Happy July 4th.

July screenshots from my new computer:

From now on, all email I send from both my desktop and my laptop will be digitally signed with PGP.  My public keys are here on my site at pgp.jwcxz.com.

In order to be able to sign and encrypt messages, I needed to install and configure gpg to run correctly.  I initially had a little trouble getting it to connect to KMail, so here are the steps I ended up taking after creating my keys.

Configuration Files:

~/.gnupg/gpg.conf:

use-agent

~/.gnupg/gpg-agent.conf:

pinentry-program /usr/bin/pinentry-qt
no-grab
default-cache-ttl 86400
max-cache-ttl 86400
disable-scdaemon

Then, I created a start script in ~/.kde4/env/ that does eval `gpg-agent –daemon` and logged out and back in.  I have set gpg-agent to check for my password only once every 24 hours. Simple as that (well I had some complications, but in retrospect, they were trivial).

Sorry they’re a little late, but here they are.  I had to put them on another server for space reasons, but I’ll sort it out later.

http://jwcxz.com/pictures/grhr09

There are no words in the English language to describe how awesome this computer is.  But first, the build story.

I ordered everything in the evening on Tuesday, June 9th and by Thursday afternoon, everything except for the power supply had come (the PSU was coming from Memphis while everything else came from Edison, NJ).  I managed to contain my excitement on Thursday and only checked to see that everything came and read the instruction manuals.  I got up bright and early Friday morning and began building.

The whole process was fairly easy and standard.  First the CPU went into the mobo (and made a horrible crunch as it locked, but that was apparently supposed to happen), then a bit of thermal paste, and then the cooler.  Installation of the gigantic CoolerMaster V8 was a little tricky and awkward, especially when I had to flip the whole motherboard over and rest it on top of the cooler to screw in the rear plates.  the mobo went into the case without a hitch, as did all 6 of my hard drives, the optical drive, and the 2 extra case fans that I purchased.

I took full advantage of the Antec 1200’s main cable management feature: a recessed area on the other side of the moutherboard mount.  I wired up everything fairly neatly, assuming that I could then just use the remaining space for the power cables.  Originally, I had anticipated receiving the PSU that Friday, but when I checked the tracking site again, I found that the package was just a few hours too late to have gone on Friday’s local shipments.  So, I would have to wait to Monday to be able to power up my computer.

Somehow I managed to get through what seemed like the longest weekend in history (saw Angels and Demons… very good!) and Monday afternoon rolled around.  I stuck the PSU in and pretty much immediately realized that I would have to rewire the entire computer because the power cables were just too bulky and my system was becoming a mess.

So, I spent another two hours laying all the cables over again until I approached a configuration I liked.  Finally, after almost a week of waiting (well, more like two or three years), I had a computer that would be far faster and more powerful than my needs would require for probably the next decade.

I fully anticipated seeing all sorts of random problems when I started the computer up for the first time; that’s how it usually works with me.  After waiting about 20 seconds for it to POST, my suspicions were almost confirmed, but to my surprise,  everything seemed to be okay.  I went into the BIOS and made some initial settings.  I noticed that my second IDE drive seemed to be missing, but I think that’s probably because I don’t have the master/slave jumper settings setup correctly.

I installed Windows 7 RC1 onto my 160 gig SATA drive without any issues.  For the heck of it, I tired out Nexuiz and was amazed to see the game consistently run over 100FPS with all of the highest quality settings enabled.

Next, came Linux.  Setting up software RAID was a ridiculously easy task and before long, I had Arch Linux up and running.  I’m using 3 640GB Western Digital Caviar Black drives in RAID5, so that gives me about 1.2TB of space.  I gave about 100MB to /boot (ext3), 150 gigs to / (ext4), and 25 gigs to /var (ReiserFS for optimized pacman performance), leaving about 1.1T for /home (xfs with allocations properly alligned to the stripes).

After spending all week copying my old settings and files over to this computer, here I am.  I finished overclocking the system to my liking this morning.  The results are something like this (stock values are in brackets):

• Base Clock (BCLK): 160MHz [133]
• CPU Freq: 3200MHz [2667], multiplier still at the locked 20X
• RAM Freq: 1600MHz [originally 1600MHz, but my mobo didn't want to set it that high], multiplier 10X
• Uncore Freq: 3200MHz [2667]
• VCore: 1.2V (stock voltage!)
• VCore VTT: +100mV [+0mV] (had to increase it because the memory controller would fail otherwise)
• VDimm: 1.65V (OCZ’s recommended voltage, but I think I could go lower)

The VCore VTT threw me off for a while; I didn’t expect that I would need to increase it at all.  The BIOS has a base voltage of about 1.17V or so and allows you to choose how much you want to increase it by.  The fact that VCore is still at stock values is pretty nice; my CPU doesn’t heat up that much more than it did when my BCLK was at the stock 133MHz.

Before I overclocked the computer, I never paid attention to the voltages that the mobo was automatically choosing for the CPU and memory.  Thankfully, I hadn’t run the computer too long before I realized that VCore was at 1.3V and VDimm was at 1.7V!

I’m fairly sure that I could really push this system further.  With a small voltage increase, I bet I could get up to 3.8-4GHz without too much of an issue.  But why would I need to?  The system is already far faster than probably all of my other computers put together and I honestly don’t need any more flops than I have right now.  Plus, I don’t want to risk damaging any components or overly reducing their lives.

I still need to fix some of the cabling; the side panel for the cable management trough doesn’t quite fit due to some leftover pesky SATA connectors, so I have to do a little work there.  Also, I need to figure out why my second IDE drive isn’t recognized.  But other than, that, things went quite swimmingly. 

Anyways, here are some pics from the build process (I wish I had taken more ).

And a screenshot:

Edit: And also a shot of my new Conky config with box drawing around some of the fields.  It’s getting out of hand now.

Well, the parts for my new computer are now ordered, so now I begin the excrutiating wait…  Almost all of the components are from the list in my last post, except for the graphics card, which is an EVGA 01G-P3-1181-AR, a slightly nicer card than what I was originally considering (NewEgg also deactivated the 1285 anyways).  The total price (including a DVD burner) was about $1683, plus another $32 or so for shipping.  I’m now clearing off my computer desk so that I’ll have a nice big area to work on.

Anyways, I also just got back from a cruise to Croatia and Greece.  With over 1100 pictures taken between my mom and myself, I had a lot to sort through.  I also took them on very high quality (2592×1944), so the album takes up over 2 gigs.    I’ll upload a few of the best ones when I get a chance.

Here’s the current snapshot of my wishlist on NewEgg.

• Intel Core i7 920
• OCZ Gold 6GB DDR3 1600 SDRAM
• EVGA E758-A1 Mobo
• EVGA 01G-p3-1285-AR GeForce GTX 285
• 3x WD Caviar 640GB
• Antec 1200 Case
• CORSAIR CMPSU-850TX 850W Power Supply
  
• COOLER MASTER V8 CPU Cooler
• 2 COOLER MASTER 120mm Blue LED Case Fans (to go into extra slots in the case)
• Arctic Silver 5

Total cost (without shipping): $1673.88

I may also throw a card reader and/or other accessories in as well.

I’ve decided to go with an EVGA mobo because the reviews on NewEgg are pretty decent.  I was debating whether or not to go with the E758 or the E759 but eventually decided on the former because I don’t think I am ever going to use true 3-way SLI and 24GB of RAM; I simply couldn’t justify spending another $200.  Plus, 3-way SLI isn’t very good right now (bottlenecking issues still exist on most motherboards).  Unfortunately, the E758-A1 is out of stock right now, so hopefully by the time I order the parts (early June), it will be back.  If not, NewEgg sells another E758, but without all the cables and extra stuff that I may or may not need.

I found a good guide on overclocking the Core i7 920 CPU.  The disadvantage of buying the 920 (as well as the 940) over the 965 is that by default, the highest CPU frequency multiplier (20x) is selected (22x for the 940).  So, in order to overclock the processor, it is necessary to increase the base frequency and then decrease the multiplier for the memory (the multipliers for the cache and QPI bus are usually at their lowest values for most motherboards).  Keeping the Vcore value at the nominal 1.2V, the author of the article managed to clock the CPU up to 3.4GHz without an issue.  This is roughly the frequency that I’m looking to end up at when I overclock my system.  I definitely do not want to push it up to the maximum stable frequency that the author achieved with air cooling: 3.8GHz.

So here’s the setup plan:

1. Unpack everything (taking pictures!!!) and make sure that nothing looks physically damaged.
2. Read some manuals (yes, I’m going to actually try to read them and contain my excitement)
3. Install the CPU onto the mobo.
4. Install RAM
5. Apply AS5 and install heatsink (apparently, since the heatsink is so huge, it’s pretty difficult to install the RAM afterwards).
6. Install mobo into case.
7. Install PSU, hard drives, DVD drive (I have some old ones that I can use).
8. Turn it on and hope like heck it works.

I’ll probably start by installing a non RAID drive for Windows so that I can use Prime95 to make sure that my attempts at overclocking resulted in a stable system.

Afterwards, I’ll install the 3 WD drives in a RAID 5 setup and install Arch Linux on them.  I’ll probably use XFS for my filesystem, but I’m not quite sure yet.  I’ll also install another drive for Windows and/or other operating systems, but it won’t be part of the RAID array.

I’m worried about the performance of RAID 5.  Under certain conditions, a RAID 5 array can suffer huge performance losses.  However,  these losses are mainly limited to situations where there are multiple small simultaneous writes, something not really seen in normal desktop use.  Another option would be to buy a fourth drive and use RAID 0+1 or something.

I have a feeling that by the time I order the parts (early June), the list will change again.

I usually try to keep my blog as commercially neutral as possible, but sometimes, if I find something really good, I can’t resist.  TradeKing is a perfect example.

Back in the fall, I began looking for a good cheap broker.  I have an existing brokerage account that was also tied to my college fund and my IRA, but that’s only for ultra long-term growth.  Instead, I wanted a broker with cheap commissions; in fact, that’s really all I wanted.

I came across several brokers, including some that offered free trades so long as you made a certain number of trades a month.  TradeKing seemed like the least gimmicky out of all the brokers I looked at.  Their pricing scheme is simple: if you trade 10 shares, you pay $4.95 commission.  If you trade 100 shares, you still pay $4.95.  If you trade 100 million shares, you pay (yep, you guessed it) $4.95.  The only exception to this rule is that stocks trading for less than a buck cost an extra dollar for every hundred shares.  Options contracts are $0.65.

The other thing that I really like about TradeKing was that in addition to providing rock solid trading services to independent traders, it also boasts a large community website where traders can share tips and examine strategies.  The community site has a very active forum, a leaderboard, statistics on recently traded companies, etc.  You can elect to display the investments you make (as well as to post notes for each trade) so that others can follow your line of thinking.  Every member gets their own trading blog where they can post more detailed analyses.

So, I signed up for an account and deposited a small amount of money so that I could start trading.  Right now, I’m just investing in stocks and ETFs; eventually I’d like to start trading options contracts too.  I have been following many of the community discussions in hopes of learning how to effectively work with options.

Above all, TradeKing boasts an absolutely spectacular staff.  The CEO, Don, routinely posts interesting news stories to his blog, which I read daily.  Every Friday, he posts his weekly shout-out, highlighting the week’s top performers and interesting discussions held on the forum.

A few months ago, TradeKing held a nice promo where they gave out nice “Elvis-style” sunglasses to people who filled in their community profile.  It ended before I joined TradeKing, but when I contacted Don, he was kind enough to reply back personally and have his staff ship out a few.

In summary, if you are looking for a dependable broker with excellent customer service, yet you wish to be an independent investor, TradeKing is absolutely perfect.  Over the past year, I haven’t had the chance to really use all of the tools they offer, but hopefully, I’ll have much more time this summer to do so.

Important note: this entry was not endorsed by TradeKing in any way.  I wrote this article solely as a happy customer and was not reimbursed by any means for this post.  If you have questions about TradeKing, please contact their customer service desk (contact info is on their site).

I own several cheap green laser pointers and I’ve noticed that some of them behave very differently from others.

I bought my first green laser pointer back in September for a little over $20.  It was this one.  At that time, I had no clue of what a typical 5mW pointer looked like, but I expected that I received somewhat of an overpowered unit since it could burn through trash bags and had a clearly visible beam (as it turns out, the pointer is about 50-80mW based on a few readings I took).

I then purchased a few more pointers for my family for Christmas a few months later.  The units I received were nothing like what I had originally bought.  First of all, they had different stickers on them and the size of the apertures was much smaller.  Secondly, not only were they extremely weak, but on rechargeable (NiMH) batteries, they needed about 30-45 seconds of warm-up in order to produce a decent dot.  With alkaline batteries, they were a bit better.

A few months ago, I purchased a 50mW pointer from LEDshoppe.  It was also bright on alkalines but miserable on rechargeables.  It occurred to me that the driver seemed to be the problem.  Alkaline batteries probably allow the driver to just cross the threshold voltage required by the IR pump diode.  So, rechargeables, which have a nominal voltage of about 1.2-1.3V, cannot quite reach this barrier and as a result, the laser will barely lase.

The driver used in the first laser I purchased clearly had to be a boost driver; i.e. it boosts the voltage output to compensate for low voltage input.  The other drivers were just linear constant-current drivers.

So, since I don’t really want to pay for new alkalines constantly, I decided to look into another method that was already being used by a few members of LaserPointerForums.com.  10440 batteries are 3.6V lithium cells that are the same size as standard AAA batteries.  Therefore, if I use a single 10440 with a spacer in my laser pointer, the output should be at its intended maximum.

There are a few downsides to this:

• Assuming that the driver is linear and outputs constant current, a significantly higher input voltage will result in excess heat dissipation.
• These batteries are typically only around 300mAh, so lots of recharging will be needed, as a typical ~100mW pointer will draw up to 600mA.

While discussing these issues on LPF, member Warske recommended using a diode to lower the voltage going to the driver so that heat would be dissipated in the spacer instead of on the driver board.

I got my 10440s and charger today, so I quickly built a spacer with some cardboard, wire, and a diode.

The results were spectacular.  The laser is drawing its full 600mA or so and is definitely over 100mW (I’ll test it on my LPM when I get home).

Here’s a comparison between my improved laser (on the left) and one that’s doing roughly 50-60mW.

This picture was taken in broad daylight with a limited exposure time, so that's why the lasers aren't particularly bright.

So here’s what you can do if you want to make your green laser pointer more powerful:

1. Buy a pack of 10440s.  I know it’s tempting, but don’t put both batteries into the pointer because you’ll fry the driver.
2. Buy a charger.  I used this one, which seems to be the best.  Be careful though, if you’re using your own lithium charger, make sure it is designed to charge 10440s.  They need a significantly lower charging current than CR123As or 18650s.
3. Buy a regular-ol’-diode from Radioshack or anywhere cheaper.
4. Get some ~22 gauge wire and make two little coils like I did in the first picture.
5. Solder the ends of the coils to each end of the diode.  Make sure that the total length is about the same as a standard AAA battery, but err on the side of making it a little longer.
6. Get some cardboard or thick paper and wrap it around the wire/diode setup so that just the coils stick out on either end.
7. Finally, make a little paper tube so that you can connect the positive end of the 10440 battery with the coil without the chance of that coil touching the side walls (otherwise, the diode is useless!).  Make sure that the diode is in the correct way; i.e. the white painted stripe should be connected to the coil that touches the end of the barrel.
8. Put it all together, and enjoy!

Well, it’s May, so I have about two weeks of classes left followed by a week of finals.  The weather around here is finally decent and all of the trees are in bloom, so I suppose that Spring has officially arrived.  To celebrate the change of seasons, I decided to take some pictures of my room:

The view outside my window is no longer snowy. 

Oh, and here’s a picture of my desktop.

As for the wallpaper, I’ve been messing around with The GIMP’s rendering capabilities.  Right now, it looks a little too cluttered…  I plan to work on it this summer.

I noticed that the newest version of Firefox seems to be significantly faster than the old version in that the random hangs that used to occur have stopped.  This made me happy, but I wasn’t planning on switching back from Opera, whose use of Qt has made my system seem very unified.  Then, I tried Vimperator.

Vim, for those who aren’t familiar with it, is a powerful console text editor designed to provide developers with an elegant interface with a ton of context-sensitive shortcuts.  I’ve never been a particularly big user of Vim because I have always preferred graphical editors.  But in a pinch, it’s definitely a great tool to use.

Vimperator is an extension for Firefox that tries to harness the elegant Vim interface.  First, it gets rid of all the toolbars and replaces them with a single status bar with a command line below it.  It works just like Vim in that you can type : and then a command to do something, or you can use a single keystroke as a shortcut (e.g. t invokes :tabopen, which allows you to open a new tab with a URL).  Autocompletion is nicely implemented.  Theoretically, using Vimperator, you can browse the web without ever having to touch a mouse (no, you don’t have to hold tab down to go from field to field, there is a more elegant solution involving pressing ‘f’ and then typing a corresponding number).

I still like to use a mouse on occasion, so I’ve create somewhat of a hybrid environment.  For example, I added :set guioptions+=B to my .vimperatorrc.local config file in order to leave the bookmarks toolbar up at the top of my screen (I haven’t restored any of the other toolbars, though).

Oh, also a view at night:

I bought a pack of C-Mount laser diodes recently on LPF, but haven’t really had the chance to test them much.  Since I’m feeling pretty sick right now and my headache is preventing me from focusing on any real work, I thought I’d test them out.

I constructed a very simple constant current LM317 driver a while back:

(Since this pic was taken, I’ve replaced the trimmer resistor and added a little switch).  I tested the 20 diodes at a really low current.  The results were pretty impressive, a bunch of 808nm IR diodes, longwave IR diodes, a 635nm diode (!!!), and only 2 dead ones.  I wanted to test a few at their full current though (~1.2-1.3A), but without proper heatsinking the diodes would die.  Suddenly, I got an idea.

That’s right.  I submerged the diode partially in a cup of water.  Poor man’s water-cooling.  Keeping to a fairly short duty cycle (for the sake of the LM317 chip, which dissipates a ton of heat), I found that this little guy burns right through cardboard, dark plastic, etc. even when it’s uncollimated!  The diode was fairly robust, too.  I managed to drop it in the cup of water without it dying.  However, I think my grounding strap wasn’t quite attached to ground because I managed to kill the diode a little later.

Here are a few pics I took with the second diode (~300mA in the first pic, 1.3A in the second):

So, it was a fun experiment, even though I lost a diode.  I can’t wait for my TEC and collimator… Then, I’ll be able to install and run my precious 635nm diode.