I have been doing a bit of research in my next notebook because the current one is starting to fail (hp8440p), and I think that it is worth to share the insights.

Display

The display is one of the main characteristics. After all, it is the think that I look most of the time.

I like large displays but they make the notebook too bulky to carry. After all, I use a large 30 inch display at work. The main point of the notebook is for my work at home, coffee shop, or when I travel. The constrain is between larger than 13 and less than 15 inches.

• Between 13 and 15 inch display.

• 300 or more nits or cd/m^2. My current display is just 220, and it sucks.

• Super thin bezel. This is not a must have feature, but they are a way to save real state (1 inch more display with similar notebook size)

• At last 1600x900 resolution. A larger resolution may be good but problematic. The reason is that the 30 inch monitor has just over 100PPI. I can see up to 150PPI in the notebook, but anything larger will complicate the setup because it may require different setups.

Processor/GPU

I am an architect, so I feel bad saying this, but I would go for a Core i5 2540M or a Core i7 2677M. I would not go for the fastest versions due to power, and I would wait for the Haswell, but it is not a hard requirement.

I would expect a 25% power decrease for the same performance budget, but not clear if I can wait the 6 months (early 2Q2011 that I expect for the Haswell).

For the GPU, the same but even more conservative. I would get the bottom GPU cards from nvidia (315M or 410M). I do not play games, so even the integrated processor could work, but I prefer a nvidia because I tend to run CUDA simulations.

I/O

The main constrain is to have a displayPort. The reason is that hdmi can not drive a 30 inch monitor.

Unless the manufacturer offers a 256GB or larger SSD, the notebook should also provide the option to replace the HD. More likely, it is a 7mm 2.5” drive, and I can use some micron 512GB or a 320GB Intel.

No need to ask for USB3 because it is going to be there.

Keyboard/Chassis/Other

I am not a big fan of the ultrabooks because they are not easy to replace the HD, battery, and they do not have docking stations. The key advantage of ultrabooks is the additional thinnest and structural strength. I would not mind an ultrabook if it already has all the HD specs that I want. Clearly, it is not a must have feature.

At work, I use a happy hacking keyboard, which is difficult to beat by any notebook. I like the keyboards from the ultrabooks, and my current HP elittebook is also pretty good. Difficult to say anything unless you try. The reviews for the Thinkpad X1 keyboard are good all over.

I ran Linux, which means that the notebook should ran it. Some brands like Lenovo or HP should be pretty good, but others may require to be in the market for a while so that I can check that the notebook works fine.

In relation to the looks, I like different and stylish like the Sony Vaio Z. The apple products are nice and well engineering but they are something far from different. Apple has become the brand of the sheep. I may have some of my previous life hacker because I want the notebook to be more elite.

Battery

The battery life should be over 5 hours with normal use. I tend to have meetings and classes, and the last thing to have is a dead piece of plastic.

For longer trips and conferences, I would like much longer battery life. At least 10 hours with an additional battery slice or equivalent.

Current models

This is a list of nearly good notebooks, but with some issues with my setup:

• Sony Vaio Z is a perfect option, but I have a big no no to the fact that it just has HDMI. What was Sony thinking? No display port, no 30 inch monitor. Additionally, the linux support is not ideal as the docking station does not seem to work.

• Thinkpad T420s is pretty good but the display is not good enough. It seems the same as my current hp8440p which is not good enough for anything but working in an office environment. The bezet is a bit thick. Maybe the T430s solves the issues with a nicer IPS display and smaller bezels.

• Thinkpad X1 is fairly good, but the display resolution is not good. The Corning glass is neat, but it seems to add too much reflection.

• Samsung series 7 is still not released. The main issue is that the HD can not be replaced. I am not sure about the looks as I do not want to be too close to apple. Not clear how to have an additional battery for trips.

• ASUS U46SV is a pretty neat new ASUS notebook. The main issues is the low res display (not clear about the nits). Also, it is an ultrabook, so no clear how to replace the HD for a good SSD without breaking the guarantee. No clear how to have an additional battery for trips.

• ASUS UX31 is a unibody. Nice with a low power CPU (2677M?), good display and display port. The main issue is the SDD replacement, and that it looks like a Mac Air (sheep). Not clear how to have an additional battery for trips.

The goal of this post is to try to find a politically agnostic method to decide the funding level for major US Government agencies.

Funding level for research has a huge potential impact in the future of USA. A major player in scientific and even engineering research is the US Government with programs like NSF, DARPA, and department of energy programs.

Traditionally, programs like NSF had bipartisan support from the Congress and the White House, but lately their budget is being more politicized. A libertarian politician could say that government (little g) should not get involve and that private enterprise should cover the costs, maybe providing tax breaks for research intensive industries. At the opposite side , another politician could propose larger taxes in all companies and directly fund research significantly boosting Government (big G) research expending.

Deciding between the two approaches without data is not really scientific because they are more about dogmas than measurements. Both approaches rely on “a way to see the world” and shape it without considering its efficiency. This is not necessary for a most scientific and engineering support, this post just focuses on NSF. The reason is that we can estimate a return of investment (ROI) and decide the funding level.

Research Funding ROI

Research by products from NSF and DARPA programs have helped to create many technology jobs in USA. At the same time, NSF projects mostly covers student support and equipment. This is the list of return of investment (ROI) from NSF funding:

• New jobs

• New companies created

• Increase national security and independent (assumed zero in this post)

• Increase quality and lifetime in US citizens

Raymod Orbach said that “Well over half of our economic growth in the last century came from investing in science and technology,” Orbach served as Under Secretary for Science in the United States Department of Energy from 2006 until 2009. I have not seen data for this statement, but clearly a “portion” of US GDP growth is due to science and technology investment. I would just call unadjusted ROI the ratio between the dollars invested and the GDP delta due to investment.

A rough approximation or back of the envelope calculation looking at 2011 budget, all the research agencies in the US Government including defense, NSF, NIH, agriculture, NASA… have close to 150 Billion budget. NSF budget is just 5.5 Billion.

Looking at the US inflation adjusted GDP, the average GDP growth for the last 50 years is 2.7% approximately. Assuming Orbach 50% contribution, half would represent over 200 Billion annually. This still has over 33% ROI even after including all the funding sources.

Ideally, each agency NSF, DARPA, NIH should have an approximation for their return over the last years. I understand that investment in the last 5 to 10 years may still not be fully deployed. This is why it may be good to look at long term 10 to 20 years GPD inflation adjusted contribution and their budget for that time period.

Research Funding Level

Any investor with a 33% return would invest as much as possible, much more if you consider that interest rates are very low. Nevertheless, we could not invest as much as wanted because there the efficiency decreases as the investment level increases. This is the case because there is a limit on currently available researchers. I am a faculty, and I have a large research team, but I could not triple my research team size no matter how much money I can get because I do not have enough students or time to coordinate a triple size student group. The only way to do it will to build a hierarchy which effectively will mean hiring some post-doc or faculty. As more PhD students graduate and became researchers themselves, the overall amount of researchers available increases.

Agencies like NSF can estimate the amount of too lax or too competitive budgets. If there are too many unfunded “competitive” projects, it means that there is not enough money for the researchers. If all the “competitive” projects were funded, it would mean that NSF had too much money.

Currently, NSF had around 20% funding rate, and this is after constraining the amount of allowed NSF proposals under review. In practice, I can just send 4 or 5 NSF proposals a Year even with my most aggressive alternative. Core programs constrain to two NSF submission per Year, and only a few non core NSF programs may apply to my research topics.

If the amount of unfunded competitive projects were a major source to device there should be more checks to verify that not so good proposals were push to competitive. Some communities like computer architecture are famous for being very competitive and even mean reviewing other researcher proposals, but other communities may have different culture. The ROI for each community and additional checks and balances could help to monitor this possible source of problem.

Why not Industry?

Long term research is a risky business if not done at large scale. Many large labs from the past like Bell Labs and Xerox Park are practically gone. The only exception is SRI and IBM that continue having large labs. Intel also has significant research effort but focused on their products and fabs.

The reason for the dismiss is the high risk. It can be seen in the Xerox Park that developed the windowing system used by Apple and Windows, ethernet, laser printers, and many other extremely successful technologies but it was not able to commercialize. Companies like Apple and Windows may exist because Xerox Park, but Xerox Park itself is gone.

A similar problem happens in the CAD industry were a small number of players (Synopsys, Cadence, Mentor) control the majority of the market. Someone may think that they have huge investments in research, but this is also not the case. They do not have large research teams but large development teams.

If we look at computer architecture, we can see that the majority of architectural ideas and proposals did not come from industry, they came from academia (branch predictors, multiprocessors, out of order processors, energy and thermal work…). Without the NSF and DARPA investment in research, the main companies from the silicon valley may not have ever been created.

Overall Recommendation

Politics could decide to add additional funding to research agencies, but a bare minimum should could be decided so that it has a good ROI and the % of unfunded competitive projects. These are my suggestions:

• Make sure that agencies like NSF have enough resources to fund most of the competitive proposals to avoid wasting opportunities.

• Keep track of the ROI for research agencies in a 10 to 20 years time frame, and adjust the agency budged accordingly when the economy needs it.

• Do not think that industry will provide significant parts of the long term research or even 5 to 10 years time frame.

I just got my T-Mobile G2X as an upgrade to my old Google G1 phone. I enjoyed my phone, but it was time for an upgrade. Being an architect, I could not resist to upgrade from a slow 528MHz Cortex A8 (Qualcomm MSM7201A) to a more decent dual core 1GHz Cortex A9 (Nvidia Tegra 2).

The phone is nearly a copy of the LG Optimus 2X. The reference is LG-P999 instead of the LG-P990. Mostly, it seems to be a change in the radio frequencies.

Some nice features:

• Much faster. The G1 was upgraded to CM6, but it could not manage it. It was a pain to use.

• New T-Mobile wifi calling seems a plus (but it has timeout and connectivity issues).

• Nicer screen.

• Five touch points vs 2 touch points.

• It can be rooted and it does NOT have a locked rom. This is a must have for me. Now, I just need to wait for CM to get me a 2.3 upgrade.

There are tons of benchmarks posted around comparing the Tegra to the competition, so no need to repeat. I just wanted to go over some not so easily found details. As a reference, the boot time is 35 seconds with the default 2.2.2 android.

The display supports 5 touch points. This is a plus compared with several android devices that just support 2 touch (multi-touch of two). Even the Viewsonic Gtablet has just 2 touch points (same for the Google G1).

T-Mobile has a Wi-Fi calling app that it is very useful for places with low coverage. I thought that it would be great, but it was giving me some problems because it disconnects. If I check the application log, I get “unable to connect.” “Failed location update” after the phone goes to sleep.

To get it working, I had to disable the screen lock and the wifi timeout. To disable the screen lock go to “menu” -> Location & Security -> Set up screen lock -> “none”, and choose none. To disable the wifi timeout go to “Menu” -> Wireless & networks -> Wi-Fi settings -> “Menu” Advanced -> Wi-Fi sleep policy and set “Never”. Interestingly, I could not get it work without disabling both options.

update

The T-Mobile wifi still fails after the latest suggestions. It works better but it still fails. Since it seems to be a wifi problem, I am going to root the phone and check the problem. Since I do not use Windows or OS X, the best way to get root is to just download psneuter and run this script.

update 2

I found a way to keep the wifi up. Install the free “Wi-fi Keep Alive” from google market. It does the job once you enable the workaround option 1 and 2 (Inside advanced options). Now, I do not miss calls, but I have to figure out how to increase the battery life which has decreased to less than half a day as a result of the wifi keep alive program.

to root T-Mobile G2X from a Linux terminal

Get root access:

adb push psneuter /data/local/tmp
adb shell chmod 777 /data/local/tmp/psneuter
adb shell /data/local/tmp/psneuter
adb shell

The previous script roots the system only once. Every time that you reboot, you will need to run the psneuter script again. If you want to keep the root without needing to do the previous thing all the time, you should install SuperUser. The instruction is also pretty straight forward.

# Get su and SuperUser.apk
mkdir tmp
cd tmp
unzip su-2.3.6.1-ef-signed.zip

adb push system/bin/su /system/bin/su
adb shell chmod 4755 /system/bin/su
adb push system/app/Superuser.apk /system/app/Superuser.apk

Install busybox to get more decent commands. Just get the latest busybox for your specific ISA. Tegra is an ARMv7, but the prebuild binary was not available, so an armv6l is good enough.

wget http://busybox.net/downloads/binaries/1.18.4/busybox-armv6l
mv busybox-armv6l busybox

# remount /system as rw
adb remount
# if it fails, try mount -o rw,remount /dev/block/mmcblk0p1 /system

# upload the busybox
adb push busybox /data/local/

# Connect with root (adb shell)
chmod 755 /data/local/busybox
/data/local/busybox mkdir /system/xbin

/data/local/busybox cp /data/local/busybox /system/xbin/busybox
cd /system/xbin
chmod 755 busybox

# Create a shortcut to the commands that you may want
./busybox ln -sf ./busybox ls
./busybox ln -sf ./busybox df

# or link everything
./busybox --install -s /system/xbin

Remember that /system/bin is earlier in the path to the /system/xbin, this means that by default you are going to have the not so nice ls and df commands.