taks

i'm jealous. i don't sleep much anymore at all (not that i ever did, but now it's worse). not sure if it is withdrawal from xanax or a side effect of zoloft. grrr. at least quitting smoking has been real easy to do. 10 days and counting. taks

keep in mind, it is much easier for a processor IO bus design (e.g. the FSB) to catch up with memory speeds than vice versa. the iteration time between updates is simply slower with the former. the processor itself sometimes spends more time waiting for data than doing actual work (app. dependent, of course, particularly with non-contiguous data locations in memory). i don't know what it would really take to get a genuine performance improvement since cached operations all typically function at half CPU speed (well, the MIPS does/can). doing operations on large blocks of contiguous memory result in 99% of your work being conducted directly out of cache, so as long as the work you're doing takes longer than the time it takes to pre-fetch all the data you need (into cache), memory speed won't make a bit of difference. taks

good luck, btw, GD. we think our phase II prototype development will be funded now which is a big load off of my mind. of course, we've already spent most of that funding in IR&D over the last 6 months. taks

not much you can do inside the box. perhaps putting a filler board in between the sound and video cards, something you don't intend to use or something that does not run at high speed. a LAN card maybe... i would NOT recommend putting anything metallic in your box on or around your components. the risk of shorting something out is just too great if you are not trained to do it correctly (and safely). more than likely a large portion of the noise is conducted anyway, which is a mobo issue that you will not be able to resolve easily. taks

find something new then GD? today i had an EGD - esophagogastroduodenoscopy. basically, they tried to knock me out then stuck a tube down my throat to look at my esophagus, stomach and duodenum. i have a mucosal abnormality in and around my duodenum. in other words, it looks like an ulcer. he's testing for h. pylori but i won't find out for 7-10 days. grrrr. btw, they "tried" to knock me out with versed (like valium). apparently i took twice the dosage as most patients (9 mg instead of 5 mg) as well as 100 mcg of fentanyl (pain killer, normally only 50 mcg is given) but managed to stay awake for the majority of the procedure. it was weird. very weird. either way, versed is a really, really, really killer drug. too bad it acts/wears off so fast and knocks you out for the most part. the after-buzz was great, however. taks

yeah, same here. not a bad adaptation. i saw the original years ago and can't recall how close they stuck to the first screenplay, however. i thought they whacked the one scientist in a completely moronic fashion. it was heroic, but watching i couldn't help but think "why jump in the water when the guy's hand is right there next to the edge?" i also thought they borked the very ending... rushed i think. overall, an entertaining movie. taks

man, why would you waste so much money on best buy? add in tax and you're probably out $75 over newegg on the graphics stuff. taks

i was doing a presentation at the GSPx conference in 2006 (the multi-core applications section) and one of the guys that was porting gcc to work with the cell was there. his work was about as far along as the work that had been done for the BCM1480 (quad-core MIPS), i.e., not very. he tried to turn it in to a bit of a competition during my speech (which was about parallelization), but we had different target applications: i was targeting under 30 W with a high FLOP/W ratio, he was targeting over 100 W, just for the freaking chip, with a relatively low FLOP/W ratio. thanks di. taks

engineering theses/dissertations are typically much shorter than liberal arts versions (e.g., history, english, etc.). the hardest part with the engineering versions is advancing the theory, i.e., you sort of have to invent something (i invented two somethings, and tied them together). taks

ahhh, thought so. even getting compilers to properly deal with SIMD is a pain. at least, gcc (and similar) is incapable of vectorizing code on its own, hence the assembly stuff i'm writing as we speak (figuratively). threading on top of vectorizing is just asking too much for what we've got today. i guess it's a huge AI problem underneath it all. taks

my MS thesis took a month, too. it was about 70 pages or so before i added the software in an appendix. i included none of my software in this document. i would recommend MathType, btw. very easy to use, and much faster than LaTex. there are compilers that actually work for multi-processor architectures? taks

i'll check tonight. i haven't had it in a while here at the office. sorry it took so long to respond. weekend and graduation parties and all that and i spent zero time on the web at home (which means i'm doing this from the office, hehe). i played some NWN2 yesterday, however. taks

doctor of philosophy, i.e. a Ph. D. already got terrible handwriting, so it can't get worse. technically my degree is "doctor of philosophy in engineering with an emphasis in electrical engineering" but realistically it is "doctor of philosophy in electrical engineering with an emphasis in communications and signal processing." our school only offers PhDs in the engineering department (though a nursing one starts this fall), and the other engineering areas (mechanical, computer, aerospace and computer science) were not granting enough degrees to maintain their status. as a result, they combined them all into one to guarantee the state* would not yank the smaller department degrees (EE typically has 3 or 4 per semester, so we were never in danger). taks * note that there is no such thing as a national accreditation program such as ABET for graduate schools. they are all monitored by their respective states or not at all.

so what, judge pope? taks

ah, i'll give that a try when i get home, tigranes. that's about what my hood and gown looked like meshugger, but the gown had black velvet stripes, not blue. i got to keep the hood (which they pinned on during the ceremony) and mortarboard/tassle. the gown was a rental. 115 pages total, including all fluff such as title, lists of tables and figures, copyright page, etc. the actual meat of the writeup describing what i did was probably 30-40 pages with another 30-40 explaining all the simulations. if i had another 6 months, it would have been 200 pages or so since there were a lot of things i wanted to show that i did not have time for. basically, my idea was a means for estimating multipath (aka channel estimation) coupled with a means for adaptively detecting symbols (not unlike an equalizer) in the presence of multiple access interference (multiple cell phones talking to the same base station, for example). taks

i'm slow and large and can't fit anywhere. dead meat, too. if i keep dropping the weight like i have been, however, then i'll be slow and skinnier than a rail, but still dead meat. taks

the insert image link wants a URL and does not provide any "browse" option to pick up a local file. taks

no post doc work other than a paper my advisor and i have planned. i'm apparently the de-facto comm-lab instructor now, so i get to teach a 1 credit hour class once per year. i'd like to pick up the actual comm class as well, but they have the new assoc. professor teaching that. my dissertation was a DS-CDMA receiver. i had an adaptive channel estimator and an adaptive symbol detector combined. DS-CDMA is what most US cellular phones use. taks

nah, i'm arrogant enough as it is, mkreku. you of all people should know that! (it's all in good fun). thanks all. after today, btw, i won't sign as drtaks anyway. one day i've been saying... one day only. never made it home yesterday (ceremony, dinner, pool, crashed on the couch too drunk to think) so today was the day. taks

thanks guys, i appreciate the support. ok, how do i insert an image from a CD? got one of me walking... taks

i woke up with a severe hangover resulting from my initial graduation party yesterday. i have to go get the keg for today's party. oh, i guess that means yesterday i graduated and became dr. taks. drtaks

fortunately, i believe i have finally met my insurance deductible. it has been expensive... drtaks

make that three. drtaks

as if i needed to be any more arrogant than i already am. wrote it that quick, yes. i spend over a year working on it, however. some of the writeup was left over from my comprehensive last april, but i had to retool that pretty significantly anyway. i can write, if nothing else, very quickly. too bad i can't do creative writing stuff... it's all technical. drtaks

twice just now, though with two different tabs. drtaks