| Personal background |
|---|
Software Developer by trade since 1994 professionally specializing in MIS/IS/IT programming (the bread-n-butter/steady eddy end of the field, no shortage of work ever) and a 2 year stint as a network engineer/admin during my 1994-1996 period to round myself out better and this has helped over time as a developer as well (especially regarding deployments).
Currently I am using Microsoft Visual Studio 2005 + SQL Server 2005 on Windows Server 2003 SP #1 (with IIS 6.x in the mix of course) fully hotfix patched @ work and love it! Excellent IDE. IMO, Visual Studio 2005 provides very safe code for business plus fast build cycles and the MOST elegant feature-filled IDE I have ever seen! Good combo w/ SQLServer 2005 riding on Windows Server 2003 SP #1 fully hotfix patched.
I use mostly VB6/VB.NET work for Windows clients and thin-client ASP/ASP.NET type apps for business reporting and DB front-end application (typical business MIS/IS/IT fare) in combination with Crystal Reports and SQLServer Reporting Services.
(Good job Mr. Heijelsberg is all I have to say, he developed/architected my FIRST and FAV tool in Borland Delphi 1.0 onwards (my_fav_vers:= 3.0-7.0;) during his tenure @ Borland (which I use for shareware/freeware creation since I have been into that since 1994-1995 or so)).
PLUS, occasionally, I develop w/ Access as well, it is what I started on back in 1994 w/ VB 3.0 in 16-bit Windows 3.x with DOS development efforts!
(And, of course, SQL query and T-SQL usage occurs regularly on the job also as std. fare! I like SQL...)
Anyways - I am continuing the SETI@Home project, from the original one which I joined back in 1999 and this round, this is the system running things @ home:
===========================================================================
---------------------------------------------------------------------------------------------------------------------------------------
HARDWARE SIDE:
---------------------------------------------------------------------------------------------------------------------------------------
AMD Athlon 64 x2 4800+ Dual Core CPU (1mb L2 cache per core = 2mb total)
ASUS A8N-SLI Premium motherboard
PCI-Express 16x/4x/1x w/ PCI-X slots ontop of the normal 3 PCI ones: The real plus part of this mobo is that it allows SLI-ing the GeForce too one day w/ yet another GeForce 7900 GTX OC by BFG no doubt one day here!
NVidia GeForce 7900 GTX OC by BFG (512mb GDDR3 RAM, x16/x8 PCI Express SLI-able type video card)
Dual Western Digital Raptor X 10,000rpm 150gb SATA 2 16mb buffers onboard each harddrives (for RAID 0 via controller listed next)
Promise Technologies SuperTrak EX8350 PCI-Express x4 slot (128mb RAM onboard of ECC type) Caching SATA HDD Controller (with Intel IOP333 XScale I/O SubProcessor to offload CPU and manage RAID functions AND other I/O)
INTERESTING PART, THAT LAST ONE: Here is its technical readout:
http://www.promise.com/product/product_detail_eng.asp?segment=RAID%205%20HBAs&product_id=156
CENATEK RocketDrive 2gb Solid-State PCI RamDrive card (1gb partition for pagefile FIRST, w/ 1gb second partition for: temp ops of apps and OS (like services logs + system eventlogs placements via registry hacks and app internal settings via reg hacks, gui options, or ini file edits), logging by apps + OS, AND webpage cache placements... plus, for running SETI@Home 2 from as well)
* The CENATEK RocketDrive Solid-State 2gb Ramdisk is where the application suite for this project runs from, to avoid std. SATA/ATA-IDE/EIDE (or even UltraScSi disks) weakness of mechanical head movement slowdown vs. running in RAM and to have the FASTEST possible access to files (especially for writes, with me upping the SETI@Home 2 models write to disk settings from the std. settings in the client)!
AND to have fastest possible seek times of the files as well... 1000x faster than mechanical HDDs have here, because of no moving parts and read/write head movements latency.
SONY DvD Dual ( + - ) Format RW DRU-810a double-sided capable burner unit (lol, definitely has nothing to do with speed of things on this project but, it is there for FAST backup!)
I.E. - Every little bit helps imo!
(E.G. - The slowdown/bottleneck inherent to std. harddisks is not a problem on one of these units on file seeks and accesses (since there is NO moving parts), and it has its own backing powersupply, retaining state of the data on its partitions between reboots of the system as well)
Again, also the FASTEST possible access/seek times is possible using ramdisks (of ANY kind, not just solid-state disks but also software based ramdisks), and even vs. the fastest std. HDDs out there from the most modern of the SATA/EIDE or UltraScSi camps - 10,000 rpm AND 15,000 rpm disks of any kind, notwithstanding, the solid-state rocketdrive outperforms them hands down...
Yes, I have compared this unit to my WD RAPTORS above (@10k rpm + 16mb buffers onboard them EVEN IN RAID 0 riding on a caching controller) and they absolutely, fast as they are (which is the currently 2005 fastest overall SATA/EIDE drive which beats most UltraScSi models)?
Well, even those raptors of mine @ 10k rpm get smoked in benchmarks tests by the results this rocketdrive solid-state ramdisk by CENATEK produces for results on the benchmarks I have run, everytime.
---------------------------------------------------------------------------------------------------------------------------------------
SOFTWARE + OS OPTIMIZATION SIDE:
---------------------------------------------------------------------------------------------------------------------------------------
** Running on Windows Server 2003 Enterprise Edition SP #1 + FULLY patched/hotfixed updated AND fully tuned (per the settings in this URL for both security AND speed) to-the-max for performance in either area:
APK Internet Speed and Security Tuneup:
==============================================
http://www.avatar.demon.nl/APK.html
and
APK Local System Speed Tuneup:
==============================================
http://www.avatar.demon.nl/APKTuneup.html
*** Running BOINC latest client 5.2.13x with Intel C++ 9.0 X86 Pentium 4 SSE/2 Optimized via IPP 4.1 libs modded clients portions!
(E.G.- seti 4.18 exe (the workhorse program that does the actual unit processing) showing easily 25% gains already over the originals running on the CPU listed above runs here, in 2 instance, @ HIGH CPU PRIORITY by setting it manually via taskmgr.exe as they both run).
(I do NOT set HIGH CPU PRIORITY on any other program in the suite for this project - just on the seti*.exe, but not on boincmgr.exe, or boinc.exe. Only seti 4.18 (HIGH on both instances of seti*.exe running here @ once via boincmgr.exe and boinc.exe) gets that and boinc.exe and boincmgr.exe BOTH run @ LOW CPU priority and I stop the Explorer.exe shell even, plus use the tuning tricks for IP speed online AND local system speed from the 2 urls I listed above - they do help. When you optimized those, other apps and the OS itself get sped up, and everything running gains imo thru greater overall system efficiency improvement).
===========================================================================
:)
APK |
| Thoughts about SETI and SETI@home |
|---|
IMO, SETI is not only a noble effort that one day hopefully will help us travel the stars AND that its data will act as a guidepost/map which hopefully will help humanity find/trackdown other civilizations out there to communicate with them via finding them first (hopefully E.T., NOT THE BORG), and then secondly benefit hopefully from it!
It also functions, imo, as a pretty good benchmark of your computer speed as well (bonus).
E.G.:
My initial 50 units were done on an AMD K6-III @ 450mhz, taking 24 hours each.
Then, my next 310 were done via a Dual CPU/SMP 1ghz cpu rig, taking my per-unit processing times down to 6 hrs 58min each!
Now, my latest rig (Intel Pentium 4 std. edition 512k L2 cache + running the units from a CENATEK RocketDrive Solid-State disk partition (ramdrive in hardware)) which got my times down to:
2 hr 59 min 17.6 sec
Per unit on the original project. So, like I said earlier - this project and its programs can be viewed as a good benchmark test in and of themselves!
FINDINGS FOR SETI@HOME 2 PROJECT USING BOINC MODDED CLIENTS FOR FASTER PROCESSING:
The Crunch3r modified/highly-optimized version of the BOINC + SETI 4.18 executables optimized for my Pentium 4 CPU type + SSE/2 proved itself FAR faster than the stock/oem builds from SETI@Home, @ 1hr. per 2 units being finished, by using Crunch3rs build of SETI 4.18x...
So I went with using Crunch3r builds, naturally.
Although, later (update) below, you will read that I ended up using portions done by Trux too! See below for more on that account/note...
This superiority (for now) in Cruncher builds here is doubtless because his optimizing the client via Intel C++ 9.0 (and the IPP 4.1 libs used for the SSE/2 instruction set optimization) does a better job on floating point operations data than the stock/oem models, which is MOST of what goes on in SETI 4.18x (fpops) as it is the actual processor/workhorse on said data in the units we receive.
MSVC++ optimized builds tended to show superiority in INTEGER ops (iops) and may in fact be a better choice for reoptimization of boincmgr.exe and/or boinc.exe though. Something to keep in mind.
UPDATE - on that LAST note, about using MSVC++ rather than the Intel compiler to optimize boinc.exe I have changed to Trux builds for that - higher benchmark results and also VERY CLOSE in fpops, and FAR better in iops results (12/30/2005).
So, for right now, I run the units AND application + dlls AND datafiles from the Solid-State disk and give the BOINC.EXE normal cpu priority and SETI 4.18 executeables processes (2 run here @ once always) @ HIGH cpu priority, AND I do the OS + process tuning noted in the URL above as well, lastly running boincmgr.exe @ LOW CPU PRIORITY.
This type of setup here seems to have worked out for the MOST efficiency/speed possible so far in per-unit processing times being kept as low as is possible.
I have gained in my experimentations and nagging folks @ the community messageboards for this project for their knowledge of this projects code and dataset processing, and that was the point of it - faster per unit process times!
Above all - The folks @ the messageboards for this project were all very helpful and patient answering my queries and more about this project and I highly recommend them as people to ask about this project and how it works and can be improved performance-wise, by ALL means.
A bonus in addition to contributing to a future potentially GREAT cause imo.
AND, lastly - My interest in participation in this project is in part largely simply because I watched Star Trek on T.V. as a kid in the late 1960s: Capt. Kirk, Spock, AND Bones were some of my early heroes!
(So, hey: You never know but, just maybe someday? A real Mr. Spock type will be using my data I generated here (along with you all, early stellar cartographers all in a real way, imo) to pilot a real enterprise... who knows)! apk |
| Your feedback on this profile |
|---|
| Recommend this profile for User of the Day: |
I like this profile |
| Alert administrators to an offensive profile: |
I do not like this profile |
|
|
