CoffeeFiend

Trying it with Windows in safe mode is another option. Why not I just wanted to check if some word addon slows it down (easy and quick way to rule something out)

That isn't what I mean. Not running Word from Windows running in safe mode but rather Word running in safe mode i.e. start > run "C:\Program Files (x86)\Microsoft Office\Office14\WINWORD.EXE" /a (fix the path according to OS language, Office version, etc) This opens word without starting the addins which could be causing problems. Completely pointless IMO. I can't see how swapping CPUs would corrupt the drive's contents or similar. Swapping the CPU itself is trivial (ZIF socket), but the heatsink's pushpins are a bit flimsy so you have to be somewhat careful (and follow the instructions -- rotate to release, etc)

The C2D E4400 is actually a hair slower than his existing E2200 (around 10%; not uprising as they're both allendale, and the E2200 is also clocked faster which more than makes up for the smaller L2 cache). Seeing how it has a fairly nice ICH9R chipset he can upgrade to just about anything that still uses Socket 775. The Celeron E3300 is around $50, and it's about 35% faster (faster but probably not enough to see a big difference); The Pentium Dual-Core E5500 is around $80 and it's almost 60% faster (not too shabby); The Core2 Quad Q8300 is around $140 and it's almost 3x the speed of his existing CPU if you're using all cores (twice as many cores which are about 40% faster each -- so not as fast as The E5500 in single-threaded perf). That's just about the fastest CPU you can still throw on there to get a longer life out of it which doesn't happen to be horribly overpriced for what it is. Then you have the Q9550 and the like which is only 20% or so faster than the Q8300 but costs about $275... or the $330 Q9650 AMD has CPUs that are 50% faster than that for $100 less, and even Intel has faster i5's for $200... The pricing and availability of Socket 775 quad cores sucks. I don't plan on ever upgrading this E7500 because of that. It's literally cheaper to buy a faster CPU and motherboard with it instead. P.S. you still haven't told us about safe mode results.

Sounds like it's not malware or virus/spyware/malware-scanner apps then. You can peek at the bottom, it should say so in the status bar. A network printer perhaps. Lots of them have misbehaving drivers too (I've seen several that made Photoshop crash instantly when creating new documents...) You could try to change your default printer to something else temporarily, just to see if that fixes it. I would also try to open word in safe mode (using /a) then opening the document (file>open or using its icon) to see if that changes anything. if the delay persists, then we also know it's not due to word itself (the app or related "extensions") loading slow but rather the document loading slowly. I don't think I've ever come across a word document that was this big though so I'm not totally sure what kind of performance to expect then (depending on the document complexity of course)...

I would definitely NOT buy one based on the info we have so far. That suggests the drive is NOT your bottleneck. If it was just loading slow from disk, here you'd be talking about 68 kilobytes per second which beyond ridiculously low -- we had FAR better performance on drives that are over a decade old. Even crappy old floppy drives are faster than that. In fact, I've written a totally trivial C# app that checks how fast I can load a 15MB file (well, whatever file you happen to name file.ext) into a byte array, just to see how long it could really take: DateTime before = DateTime.Now; byte[] buff = File.ReadAllBytes("file.ext"); DateTime after = DateTime.Now; TimeSpan ts = after - before; Console.WriteLine(ts.TotalMilliseconds); It doesn't get much simpler than that... The result on a 15MB file (on the first run)? 150ms or so is the absolute max I've seen. Even on a slower drive it wouldn't be much higher than that. So I can't see file I/O accounting for more than perhaps a second or so of that 3 minute 20 second delay, unless there is something wrong with your I/O subsystem at some level (malware? misbehaving driver? security suite slowing disk access a lot? -- although all these would make the computer really sluggish as a whole). Long story short, buying a SSD may potentially reduce that second or less worth of I/O latency, but as is it won't really help with the rest of that 3 min 20 sec you're waiting. Something else must be causing this. Such very large office documents tend to be fairly complex, but it's not like it should peg your CPU to 100% for over 3 minutes. Perhaps it's an antivirus (or malware-related app) that's causing the delay. Or office checking if someone else on the network has it open already (this can take forever sometimes). Hard to guess.

Just by looking (haven't actually tried any of it), this seems to be the culprit: oShell.run "msiexec.exe /i" "Z:\VerdiemSurveyor\Clientx64\Verdiem Surveyor Client.msi /qb SERVER_NAME=myservername.ca SERVER_PORT=5600" There are spaces in the MSI's file name so you'd have to wrap that around with quotes, and you'd want a space after /i too, and you can't just end a string with " add a space and continue by adding another " (if that actually was your intent) -- you'd have to add a & between both for that to work. This fails parsing, and if you added the ampersand to have oShell.run "msiexec.exe /i" & "Z:\VerdiemSurveyor\Clientx64\Verdiem Surveyor Client.msi /qb SERVER_NAME=myservername.ca SERVER_PORT=5600" then it would pass these as arguments to msiexec.exe: 1st arg: /iZ:\VerdiemSurveyor\Clientx64\Verdiem 2nd arg: Surveyor 3rd arg: Client.msi 4th arg: /qb 5th arg: SERVER_NAME=myservername.ca 6th arg: SERVER_PORT=5600 which is not what you want. I would try this: oShell.run "msiexec.exe /i ""Z:\VerdiemSurveyor\Clientx64\Verdiem Surveyor Client.msi"" /qb SERVER_NAME=myservername.ca SERVER_PORT=5600" which passes these arguments: 1st arg: /i 2nd arg: Z:\VerdiemSurveyor\Clientx64\Verdiem Surveyor Client.msi 3rd arg: /qb 4th arg: SERVER_NAME=myservername.ca 5th arg: SERVER_PORT=5600 That should work.

Actually, it's kind of neither cryptography (in the sense of encryption) nor stenography (again, subtle differences in meaning which are arguable). This merely hides a file (you could call that cryptography) by appending a file at the end of a jpeg file and nothing more. It's a pretty darn old and well known trick, very much like using NTFS' alternate data streams (which doesn't actually make the files seem WAY larger unlike this). Stenography rather hides the files in the pixels by slightly altering their color (you extract the infos by comparing to a reference picture). Personally I just use truecrypt.

Silverlight is used in some places. The sliverlight version of Bing maps is pretty slick (although I use google maps most of the time). It was also used to stream the Bejing and Vancouver olympics. It's also used extensively by Microsoft for all their developer video series (from various events, tutorials, etc) Let's hope not! Flash *needs* competition, otherwise it'll just stagnate, very much like IE6 did. Thanks to Silverlight, we're now seeing a lot of extra features being added to Flash -- things like video decoding acceleration (H.264). And it definitely won't. HTML 5 has support for video which could indeed kill a large chunk of Flash usage on sites like Youtube, but with browsers supporting different codecs the future isn't so simple (most likely serve H.264 to everything that supports it, and H.264-played-by-flash to everything else). And that doesn't really address the rest of its uses. And no, HTML5-only features like canvas/video/audio tags + SVG + JavaScript + CSS isn't a replacement in many ways (like for dynamic audio in a game) -- nevermind that all of this runs on a very small portion of web browsers right now whereas Flash works with pretty much everything (nevermind the cross-browser quirks). Also, Flash has a large and active community based around it, a lot of people already know actionscript, it has great development tools for animators, it's easy to use and rapid to develop with, it's a supported output format by a LOT of other tools used by non-animators (like InDesign), etc. Silverlight also has its strong points like being able to reuse existing C# or VB code instead of actionscript. I thought about using silverlight before as you can have pretty amazing user interfaces (on the web) using it (just check out the demos of these controls for example) but I just don't have the time to look into it. I very much dislike Flash: slow loading, CPU/memory hog, browser crashing, battery draining, where bookmarking or the back button don't work, nor does it let you open links in other tabs, nor can you adjust the text size with ctrl + or -, forget about using ctrl+f to find something, can't save pics, it's not indexing friendly nor really accessible, it doesn't resize with screen resolution like HTML does or at least can, it seems to be used mainly by highly annoying slap-the-monkey style ads that sometimes even have sound, it's not really meant for touch devices, it doesn't run on a lot of mobile devices, it has those flash cookies your browser won't delete, judging by its track record it's a liability (insecure), etc. Most of the time it's just used to add very distracting and somehwat annoying animations all over the place and it just gets in the way of actual content. But unfortunately it's not going away anytime soon. I wish nobody would develop those really, REALLY awful flash-only sites anymore (I know a guy who swears by that and thinks this is the future... you should see his sites ) and I really hope that devices like the iPhone/iPad/etc that don't support flash will help, but for some tasks (e.g. games) it's still the best tool for the job.

Writing a carriage return into the registry is pretty easy -- at least inside a REG_SZ value. It's like 3 lines of vbscript... No need to use any fancy API to do it (regedit won't show it but it will definitely be in there). However, if it's stored inside a REG_MULTI_SZ value than that's another story (forget about vbscript then but C++ will do the job just fine using the usual RegSetValueEx function)

There is no such thing as Winows (or Windows ) AntiVirus. There is none. There is a built-in firewall though (which is plenty good indeed). It totally depends. Symantec Endpoint is an enterprise product, meant for centralized management and centralized updates and such (and it's not dirt cheap either -- starting at over $50/year/PC). MSE is a product licensed for home use only. There are no centralized updates or management as it's a home product. I'd still rate it WAY better than Symantec myself despite being free. If you need/want something for work, Microsoft also makes something similar to Endpoint (i.e. the same type of features/licensing) called ForeFront. Windows Defender is something rather different. It's mainly an anti-spyware tool, not an antivirus.

It very much depends what you mean by "lame router". The firmware on every single consumer router (Linksys/D-Link/Netgear/etc) I've ever seen is REALLY lame compared to DD-WRT. It sounds like you're trying to say Linux sucks at routing (definitely not what I'd call a "hack"), where it's pretty much the very thing it's best at IMO (BSD is good there too). Under $500 I'd pick DD-WRT over pretty much anything else hands down (if you tried it you'd understand too). And even if you offered me a fancy $1000+ Cisco or Juniper router, I'm still not sure it would be a better pick in a LOT of cases. Yes, the fancy router would have a faster switching fabric (fancy backplane, custom ASICs, etc) in case you're running an ISP, and IOS/JUNOS have some advanced features that Linux may not have but those typically won't be useful on a home LAN (yet lack plenty of useful ones) and require very advanced networking knowledge (like a CCNE) to setup and configure properly. For a home user, there's very little to gain from using these, nevermind the exorbitant costs and training required. The cheapest pre-built thing I'd consider to replace DD-WRT would be a $950+ Cisco 1811. And even then, DD-WRT Pro on a plain old x86 system (with good NICs and Wifi NICs) would probably be cheaper and better in every single way or just about. BTW, DD-WRT also supports some dual band units, mind you a lot of those suck (crappy built-in antennas and what not). A lot of them also only use a single radio for both bands ("non-simultaneous" dual-band). And there's a LOT of wifi N models that don't support the 5GHz band at all.

That's uncommon in most notebooks because it usually means very short battery life and a lot of heat, and that there are other bottlenecks making performance of many tasks slower regardless (very slow GPUs are extremely common, 5400 rpm or slower drives, often less available RAM, etc) There's very little difference between XP Pro and Win 7 Pro cost wise to most large OEMs (most likely within $10). You're looking for XP which often means extra downgrade costs. Again, hardly any price difference. A 5400 rpm 160GB'er like you're looking at is worth $40 or so. A 5400 rpm 500GB'er can be had for $50. Tiny hard drives aren't a big selling point and there's not much savings either, that's why you see bigger hard drives everywhere. I can't stand glossy screens either...

What do you mean? A wireless N router running DD-WRT will do just about every "managed" feature I can possibly think of through its GUI (very advanced stuff might need to be done from the CLI though). You can access it via telnet, SSH and a web interface. It does VLANs (and you can pick each port's speed and duplex mode), SNMP (syslog too), you can do port mirroring (using iptables), it supports 802.1X auth, you can setup some fancy QOS rules, it supports STP, link aggregation is pretty much pointless on a typical router with only 4 ports but it's still supported (along with other fancy stuff) and TONS more. You get full access to everything, you can mount shares using samba, you can add SD cards (simple mods, useful in several ways), you can increase wifi power (and decent models let you change antennas too), you can tweak the routing tables to your heart's content, there's a fancy DHCP server (actually, 2 of them, and you can use it for PXE and such), TONS of wifi options (including some to setup hotspots) -- more than you'll ever care for, it does bandwidth graphing, some builds have OpenVPN built-in, there's filtering (for P2P, keywords, URLs, etc), access policies and radio scheduling, cron jobs, you can do prerouting towards a squid install, it has fairly extensive IPv6 support (you can run radvd and 6to4 on there, setup a tunnel, etc), there's a decent dynamic DNS client, some have USB ports you can use, you can run several other apps on it, there's the nice wiviz scanner, etc. I honestly don't know what else you could ask for.

$162.51 USD

That's a far more honest rating than most though. Anyway. Often, when a cheapo PSU fails like that, it makes really nice voltage spikes. Those often take out the motherboard (if often doesn't take much), sometimes more... I'd still start by clearing the CMOS like puntoMX said.

They're based on the Pentium-M line. Nope. It's a super-low power and low cost CPU that's totally NOT meant for performance. It has its very own socket 441 which no CPUs besides the Atom uses. The Atom ranges from very slow, like the Z510 which is pretty much on par with a P4 1.3, to the somewhat faster than a P4 3GHz but still very unimpressive, like the most high end D525 which roughly on par with a Pentium D 2.8 ... Not much of an upgrade over what you had, if at all. The absolute most high-end Atom is still fairly slow by today's standards. AMD's low end like the $60 Athlon II X2 250 is almost 3x the speed. Plus, it's not crippled like the Atom (e.g. it has VT-x/AMD-v) and it's easily upgradeable (up to 6 cores -- a drop-in replacement). For a Pentium D with a broken PCI slot I'd say the final price really isn't that bad. What he said.

...right back at you? Extended form? I'd like to see references to that stuff. I never heard of anything like this, much less that is commonly used today. Hmmm, what am I missing here? You can make API calls from plain old asm no problem. You declare the call as extern and compile using the right .lib's -- sample code on request. And you say this based on what evidence? Most "top programmers" I've seen or heard of don't write in assembly. Not because they're unable to (seriously, writing basic asm isn't hard at all and there's plenty of n00bs who are into it too) but rather because it's a waste of resources (time and money) in the vast majority of cases. And even in the "top rated" places that supposedly grab all the new super great programmers, they're still into other languages. Like Google (C++/Java/Python/JavaScript mainly), or Microsoft (mainly C/C++ AFAIK), or Apple (C/C++/Objective C), several others like IBM and Oracle (C/C++/Java), etc. There's plenty of great programmers (who know several complex languages better than I ever will, that seemingly knows every algo and pattern and when to apply them, etc) but just aren't quite as good as most good modern compilers at writing fast asm code. It's not like the the early DOS days where it was trivial to completely PWN turbo pascals' speed (although some BASM or OBJ's compiled in TASM still did the trick to speed up the slow parts without quite having to write every single thing in asm). These days are long gone. Unless you you can write SSE2+ level assembly (including all the fancy SIMD instructions), can do instruction ordering, can keep the pipelines full on fancy multi-core CPUs (and keeping in mind things like how NUMA architectures work and other fancy things), preventing cache misses, etc. Most people just aren't able to match a good C/C++ compiler these days (myself included). It takes FAR more time to write anything this way (massive waste of time) with often VERY little gain (it's really premature optimization), it's not portable between different CPU lines (yes, there is life besides x86 -- ARM, MIPS, PPC and plenty of microcontroller series), or even different "bitness" of the same CPU line (e.g. 16 bit code on a x64 OS, or x64 code on a x86 OS), between different OS'es, etc. So you have to keep several versions of your code which differ quite a bit from each other, and manually optimizing your code for each major CPU "revision" too (different instruction sets available, different architectures, etc) which is extremely time consuming and rather hard. Nevermind the extra complexity it adds (requiring higher qualified programmers to do a great job, which cost significantly more and are a lot harder to find), the very high overhead of maintaining and debugging it (various versions which each have many times more LOCs), etc. So it's no wonder hardly any companies want to make use of it (besides those making specialized things, like compilers or 3D engines for games) as it's totally not cost effective in the vast majority of situations (it mainly seems like a way to ship late & over-budget, and making it even harder finding qualified personnel). That makes it a skill that's not all that marketable, unless you intend to apply on a rather specialized job. Even in the embedded world we're writing a whole lot more C than asm these days. And on the "desktop apps" side, most (90% perhaps) of what I see is still plain old C++/MFC with some .NET and other stuff. Edit: Bored. Here's a really simple hello world type app, written in x64 asm. It pops up a messagebox, and sets the exit code to whichever button's value was pressed. Compile with ml64 (found in "C:\Program Files (x86)\Microsoft Visual Studio 10.0\VC\bin\amd64" for those using VS2010). Assuming your lib search path ("LIB" env var) is set properly (i.e. includes "C:\Program Files (x86)\Microsoft SDKs\Windows\v7.0A\Lib\x64" or similar), you can compile it with (you could use includelib in the file too): ml64 something.asm /link /subsystem:windows /defaultlib:kernel32.lib /defaultlib:user32.lib /entry:main extrn MessageBoxA: PROC extrn ExitProcess: PROC .data capt db 'Hallo Welt', 0 msg db 'MSFN',13,10,'ROX!',0 .code main proc sub rsp, 28h ;adjust/align stack ;1st arg in rcx, 2nd in rdx, 3rd in r8, 4th in r9; FP in XMM0..XMM3; remaining on stack; return val in rax mov r9d, 43h ;uType MB_YESNOCANCEL | MB_ICONINFORMATION; 32 bit value so r9d lea r8, capt ;LPCSTR lpCaption lea rdx, msg ;LPCSTR lpText mov rcx, 0 ;hWnd HWND_DESKTOP (xor rcx,rcx would work too) call MessageBoxA mov ecx, eax ;set 1st param to MessageBoxA's returned value ;add rsp, 28h ;clean up the stack (purely optional before calling ExitProcess) call ExitProcess main endp End While this is trivial, writing a complete desktop app this way (all the controls, all the message handling, all the code to do what it's meant for, etc) would be totally insane. It would literally take me over a year full time to accomplish what I can do in a couple hours using c# and winforms (but yeah, it would use 5MB less of RAM... and probably be full of bugs, not feature complete, late, over budget, not available in native x64, etc)

I'd recommend DD-WRT or Tomato over any OEM firmware. It doesn't even compare.

A lot of us were into it back then (I mean, in the DOS days; I was more of a TASM guy though). There might be some people who are still into it. I did have some fun with ml64, but there's just no way I'm still going to develop PC apps in assembly in this day and age. There isn't one. The closest thing is this place, which is mostly used for batch files and simple scripting anyway (not a whole lot of "real" programming)

There you go. Cheap PSUs VERY often do just this. Quality is everything when it comes to PSUs. There's some quality PSUs under 550W that would easily handle that load, and then there's the no name cheapos... Like this 650W unit that dies at 184W in a load test (more LULZ from other cheap units in the following pages). When it comes to cheap no-name units, wattage claims are little more than marketing lies (lots of under-spec'ed parts, inadequate cooling, etc). It's not uncommon for them to take out the rest of your computer with them when they die either.

If you really want support for a large number of older joysticks, have a peek at this Getting old joysticks to work on a PC is fairly simple in the vast majority of cases. The hard part is acquiring the old joysticks (if you're into that) -- preferably in good condition, and also finding the mating parts for all the proprietary connectors. A lot of them very much sucked, but I guess using the authentic joystick adds to the experience when you're playing an old game. That thing is also using an open hardware design, will open source code as well. You could very easily make something out of a slightly fancier atmel (like a mega128) and be able to support even more and fancier stuff... And it uses the V-USB lib too, so it's quite straightforward.

Such joystick-to-HID (USB) "adapters" are pretty trivial to build these days. There's dozens of hardware designs out there you can reuse (from very popular NES HID conversions to boards that can connect to ATARI/COLECO joysticks and plenty of others). And on the software side, there's plenty of projects with full source code too, although a lot of them just rely on the popular V-USB lib to do the heavy lifting. Even V-USB's website has a whole category of example projects dedicated to this kind of stuff. PCBs of that kind are very easy to route as well (even for a n00b), but unfortunately making PCBs available for a decent price (pennies) requires having to get a lot of them build at once... Nah. You don't need a "real" programmer for this. You can build a very simple ISP programmer out of little more than 3 resistors and a DB25 connector (assuming you still have access to a machine with a good old parallel port). Like this one for example. Not that I've had any gameport joysticks in 10+ years...

That's REALLY weird. I had a very similar problem this week. I couldn't delete/rename/move a couple files. However, "try again" was no help. I had to open a cmd prompt, do dir /x to see the short file names and delete them using that... Like if there was no LFN entry for the files somehow. I should have captured it all using procmon but it's too late now as the files are gone. You might want to use procmon to see what's hapenning (you can attach the log here too so we can look at it, if there's no sensitive information in it)

RAID is NOT a backup! If a accidental deletion, fire/flood, voltage spike, disgruntled employee, virus or burglar makes your files vanish (basically everything, save for a single drive failure which is FAR less prevalent than many others on this list) the files will be gone regardless. Quickbooks is pretty much the standard tool for this. Accountants know it (that's fairly important) and it actually works. I wouldn't waste a massive amount of time on trying to self-develop a sub-par solution over time myself. Too late for a refund, is it? You should have used Bizspark It's meant for startups exactly like yours. It might not be the absolute best choice (e.g. the Phenom 86xx being cheaper and also faster) but overall it's still not a bad pick. I wouldn't exactly say it's a great suite myself (I truly don't care for it -- I'd MUCH sooner run an older 2nd hand version of MS Office). Most businesses rely VERY heavily on Outlook too. As for price, Bizspark would have made that pretty much irrelevant.

There aren't any great solutions that are dirt cheap. We're still using an old version of BackupExec at work (also expensive & not exactly great) but we're likely moving to Acronis Backup & Recovery 10 Server sometime soon, even if it's a bit on the expensive side. That already makes the list of suitable apps pretty short. Those editions aren't meant for servers, and as such will likely cause you problems when the time to restore comes (lost and/or corrupted data is likely -- we're definitely NOT risking losing our data to save $800!) Many apps that runs on servers need specialized backup apps, such as Exchange, SQL Server, Virtual Machines and others. Nevermind all the files that are open, etc.