

• No Int 13 drives to support
• BIOS not installed!
• Not enough space to copy PCI option ROM

The final symptom/problem is that even though I was typing CTRL-A to get into the configuration manager for the controller. The configuration manager would never load.

At various times with various motherboards, I had one or more of the above error messages from the Adaptec controller. In each case, soon after these error messages, the computer would continuously beep.

After many night of trying many different BIOS settings, moving the card to different slots, buying another 3805 thinking that the first one was bad...the problem was the firmware for the 3805 that came with it from the factory would not work on my new motherboard P5Q-E (it wouldn't work on my previous motherboard either P5K-E). This was very confusing to me since my old computer with the same P5K-E motherboard had a 3805 in it and is working just fine -- no BIOS setting changes required.

Finally, I noticed that my old computer's 3805 had an old version of the Adaptec firmware on it and the new ones had a more recent but not latest firmware version on it. I doubted this could be the problem, aren't all new firmware versions better than older versions? Perhaps the software developer in me was forcing this assumption on me...

Now the problem was actually upgrading the firmware on the controller. I had to install a floppy drive in this system. The firmware is too big to fit on one floppy. Nicely, Adaptec had broken it into two firmware files and afu.exe. I put afu.exe and the first smaller file on the first floppy and then the second BIG firmware file on the second (it barely fit). Finally, I booted an old windows 98 floppy, put in the floppy containing afu and ran it. Eventually it ask me for the second file. Finally, it upgraded the firmware on the adaptec RAID controller.

Holding my breath, I anxiously rebooted the computer, waited for the Adaptec messages to show on my screen, and hit CTRL-A to enter the array configuration manager.

VOILA! Life is good! Thank you Jesus!

You MUST use the corresponding Adaptec driver version instead of the drivers that will be installed automatically by Windows 2008 (or 2003) in order for everything to work properly. The only way to do this is to install the driver manually:

Device Manager/SCSI and RAID Controllers/Adaptec RAID 3805/Update driver
1) no, don't search for software
2) install from a list or specific location
3) don't search, I will choose the driver to install
4) have disk
5) select AMD64 subdirectory to get X64 drivers

Take-aways:

1. if it don't work, check your BIOS version and Firmware versions
2. if it does work, don't change your BIOS version or your Firmware versions, you could end up with an incompatible version (uh, if it ain't broke, don't fix it!)

if this helps you, paypal me $1 cause I have an extra 3805 now that I have to pay for and they ain't cheap!

When you are editing preferences for your Mac Remote Desktop Connection, there are limited display resolutions you can choose for the remote desktop size.  You can also choose “Fit to Window” but this scales the window instead of changing the screen resolution which sux.

However, there is a way to define your own preferred resolution (assuming the windows box you are connecting to can support it). Use TextEdit to change the saved rdp file for the connection you want to change. See below for an example where I set the resolution to 3820x1150 (this works good with my Dual Head2Go on my MacBook).

On another note, Visual Studio development via VMWare or Parallels is just not fast enough for me on my MacBook Pro. Even though it probably does not significantly reduce productivity, it is frustrating to have to wait for windows to pop-up and builds to complete. I’m just not used to having to wait for anything in my dev environment. I’m thinking this is due to running on a laptop not due to running in a VM on Mac since running a VM on my Win server is actually acceptable. The good news, I suppose, is that this means there is no reason for me to get a high-performance MacBook/MBP when I give this one to my wife the end of this year. Maybe even a MacBook Air will do...

Indexed Views can be now used in any version of SQL 2005 but will not get used automatically by the query optimizer unless you’re running enterprise edition (SQL2000 required the enterprise or developer edition to even create them).

Example (not a good one since this query is not very expensive):

IF EXISTS (SELECT * FROM sys.views WHERE object_id = OBJECT_ID(N'[dbo].[INDEXED_VIEW_TEST]'))
DROP VIEW [dbo].[INDEXED_VIEW_TEST]
GO

CREATE VIEW [dbo].[INDEXED_VIEW_TEST] WITH SCHEMABINDING
AS
SELECT  A.[Name],
                  AMC.[AudienceID],
                  AMC.[CompanyLocations],
                  AMC.[Sections],
                  AMC.[Departments],
                  AMC.[JobTitles],
                  AMC.[CustomFieldValue1],
                  AMC.[CustomFieldValue2],
                  AMC.[CustomFieldValue3],
                  AMC.[CustomFieldValue4],
                  AMC.[CustomFieldValue5]
FROM    [dbo].[AudienceMemberCriteria] AMC
INNER JOIN [dbo].[Audiences] A ON AMC.[AudienceID] = A.[ID]
GO

IF EXISTS (SELECT * FROM sys.indexes WHERE object_id = OBJECT_ID(N'[dbo].[INDEXED_VIEW_TEST]') AND name = N'IDX_INDEXED_VIEW_TEST')
DROP INDEX [IDX_INDEXED_VIEW_TEST] ON [dbo].[INDEXED_VIEW_TEST] WITH ( ONLINE = OFF )
GO

CREATE UNIQUE CLUSTERED INDEX [IDX_INDEXED_VIEW_TEST] ON [dbo].[INDEXED_VIEW_TEST]  ([AudienceID] ASC)
GO

-- this query does not use the index, this is query 1 below
SELECT * FROM INDEXED_VIEW_TEST
GO

-- this query DOES use the index, this is query 2 below
SELECT * FROM INDEXED_VIEW_TEST WITH (noexpand)
GO

You can see that query 2 is much more efficient. Query 1 takes 70% of total time to run while Query 2 takes only 30% of the time. Also, there is no table join for Query 2.

I can vouch that this actually works: http://lifehacker.com/398906/jailbreak-iphone-20-with-pwnagetool

I used the previous version (2.0) of the pwnagetool (macOS version) with no problems whatsoever (there is a newer release) on one iPhone with 2.0, one iPhone with 1.1.4 and on one with 1.1.3 installed prior to using this tool.

I created only 1 patched 2.0 ipsw file using this tool. And then I restored the phones from DFU mode using this same ipsw file. You have to hold down option and click restore so iTunes will prompt you for the file to use. It’s all documented on the lifehacker link above.

Finally, I recommend getting the $19.99/month Blackberry Internet Service from T-Mobile. Then all you have to do is set the APN for the network on the iPhone to “wap.voicestream.com” and everything will work, yup Safari, YouTube, Google maps, App Store, … it all just works.

Good Luck, all I can say it worked for me with three gen 1 iPhones. This does NOT work with the new 3G phones. Wife and kids are happy with them. I gotta get one more so I can pwn it for me!

I love the pineapple dudes!

I’ve had MagicJack for over 6 months now and have been using it as a second line at my house. I must admit that I am underwhelmed. Sound quality is not great, neither when using a headset on my computer with it nor with a regular phone plugged into it. It probably sounds as good as a cell phone with 4 out of 5 bars except for all the occasional burps and squeeks that come from MagicJack. Skype sound quality is much better. However, the price of MagicJack is really cheap. There is no free ride, you get what you pay for. I guess the real test will be when renewal comes up, will I renew it? Don’t know yet…

Anyway, I’ve found that it works better for me just to connect it to a regular phone and not use the headset on the computer where MagicJack is being run. So I have it hooked up to my Windows 2003 X64 Server and am running it as a service. This works as well as any other scenario for me. I found instructions on how to run MagicJack as a Windows Service here, it’s really pretty easy to do.

I have been using Firefox 3 on both PC and Mac since soon after it was released and have to say that I am pretty impressed. As a developer (and also when I am testing) I need to have “display a notification about every script error” turned on in IE so that I know if any JavaScript errors occur on the code I'm slinging. Unfortunately, this causes errors to pop up regularly when I am using IE on other sites that I read or search which really sux. So now I use Firefox for searching and reading and IE for developing/testing which is much better.

Also, there are some pretty cool add-ons to make Google stuff much easier to look at. Google (gmail, reader, gcal, …) stuff usually works really well but it’s just plain ugly. The add-ons: Better Gmail2, Better GCal, Better GReader make these things look much better and add more functionality.

I have to admit that I do like FireFox better than Safari as well and it is nice to have all the same shortcut keys, add-ons, etc on both platforms.

There are a lot of ways to do type casting in C#. Hopefully, these examples can help you decide what the best method for each situation you run into…

const int theInt = 12;
const string theString = "fred";

object boxedInt = theInt;               // value of theInt is boxed into boxedInt
object boxedString = theString;         // value of theString is boxed into boxedString

///////////////////////////////////////////////////////////////////////////////////////////
// this method of unboxing throws exception on failure
///////////////////////////////////////////////////////////////////////////////////////////
int unboxedI = (int)boxedInt;           // unbox boxedInt to get the int value
Console.WriteLine("(int) boxedInt={0}, isNull={1}", unboxedI, (unboxedI == null));

string unboxedString = (string) boxedString; // unbox boxedString to get the string value
Console.WriteLine("(string) boxedString={0}, isNull={1}", unboxedString, (unboxedString == null));

try
{
    unboxedString = (string) boxedInt;
} catch (Exception e)
{
    Console.WriteLine("(string) boxedInt={0}", e.Message);
}

///////////////////////////////////////////////////////////////////////////////////////////
// this method of unboxing returns null on failure (can't do value types with as)
// unboxedI = boxedInt as int; is a compiler error, int is a value type
///////////////////////////////////////////////////////////////////////////////////////////
unboxedString = boxedString as string;  // returns value of boxed string
Console.WriteLine("boxedString as string={0}, isNull={1}", unboxedString, (unboxedString == null));
unboxedString = boxedInt as string;     // returns null
Console.WriteLine("boxedInt as string={0}, isNull={1}", unboxedString, (unboxedString == null));

///////////////////////////////////////////////////////////////////////////////////////////
// this method of unboxing throws exception when it can't determine how to convert
///////////////////////////////////////////////////////////////////////////////////////////
unboxedI = Convert.ToInt32(boxedInt);
Console.WriteLine("Convert.ToInt32(boxedInt)={0}, isNull={1}", unboxedI, (unboxedI == null));
unboxedString = Convert.ToString(boxedString);
Console.WriteLine("Convert.ToString(boxedString)={0}, isNull={1}", unboxedString, (unboxedString == null));
try
{
    unboxedI = Convert.ToInt32(boxedString);
}
catch (Exception e)
{
    Console.WriteLine("Convert.ToInt32(boxedString)={0}", e.Message);
}
unboxedString = Convert.ToString(boxedInt);
Console.WriteLine("Convert.ToString(boxedInt)={0}, isNull={1}", unboxedString, (unboxedString == null));

///////////////////////////////////////////////////////////////////////////////////////////
// this method of unboxing throws exception when it can't determine how to convert
///////////////////////////////////////////////////////////////////////////////////////////
unboxedI = Int32.Parse(boxedInt.ToString());
Console.WriteLine("Int32.Parse(boxedInt.ToString())={0}, isNull={1}", unboxedI, (unboxedI == null));
unboxedString = boxedString.ToString();
Console.WriteLine("boxedString.ToString()={0}, isNull={1}", unboxedString, (unboxedString == null));
try
{
    unboxedI = Int32.Parse(boxedString.ToString());
}
catch (Exception e)
{
    Console.WriteLine("Int32.Parse(boxedString.ToString())={0}", e.Message);
}
unboxedString = boxedInt.ToString();
Console.WriteLine("boxedInt.ToString()={0}, isNull={1}", unboxedString, (unboxedString == null));

///////////////////////////////////////////////////////////////////////////////////////////
// this method of unboxing allows you to define the behavior, throw an exception if you want.
///////////////////////////////////////////////////////////////////////////////////////////
int? nullableUnboxedI = null;
if (Int32.TryParse(boxedInt.ToString(), out unboxedI)) nullableUnboxedI = unboxedI;
Console.WriteLine("Int32.TryParse(boxedInt.ToString())={0}, isNull={1}", nullableUnboxedI, (nullableUnboxedI == null));

nullableUnboxedI = null;
if (Int32.TryParse(boxedString.ToString(), out unboxedI)) nullableUnboxedI = unboxedI;
Console.WriteLine("Int32.TryParse(boxedString.ToString())={0}, isNull={1}", nullableUnboxedI, (nullableUnboxedI == null));

// TryParse for string is simply ".ToString()" ;-). There is a TryParse for Boolean, Byte, Char, DateTime,
// DateTimeOffset, Decimal, Double, Int16, Int32, Int64, IPAddress, SByte, Single, TimeSpan, UInt16, UInt32, UInt64 

/* results of running this code copied from the console window:
    (int) boxedInt=12, isNull=False
    (string) boxedString=fred, isNull=False
    (string) boxedInt=Unable to cast object of type 'System.Int32' to type 'System.String'.
    boxedString as string=fred, isNull=False
    boxedInt as string=, isNull=True
    Convert.ToInt32(boxedInt)=12, isNull=False
    Convert.ToString(boxedString)=fred, isNull=False
    Convert.ToInt32(boxedString)=Input string was not in a correct format.
    Convert.ToString(boxedInt)=12, isNull=False
    Int32.Parse(boxedInt.ToString())=12, isNull=False
    boxedString.ToString()=fred, isNull=False
    Int32.Parse(boxedString.ToString())=Input string was not in a correct format.
    boxedInt.ToString()=12, isNull=False
    Int32.TryParse(boxedInt.ToString())=12, isNull=False
    Int32.TryParse(boxedString.ToString())=, isNull=True
    Press any key to continue . . .
*/

Hey, I just found out this new TryParse method for objects that can be Parsed in .NET 2.0. It still falls a little short of what I need but it is a step in the right direction. For example, I have a bunch of utilities methods where I pass in an object and “try” to convert it to a float or int or datetime, … If the conversion fails, I’ll return the “default value”. If the conversion succeeds, I’ll return the parsed value. Here’s an example:

public static float GetFloatFromObject(object objToFloat, float defaultValue)
        {
            float retValue;
            try
            {
                retValue = Convert.ToSingle(objToFloat.ToString());
            }
            catch (Exception)
            {
                retValue = defaultValue;
            }
            return retValue;
        }

Now with .Net 2.0, I can do this instead:

public static float GetFloatFromObject(object objToFloat, float defaultValue)
{
    float retValue;

    if ((objToFloat == null) || (!float.TryParse(objToFloat.ToString(), out retValue))) retValue = defaultValue;

    return retValue;
}

Not only is this less code, but I hear it performs much better since there is no exception handling during the process.

Now, of course, if you want to do something when the parse would fail, you’ll need to do it the old fashioned way and do that something when you catch the exception.

Yeah, you want an iPhone so do I but what I need is T-Mobile@Home. Why?

1) For $10/month, all calls made at home via WIFI on my cell phone are free, they don’t count against my minutes! (WIFI phone from T-Mobile required)

2) You can reduce your plan minutes because of #1. Save money by getting a cheaper monthly plan to make up for adding $10/month.

3) #1 also solves the problem of less than ideal signal strength inside my house since the phone will be using WIFI instead of GSM.

4) All T-Mobile plans are cheaper than ATT.

5) T-Mobile SMS/MMS/… monthly fees are less than ATT.

6) T-Mobile internet monthly fees are less than ATT.

7) If you really want an unlimited plan, it is still MUCH cheaper with T-Mobile. $99/month for first phone, $49/month for phones 2-5 (ATT is $99/phone for all phones). If you get unlimited, there is little point in T-Mobile@Home except for #3.

8) After July 11, I suspect there will be millions of iPhones (1.0) on eBay for a good price that can unlocked to use on T-Mobile when I need to look kool (sp). I won’t be able to use the iPhone with HotSpot@Home though (c’mon hackers). But when I hit the road, I’ll just put my SIM card in the iPhone…

9) You can get rid of your POTS landline! Save even more money!

10) there has to be 10 items in this list right??? Because I can!

What about iPhone 2.0? 3G speed? etc? I guess you get what you pay for!

Man, I hope T-Mobile reads this and gives me a free Android phone to beta test for this free advertising!!!

Why would you need to have the full-blown version of Management Studio when you can just download and install the Express version of Managment Studio you ask? Because I want to be able to use SQL Refactor from Red-Gate. I don't really need a full-blown install of SQL Server for my development purposes, all I really want is SQL Express and the full blown Management Studio in my Developer VM.

I got this info from here (kudos to him) but I wanted to put a copy in my blog so the next time I need to do this I won't have to search for it...

After you uninstall SSMS Express:

1. Click Start, and then click Control Panel.
2. Double-click Add or Remove Programs.
3. In the list of installed programs, click Microsoft SQL Server 2005, and then click Change.
4. On the Component Selection page, select Workstation components, and then click Next.
5. Click Next.
6. On the Change or Remove Instance page, click Change Installed Components.
7. Click Client Components, and then select Entire Feature will be Unavailable.
8. Click Next.
9. Click Install.

Now you should be able to install from the full-blown edition of SQL Server Management Studio (and subsequently SQL Refactor!)