Lessons In Hardware / OS Troubleshooting

Esther Schindler writes "We like to imagine that every Microsoft OS installation will work just as well as the company promises. When things don't work out, identifying and remedying the case of failure can be time-consuming and frustrating. This lesson in how to determine why Windows 7 didn't install may help you troubleshoot a problem of your own, and save you from a Lost Weekend. Maybe you'll find this account useful all on its own. But the real key here is that the author is Ed Tittel — who's written over 100 books. If this hardware geek spends days solving a CPU-meets-Windows 7 problem, what chance do mere mortals have?"

Comments

Sooooo

[Anonymous Coward]

He has issues with an "unsupported and unwarranted engineering sample CPU from Intel" with Windows 7... and Windows 7 is of course to blame according to the OP.... *roll eyes*

Re:Sooooo

[SilverHatHacker]

This is Slashdot. Windows is always to blame.

Re:Sooooo

[Anonymous Coward]

"unsupported and unwarranted engineering sample CPU from Intel"

Is he using an IA-32 or x86-64 processor? Yes. So it is a supported CPU.

Re:Sooooo

[Ralish]

I also found it bizarre that at no point did he seem to think of checking the setup logs. Admittedly, it probably wouldn't have helped him in this case, as logs often don't reveal anything in the case of intermittent hardware failure, but really, if I have a problem with setup, the first thing I'd think to check would be the log files in case they turn up something interesting. That's, you know, kind of why they're there...

Re:Sooooo

[Anonymous Coward]

It is an ENGINEERING SAMPLE. These are NOT supported in a commercial sense, if you rang MS for support they would reject your call as using unsupported hardware.

Re:Sooooo

[Anonymous Coward]

it's great to hold ones hands over warmed up prejudices, isn't it?

cb

Re:Sooooo

[lorenlal]

Actually, let's recap the action and the missteps:
Inconsistent failure point during the initial installs. Yes, it could've been a problem with the ISO or the media. He correctly tried re-applying the image and also tested on another machine.

At that point, you don't replace the motherboard. You might as well replace everything else first... Start with slapping the HD into the machine that worked and try the install again. When that worked, that would've reduced the potential culprits to the memory, CPU, and then lastly the mainboard. Memtest would've found no memory issue (which would also indicate that the mainboard is also less likely a problem), so that's when the CPU switch should've happened... Especially since it was "an engineering sample."

Writing 100 books does not an expert make. Of course, I'll grant the guy some slack. Even the best of us have an experience where we throw our better judgment out the window. We make mistakes, or just totally forget how this is supposed to work, get into a panic, and goodness knows what else.

The difference, and where I think this guy made the big mistake? When he decided to post this experience. Would've been much better just writing it like this:

"I tried to go from x86 to x64, and it failed. I troubleshot it like a noob. I'll do better next time."

Re:Sooooo

[lorenlal]

Sorry to reply to self, but there's another problem I have with his execution:

If you're replacing the mainboard... You might as well go whole hog. Get yourself something nice...

Re:Sooooo

[Anonymous Coward]

The comment was taken directly from the site linked. Likely the comment refers to unsupported from Intels point of view.

Re:Sooooo

[Anonymous Coward]

Pahlease.. you're saying Intel couldn't make a faulty chip with IA-32 written on it?

Re:Sooooo

[postbigbang]

The logs aren't very useful, even when they trap things. Unsupported CPUs may also have unsupported or doctored chipsets. Bitching about it, 100 books or no, seems a bit silly. With fast CPUs and weird cache setups, FSB speeds approaching C, you're just going to have problems unless something's vetted.

Moaning about an engineering sample seems nihilistic to me, Windows 7 or no.

Re:Sooooo

[gman003]

Not true. We also blame SCO, the MAFIAA and, lately, Apple.

Re:Sooooo

[Anonymous Coward]

Maybe I'm, the only one, but I usually do a quick web search first, (as well as check the log files), to see if just maybe someone else has had a similar problem and can point me to a solution. It sure beats replacing a bunch of hardware at random.

Re:Sooooo

[ozmanjusri]

if you rang MS for support they would reject your call as using unsupported hardware.

Does Linux run on it?

I suspect the same request for help to the Linux community would be met with a MUCH more enthusiastic response.

Re:Sooooo

[somersault]

Don't forget Sony.

Re:Sooooo

[jimicus]

It would never have helped him - he was using an engineering sample CPU, for heaven's sake!

Having said that, I'm a Linux admin and it causes me no end of frustration when I need to troubleshoot something on Windows and I am painfully reminded that:

- The event log is a PITA to browse through, because you have to double-click on specific events to see the detail. Search doesn't work very well when you're not entirely sure what you should be searching for.
- Application software frequently doesn't write to the event log. If you're lucky it keeps its own separate log, if you're unlucky it was written by someone who thinks a log is what you get when you chop down a tree. (How the Hell any bugger ever troubleshoots during the development process I have no idea. Unless the dev build does create logs but some arse-head middle manager decided to turn them off in the production version).

Re:Sooooo

[BlueWaterBaboonFarm]

...At that point, you don't replace the motherboard. You might as well replace everything else first...

Especially not if it "took somewhere between 60 and 90 minutes". Why the hell do I care what this guy says if it takes him 90 mins swap a mobo. I'm guessing the average slashdotter could do this in under 10 with great ease.

Re:Sooooo

[gullevek]

It doesn't change the fact that this is an engineering sample. Normally nobody will have one. Then complain about Windows it doesn't work with it, is really stupid.

With linux you might get similar errors and I doubt anyone will care about. Why would anyone support an engineering sample ... useless code that might make other more important things break.

Re:Sooooo

[ByOhTek]

*typically* the a non-overheating CPU tends to work to spec (or similar to other models in the line), or not at all, without much inbetween behavior. I can see why he would replace it last, if it were a normal CPU. That being said, as the article stated, he's not using a normal CPU.

With particularly quirky errors, I would go for Memory, Motherboard and PSU as the most likely cause (add disk in this case as it is during file writes - however the disk worked fine with the previous OS, so that mitigates a lot of concern with the disk).

So, were he not using a nonstandard CPU, I would have agreed with his methodology (except replaced the PSU in an earlier step).

With the nonstandard CPU, I'd have replaced that first. There's no baseline for comparison.

Re:Sooooo

[elnyka]

if you rang MS for support they would reject your call as using unsupported hardware.

Does Linux run on it?

I suspect the same request for help to the Linux community would be met with a MUCH more enthusiastic response.

Non sequitur. I know it felt great to say that, but it is still non sequitur.

Re:Sooooo

[elnyka]

The logs aren't very useful, even when they trap things. Unsupported CPUs may also have unsupported or doctored chipsets. Bitching about it, 100 books or no, seems a bit silly. With fast CPUs and weird cache setups, FSB speeds approaching C, you're just going to have problems unless something's vetted.

Moaning about an engineering sample seems nihilistic to me, Windows 7 or no.

Or a desperate call for attention. Nerds, talented or not, are not above (ideological and possibly market-driven) attention whoring :-/

Re:Sooooo

[ozmanjusri]

Why?

Re:Sooooo

[plague3106]

Of course; the linux community has nothing else to do while they sit in their moms basement, but MS has work to do which they hope will earn them profits.

Re:Sooooo

[ta bu shi da yu]

I'm quite curious as to what it was that crapped out though. But installing using an engineering sample really is a bit stupid.

Personally, I find it more curious that if he was willing to fork out so much cash and he was writing a book that he didn't contact Microsoft support and paid for someone to assist him with finding out where it was borking, and why.

Re:Sooooo

[Anonymous Coward]

Of course, does Linux barf on it? Id like to see if it did or not .. if not then yes Windows 7 needs a fix for this one since its not within tolerance of being production ready imho.

Re:Sooooo

[WuphonsReach]

When that worked, that would've reduced the potential culprits to the memory, CPU, and then lastly the mainboard. Memtest would've found no memory issue (which would also indicate that the mainboard is also less likely a problem), so that's when the CPU switch should've happened...

Maybe, maybe not. Memtest has never been good at finding timing issues or partially good memory.

In the case of flaky hardware, your best bet is a stress/torture test of the hardware. Moving lots and lots of data around between the RAM/CPU and checking to see that it doesn't get corrupted along the way. One good tool is Prime95 in torture test mode for 24-48 hours. It'll run your memory/CPU into the ground and expose timing issues, poor cooling, or calculation errors due to the immense amount of work that it forces the CPU/RAM to perform.

Re:Sooooo

[temojen]

For a failed build, our upstream helpdesks usually recommend HDD, then MB, then PS. CPU and RAM are way down the list (strangely above bringing in a network verifier which almost always finds the problem if the HDD and MB didn't fix it and the deployment is network based). This observation is based on most of the big-time contracting services in Canada.

Re:Sooooo

[dave562]

Personally, I find it more curious that if he was willing to fork out so much cash and he was writing a book that he didn't contact Microsoft support and paid for someone to assist him with finding out where it was borking, and why.

If he did that he would have to acknowledge that Microsoft support actually does a very good job of resolving issues with their software. I'm talking about support for production environments, not patching the latest zero day security hole. The author in question probably didn't call support because he knew he was running unsupported hardware. It might be slightly inconvenient to find hardware on the HCL, but that HCL is there for a reason. It provides a known good baseline to troubleshoot against when things go wrong.

Re:Sooooo

[ta bu shi da yu]

Well, not necessarily. He would have at least had isolate the issue to hardware, and they'd have to be pretty specific about it - which would lead to the CPU. Then he wouldn't have forked out enough money for almost an entire new PC!

Re:Sooooo

[Creepy]

Well I had a supported Windows Vista upgrade a couple of years ago and the upgrade not only failed, it bricked my laptop. The repair DVD to restore XP was a waste of a couple of days and I ended up having to ship it back to the manufacturer. Two weeks after getting it back, the GPU failed (it had the notorious nVidia 8xxx GPU - it actually failed again 3 days after the warranty expired) and back to the manufacturer again for a full wipe. Then they failed to set the Vista code so the laptop wouldn't start, so back to the manufacturer again...

    So basically, due to Microsoft's flawed software, manufacturer lock codes on laptops (which are required by Microsoft so Windows can't be pirated and it allows them to bypass activation), my ex-laptop made two special trips to ASUS. That coupled with Vista being a pain in the ass made my experience with it very unpleasant. Incidentally, I do have a Windows 7 laptop now and it works quite well. I'm hoping it lasts longer than my last one (exactly 2 years and 3 days). Incidentally, I don't hate MS software in general, but I do hate their draconian DRM and business practices - in an effort to stop piracy they make non-pirated copies very inconvenient to install and use.

Re:Sooooo

[Creepy]

and RIAA, MPAA, ASCAP, God, showers, deodorant, and most squishy things.

the flame war for jokingly including God may now commence.

Re:Sooooo

[armadaindonesia]

Normally nobody will have one.I know it felt great to say that

Re:Sooooo

[Golddess]

Not God, just (some of) His followers.

Who was it that said something like "I like your Christ, I do not like your Christians. They are very un-Christ-like."?

Re:Sooooo

[ginbot462]

And Poland

Re:Sooooo

[Terrasque]

Strange, since RAM and CPU is easier to replace than MB. And in my personal experience RAM is the #1 culpit if random stuff keeps happening. So IMO if you can't point to a specific part, then try memory first. Due to how much work it is with changing MB, that would be one of the very last things I'd do.

actually

[nomadic]

We like to imagine that every Microsoft OS installation will work just as well as the company promises.

Actually around here people like to imagine that every MS OS installation will miserably crash, because then they strut around feeling good about using Linux.

Re:actually

[socceroos]

But for the generalisation, what you say is true. The more I delve into operating systems the less I like any of the current ones out there - Linux included.

Having said all that, when I need to protect highly sensitive information or critical systems then I won't be using Windows. Not because it is specifically 'teh 5U>0|2Z' but because I don't have the fine grained control I need.

Re:actually

[Hatta]

I've installed Windows 7 on my home PC. Played some games on it. I'm impressed. It's at least as stable as XP, and not noticeably slower.

I still strut around feeling good about using Linux. You don't have to hate one to like the other you know. I wouldn't use Windows every day by choice, only because the command line utilities on Linux are so much more convenient. I like the GUI better too, real virtual desktops, windowshading, the selection buffer, all great. And the repositories are great too.

So yeah, not everyone who likes linux is prejudiced against Microsoft.

Re:actually

[Ethanol-fueled]

Sarcastic? Always been the opposite. Many of us stuck with Windows because it just worked. The .iso version of ubuntustudio still fails almost every step of the install, so ya gotta install vanilla ubuntu then apt-get ubuntustudio.

Re:actually

[Anonymous Coward]

That's strange, I just installed Ubuntu Studio 64 bit on my brother's computer a couple of weeks ago and according to him, it's the most flawlessly working version of Linux he's ever used. And this is a guy that is die hard ProTools, Reason, you name it. Now he's having a ball with Ardour, Qtractor, and Hydrogen. And with jack and the real time kernel, when a not is hit on the guitar, no matter how many effects are in the chain, the note is immediately heard in the headphones.

But I guess, YMMV, right?

Re:actually

[Fallingcow]

Do what I do--run Windows, put Linux in a VM. Virtual Box is free, robust, and easy to use, or there's always VMWare.

Run the VM full screen and you can forget you're not running it natively, so long as you don't need to do anything in 3D or very processor intensive (video encoding, for example). Drop to Windows if you need a Windows app (say, a recent version of Photoshop or real MSOffice) or to play games. Plus, if your chosen distro decides to make horrible decisions that cause massive audio breakage (Ubuntu.... *glower*) you can still listen to music or watch Youtube videos in Windows without rebooting.

Another plus is that your Linux installation is all in a single file that you can back up or transfer very easily.

I find that this works far better than dual booting. Saves disk space, saves time. I felt kind of crappy at first for making Linux a second-class citizen on my machine, but this works so much better that I wish I'd done it years ago--though I supposed high clocked multi-core processors and multi-gigabyte RAM sticks weren't commonplace back then, so the experience might not have been so nice.

Re:actually

[Anonymous Coward]

Agreed. Three of my five computers dual boot Linux (Ubuntu) and Win 7. They're not mutually exclusive... people should realize that.

Re:actually

[Jettamann]

If I had mod points.. i would devote them all to you..

I do exactly what you do in ALL my workstations.. from my 8 year old 1.7Ghz single Core Pentium-M with only 2GB Ram (Run windows 732bit Host) and use VMWare Workstation 7 to run latest ubuntu in full screen... its fast.. all the way to my latest 64bit Core-i7 HyperThreaded runing 64bit Windows-7 with 8GB Ram.. with 4GB devoted to my 32bit paravirtualized kernel of Ubuntu 9.10 for development in full screen on a tripple monitor system!

Can't beat it for flexibility and maximum choice of tools for everything.

Its really too bad that Apple never lets OS-X run in VM on non apple hardware.. then I would be in heaven!

Re:actually

[Anonymous Coward]

Thank you so much for posting that in the first story I read after Raising Skinny Elephants...

Re:actually

[Sycraft-fu]

I love that idea and tried to do it at work to help myself learn more Linux, but I just couldn't. Part of the problem was that I'd drop back to Windows when Linux was being a pain and just not go back to Linux since there was nothing Linux did that Windows didn't. The other problem is that a large part of my job involves running various VMs on my system and as anyone can tell you, running multiple VMs in parallel that hit the disk hard on consumer hardware is a world of pain.

You are correct though that on modern hardware, it works great. Part of it is just that modern CPUs are so fast, another part is that VM software has improved a lot. However an even bigger part is that modern hardware has special VM support. If you get a processor with VT-x or AMD-V, it helps. Get one that supports VT-d or AMD-Vi and nested page tables and damn, you are talking near native speeds to within a couple percent in most cases.

A single VM can run in such a way as to fool you in to thinking you are running on native hardware in most situations.

Re:actually

[Hatta]

No thank you. 95% of my time is spent in linux. It runs my RAID array, my torrents, I do a lot of emulation on it. A VM just isn't going to cut it. Maybe once a week or so I'll reboot for some modern 3d gaming, and that really isn't that inconvenient.

Re:actually

[newdsfornerds]

I for one hope your next attempt to install Windows is a complete failure, and that you're up all night pulling your hair out and wondering, "what am I doing wrong?" HEH

Re:actually

[elnyka]

Do what I do--run Windows, put Linux in a VM. Virtual Box is free, robust, and easy to use, or there's always VMWare.

Run the VM full screen and you can forget you're not running it natively, so long as you don't need to do anything in 3D or very processor intensive (video encoding, for example). Drop to Windows if you need a Windows app (say, a recent version of Photoshop or real MSOffice) or to play games. Plus, if your chosen distro decides to make horrible decisions that cause massive audio breakage (Ubuntu.... *glower*) you can still listen to music or watch Youtube videos in Windows without rebooting.

Another plus is that your Linux installation is all in a single file that you can back up or transfer very easily.

I find that this works far better than dual booting. Saves disk space, saves time. I felt kind of crappy at first for making Linux a second-class citizen on my machine, but this works so much better that I wish I'd done it years ago--though I supposed high clocked multi-core processors and multi-gigabyte RAM sticks weren't commonplace back then, so the experience might not have been so nice.

Amen to this. This is specially so true for development environments. You can have multiple Linux images with dev tools specific for multiple projects (or when you actually have to have a specific version of the kernel and crap like that.)

Re:actually

[Anonymous Coward]

I did the opposite. I installed Linux as the base and run Windows (XP) in VirtualBox for those times I need to work with MSOffice documents. (Most of the time I can get away with using OpenOffice in Linux but not always.) It really depends on your goals. I have found myself getting more and more command line oriented to the point where I get impatient with Windows/OSX. YMMV.

Re:actually

[Gr8Apes]

Its really too bad that Apple never lets OS-X run in VM on non apple hardware.. then I would be in heaven!

Well, officially Apple doesn't let OSX run on non apple hardware. In reality start here and have fun.

I'm a bit of the opposite to you - I run OSX as my main OS, and run Server 2008 R2, Win 7, and Win XP in VMs (Parallels 5, VMWare 2.0 was still doing software CPUs - don't know about 3.x) and it was certainly freeing - every time windows craps out - snapshot restore and I'm back up in about 5 minutes. I've also got Linux VMs, but generally don't need to run them for testing as often as the coding I do generally transfers fluidly across and all the items I use on Linux also work on OSX.

Re:actually

[StuartHankins]

Part of the problem was that I'd drop back to Windows when Linux was being a pain and just not go back to Linux since there was nothing Linux did that Windows didn't.

If you haven't found anything that Linux can do that Windows can't, you're not trying very hard.

• Take a look at RHEL virtualization, where memory from multiple VM's can be reduced in many cases. You won't see that in Windows.
• While speaking of virutalization, you know that enterprise-grade tools from VMWare and Red Hat can use tiny Linux hypervisors right?
• With ksplice, you can run and upgrade your machines in a responsible manner without reboots.
• Take advantage of all CPU's and all cores by using tools such as taskset to move processes around in realtime
• Easily redirect a failed printer's jobs to another printer (UNIX is supreme for printer management IMHO)
• Use post-processing to modify print jobs, for instance to print to PDF and dump those PDF files onto another fileserver in the proper user's folder
• Execute a command on a list of computers or across a domain (without the restrictions of runas or psexec)
• Connect to your Cisco VPN in under 2 seconds every time vs the 15-to-20-second wait using the Windows driver
• Image your drive, partitions etc without 3rd party utilities
• Run a fully open-sourced system, with you in total control, without worry of licensing issues

Re:actually

[Gr8Apes]

...and just not go back to Linux since there was nothing Linux did that Windows didn't.

You either have a MS world view, a peculiar use case, or just don't know what you're talking about.

Try running LDAP on Windows. Can't do that.
Try running a 0 priv service that utilizes a users credentials and changes a different user's password. Can't do that without some serious gymnastics that involves using a SYSTEM or equivalent process.)
Try executing a script utilizing pipes out of a daemon process (closest thing is a service in windows land) You can't.

That's just the beginning of a long long long list of things windows can't do. The list of things windows can do? It's rather short:
Allow a virus/worm in under a service and wreak total havoc because there is no real concept of a lowest privilege process. Oh, and open windows office documents. Can't forget that important one.

You're Kidding

[nmb3000]

This is front-page news for Slashdot now? Here's the sum total of TFA:

• Guy tries to install 64-bit Windows 7 on a machine previously running 32-bit Windows 7
• Install fails over and over again
• He replaces hardware components with no luck until he swaps out the CPU
• Windows installs but is unstable
• Worthless ASUS BIOS automatic "optimizers" cause stability problems (surprise!)
• With BIOS settings changed to sane values Windows is stable

Wow, color me impressed!

How are "mortals" supposed to figure it out? I guess they buy a PC from Dell because everything in that article qualifies as "no duh" for system builders.

Re:You're Kidding

[wes mantooth]

Also, let's not forget that the CPU he used was an engineering sample: "In fact, of the 30-plus Intel processors I’ve installed Windows 7 on, this one is not only the single solitary item I’ve had any problems with at all, it’s also the only freebie engineering sample I used."

Re:You're Kidding

[Anonymous Coward]

This is the reason I will not buy a "Enthusiast" or a "Entry-Level" piece of hardware. With the former it's always the overpriced overclocking features that are wonky, and with the latter, it's the cheap hardware. Second-generation midrange stuff (with FreeBSD-compatible hardware (i.e. open source drivers available), even if it will only run Windows) has always been the best.

Re:You're Kidding

[lalena]

Exactly. He swapped every piece of hardware - saving the engineering sample CPU as the last thing he swapped. The system ran fine under Win 7 32 bit. You have to assume that hardware still works fine and that the problem was 64bit specific - which points to the CPU. Granted Intel said it should support 64bit, but it was an engineering sample.
He replaced the case, power supply, the video card, the mother board, the hard drives, and the cables first??

Re:You're Kidding

[GrumpySteen]

You should be impressed. No mere mortal would ever look at computer and think "let's replace random parts until it starts working!" This guy is clearly some sort of magical god of electronic troubleshooting. Quite possibly with a unicorn for a sidekick.

Re:You're Kidding

[Anonymous Coward]

More evidence he's a "script-kiddy": He uses Microsoft's "excellent" Windows 7 USB DVD Download Tool, instead of simply using diskpart to create a partition on the stick and copying the files over from the ISO. (In case there are more people reading this who don't know how to make a bootable USB flash drive with stuff that's already in Windows: diskpart is the command line disk partitioning tool in more recent versions of Windows. When it creates a partition on a clean drive/stick, it automatically writes a boot sector that will load "bootmgr", which will then proceed to boot from the device. That's all there is to it.)

Re:You're Kidding

[Anonymous Coward]

How are "mortals" supposed to figure it out? I guess they buy a PC from Dell

I see what you did there.

Re:You're Kidding

[bloodhawk]

The truly sad part is for a supposed techie his troubleshooting skills sucked balls. If the device worked fine with 32 bit and failed with 64 bit, it isn't gonna be the god damn cables or power supply that needs replacing.

Re:You're Kidding

[Gadget_Guy]

More evidence he's a "script-kiddy": He uses Microsoft's "excellent" Windows 7 USB DVD Download Tool, instead of simply using diskpart to create a partition on the stick and copying the files over from the ISO.

Yeah, right. He writes books on Windows 7, but he shouldn't try the official way of installing from USB. Because that would mean that he had used the tools that he wanted to write about. Shame on him!

Re:You're Kidding

[Anonymous Coward]

windows being fussy? dont start at basic defaults and work your way up, toss 4 computers worth of new hardware at it!

this guy is a nugget, and toms hardware is quickly following

Re:You're Kidding

[GrumpySteen]

Maybe if he had switched to a 128 bit power supply. That's twice what you need for a 64 bit processor, right?

Re:You're Kidding

[drfreak]

You forgot: Profit!

I can totally understand why the CPU was last on his checklist. In times past (pre-2003ish), CPU type and frequency were set by jumpers or dip switches on the motherboard; similar the the ISA days where we had to do the same for add-on cards. Most (if not all, I don't build computers anymore) motherboards after then gave an "Auto" option and people stopped paying attention to the CPU except to make sure the motherboard supported it before installing.

Now that we are in the 2010+ era, I'd imagine most motherboards can figure out what CPU you put in. My guess is that in this fringe case, it could not.

Re:You're Kidding

[drfreak]

WTF do the power supply and cables have to do with 32 vs. 64-bit? I call shenanigans!

Re:You're Kidding

[Anonymous Coward]

If your PSU isn't supplying enough current to the board, all kinds of wonderful crap can happen. I've also had cables give random problems too (though it shouldn't have worked with 32bit then in most cases)

Re:You're Kidding

[paulej72]

My first thought is why did he not try using a DVD to install as that is the supported method for the install. I would have thought that for anything other than a netbook this would be the next step, not replacing the motherboard.

Re:You're Kidding

[DigiShaman]

I could be wrong, but all production CPUs will load updated microcode at POST (stored in BIOS ROM) if available. In short, microcode is like a patch to correct CPU errata. Engineering samples are like beta CPUs though, right? It would stand to reason that no such microcode is available for them from a consumer (non testing board) Motherboard.

Re:You're Kidding

[rubycodez]

I always use gold plugged audiophile cables from my power supplies, they supply robust pure energy for perfect rendering of 64 bit flash multimedia

Re:You're Kidding

[fatp]

When the book is out, someone should write a review on Amazon to tell potential readers the author's capability of running windows on properly.

Re:You're Kidding

[Gadget_Guy]

When the book is out, someone should write a review on Amazon to tell potential readers the author's capability of running windows on properly.

So after 400 (presumably) successful installations he has a hardware failure that causes problems. Exactly what would you say that he couldn't do in your review on Amazon? That he couldn't get Windows installed? Surely not, because he did end up getting it installed after replacing the faulty part.

Sure, he replaced all the other parts of the system before he replaced the CPU, but he already had those other parts on hand. He had to end up buying the new CPU. Surely you would try the stuff that you could do for free before shelling out money on new equipment.

Re:You're Kidding

[Gadget_Guy]

Sure, he replaced all the other parts of the system before he replaced the CPU, but he already had those other parts on hand

Correction to myself. He did buy another motherboard because he had already had problems with it previously. If it had been problematic before then it seems reasonable to think that it may have been the causing this issue on the Win7 install.

Re:You're Kidding

[Anonymous Coward]

On the other hand he had a great excuse to buy himself enough parts for another computer.

"Honey, I have to buy these new parts or all the old parts will be wasted money."

Re:You're Kidding

[Barny]

Wow, he has installed 30-plus intel based Windows Seven machines? Since its launch, I have had about 40-50 DOA intel processors, NONE of which were engineering samples.

Guy is what we, in Australia, would call a "Tosser" who for lack of a better description is "Talking wank".

No, mere mortals would never be required to sort out this problem, because they would never encounter it.

Order for troubleshooting random seeming install fails is:

Install media (the disk, get a known good image/disk)
Hardware that reads that media (DVD drive, cable)
Ram
CPU
HDD (and cable)
Board
PSU

Most DOA PSU problems are dead PSU, most of the rest are rating selection errors (not powerful enough).

Also, this list is optimised not only for most likely faults but for parts that are "easy" to replace :)

Re:You're Kidding

[deniable]

Just make sure you have the directional indicators pointed the right way.

Re:You're Kidding

[IICV]

Yeah, this makes no sense to me. He basically replaced everything but the bleeding edge engineering sample? If it was my computer, that's the third thing I'd swap out once it gets to the point of potential hardware failure (ram is really easy to replace so why not, and hard drives are only marginally harder as long as you've already got the case open)

Re:You're Kidding

[Anonymous Coward]

this forum is not for liberal arts majors

Re:You're Kidding

[Hamsterdan]

Nah, he should have used a gold plated power cable. Better current quality

Re:You're Kidding

[TeknoHog]

Is that the direction of current or electron flow?

Re:You're Kidding

[rubycodez]

the important thing is to keep them straight so the electricity doesn't skid around curves

Re:You're Kidding

[Anonymous Coward]

yes.

Re:You're Kidding

[cool_arrow]

Do you really believe that? I've often wondered who buys that stuff and why.

What I love here is the part where he

[jra]

just rolls right on past the fact that, if what he was installing was -- oh, say -- a Linux distribution, he wouldn't have an opaque "I'm uncompressing files" thermometer, he'd have real progress status messages, with, y'know, *parameters* and stuff, and -- unlike me this morning with my boss's iPhone -- a hope of actually figuring out what's broken.

But he's apparently completely blind to the fact that that's the *real* problem here.

"We'll just make fault-tolerant users", indeed

Re:What I love here is the part where he

[Anonymous Coward]

And which consumers actually install an OS on their machine? Most probably not.

Re:What I love here is the part where he

[Wrath0fb0b]

just rolls right on past the fact that, if what he was installing was -- oh, say -- a Linux distribution, he wouldn't have an opaque "I'm uncompressing files" thermometer, he'd have real progress status messages, with, y'know, *parameters* and stuff, and -- unlike me this morning with my boss's iPhone -- a hope of actually figuring out what's broken.

And what specific parameter in any Linux installation error message is likely to point towards the CPU being defective? Most of them would be generic hardware-has-shit-itself errors (DMA failures, null pointer exceptions, hash failures) that could mean any of the cpu/motherboard/ram/psu/hdd are defective. It's impossible, even in principle, for any installer to be able to pinpoint with specificity what hardware is fucked.

Just for lols, I wish you would get modded up (me too, of course :-P) so that the OP can install $DISTRO on that original setup and see what error we get and whether it exactly pinpoints the cpu or whether it spits out a generic hardware error.

Re:What I love here is the part where he

[jra]

Even more reason why they should provide useful status messages: the odds are *much* higher than might be intuitively obvious that the person doing the install can make use of them.

I'm likewise unimpressed with the *order* in which he swapped components; you generally do it in descending order of "likely to be causing this problem".

Shame his installer didn't have a copy of Memtest+ on it, too, ain't it?

Re:What I love here is the part where he

[Kitkoan]

And what specific parameter in any Linux installation error message is likely to point towards the CPU being defective? Most of them would be generic hardware-has-shit-itself errors (DMA failures, null pointer exceptions, hash failures) that could mean any of the cpu/motherboard/ram/psu/hdd are defective.

That would be the P.O.S.T. which your BIOS should be checking.

Re:What I love here is the part where he

[yossarianuk]

Actually its very easy to identify if you have a HDD issue from running a Linux live CD. The other components you mention yes would be hard to diagnose, although we have identified memory bank issues also.

Re:What I love here is the part where he

[Jezza]

I temper this approach with the "this is easy as hell and very quick". So even if I think it is something, if there is something else it could be that's really quick to try I'll ignore my "brilliance" and try that. What is amazing is how often there isn't actually one problem, but two. Also helps if you have a similar working system that you can take the components from (so you know that this or that doodah actually works).

Re:What I love here is the part where he

[TheRealSlimShady]

SHIFT+F10 will give you access to a command prompt during setup, which you can then use to vew status messages and parameters and stuff...

Re:What I love here is the part where he

[cenc]

At least it would tell you everything else is fine, and using proper process of elimination work your way back to 'dam must be the processor'. That is assuming you ignore the error messages related to the processor in Linux.

Re:What I love here is the part where he

[Anonymous Coward]

yeah.
it boggles my mind that windows has no built in utilities for querying hardware sensors or SMART data.
in linux that stuff has saved my neck more than once.

its sad when I have to use a flash drive with linux on it to figure out why a machine is croaking when the already installed operating system can't check without me downloading dozens of megs of vendor specific crap.

facepalm ms, facepalm

Re:What I love here is the part where he

[Wrath0fb0b]

it boggles my mind that windows has no built in utilities for querying hardware sensors or SMART data.

Yup, it boggles my mind too. Wait, what?

smartctl 5.39 2009-12-09 r2995 [i686-pc-cygwin-win7(64)] (cygwin-5.39-1)

Re:What I love here is the part where he

[Barny]

/agree

One of the better tests for if a machine has a hardware fault or a software fault is to boot off a linux CD, because for lack of any major hardware faults, it just works :)

Some companies do have specific hardware testing tools, seagate disk tools for instance will tell you before testing a drive if smart has been tripped, for that matter, most motherboards will pause at boot time if any smart data has been tripped on a HDD.

Re:What I love here is the part where he

[Anonymous Coward]

wouldn't those errors get reported during the POST, and thus be completely OS independent?

While the article was lame, I'm assuming the guy would've mentioned if it said CPU_ERROR every time he started the thing.

Re:What I love here is the part where he

[Sycraft-fu]

That is a very basic check. You can't thoroughly check all of a system's features in that amount of time. Heck, a proper memory test that tests all different patterns of bitwise operations (as Memtestx86+ does) can take many hours on a 4GB system. Likewise it's CPU test is limited by many factors, not the least of which it isn't running in long mode. The OS switches the CPU to long mode when it starts, the BIOS boots up in real mode, for compatibility reasons.

A POST can check for extremely critical issues, nothing more. Passing a POST does not mean hardware is fine, it just means that it is fine enough to start the system.

Re:What I love here is the part where he

[GuyFawkes]

Mod parent up.

If you only found one problem, it is because you stopped looking, it is very rare to find only one problem with hardware, as in only one item may actually be defective, but at least one other item is getting close to the limits of spec.

Re:What I love here is the part where he

[Anonymous Coward]

What part of 'built-in' did you miss? And cygwin? Seriously?

Re:What I love here is the part where he

[Life2Death]

When i was younger i got my dads pentium 90 and decided that i'd put it back together (it had been parted out into a few P150's and a pentium pro system.

I found:
A bad drive, bad drive cable, bad CMOS battery, bios had bad settings, a dead fan causing overheating in some parts. It took forever to find that the cable was bad on top of the drive, but after swapping in known working parts its not that hard.

Re:What I love here is the part where he

[Gr8Apes]

And what specific parameter in any Linux installation error message is likely to point towards the CPU being defective? Most of them would be generic hardware-has-shit-itself errors (DMA failures, null pointer exceptions, hash failures) that could mean any of the cpu/motherboard/ram/psu/hdd are defective. It's impossible, even in principle, for any installer to be able to pinpoint with specificity what hardware is fucked.

Well, having experienced this particular fun with a recent CentOS distribution attempting to be installed on an AMD 939 X2 CPU, the boot and install went great. Reboot into your newly installed wonderful OS? Seems ok. Login. Success! Run a couple of commands and watch the thing pull a Microsoft by falling over dead with virtually no output except for an esoteric error and memory location which was enough to pinpoint the problem. Seems our CPU release had a small piece of errata in it that prevented a certain 64-bit optimization that the CentOS kernel tried to use. Whoops.

Not the OS

[NoCleverName]

Sounds like this system would have crashed unzipping ANY large file under ANY OS because it was the application that the CPU couldn't properly execute.

100 books?

[cranesan]

I'm a little suspicious; how much of an expert can you be writing 100 books on a variety of subjects.

Reminds me of a tech instructor I had who proudly informed the class he teaches oracle classes, mysql classes, sql server classes, cisco classes, juniper classes, .net development classes, php, etc..... Yeah he couldn't answer any basic questions that strayed from the text book in front of us.

Re:100 books?

[NitroWolf]

I'm a little suspicious; how much of an expert can you be writing 100 books on a variety of subjects.

Reminds me of a tech instructor I had who proudly informed the class he teaches oracle classes, mysql classes, sql server classes, cisco classes, juniper classes, .net development classes, php, etc..... Yeah he couldn't answer any basic questions that strayed from the text book in front of us.

Haha no kidding. I had an instructor once that taught the A+ certification class as well as bunch of other computer classes, but when I asked a question about Big Endian vs Little Endian in regards to one of the test questions on significant bits, he had no idea what I was talking about and had never even heard of Endian-ness. It was at that point I discounted just about every authoritative thing he tried to say in class.

Re:100 books?

[LordVader717]

Different tasks, different skills. If you can write good guides and enjoy doing it, why should you want any more? Having in-depth knowledge doesn't make you a good teacher.

Re:100 books?

[Anonymous Coward]

Haha no kidding. I had an instructor once that taught the A+ certification class as well as bunch of other computer classes, but when I asked a question about Big Endian vs Little Endian in regards to one of the test questions on significant bits, he had no idea what I was talking about and had never even heard of Endian-ness. It was at that point I discounted just about every authoritative thing he tried to say in class.

Mighty nice of you if you remained for the rest of the class/semester. That'd be the point where I'd laugh, inform him I would be dropping his class, say something dismissive and insulting, and walk out. I actually did that in a programming class there the teacher totally flipped out when I had to hard reset a computer that had an ejected floppy disk sticking out of the drive. I have no idea what sort of bad thing he thought was going to occur as a result of that ejected floppy that wasn't completely removed from the drive but berating me in front of the class was uncalled for.

Re:100 books?

[Anonymous Coward]

You may be right on that. Yet, lacking in-depth knowledge of the subject you are teaching is a pretty good guarantee you will not be a good teacher.

Re:100 books?

[the_other_chewey]

Having in-depth knowledge doesn't make you a good teacher.

Completely true. However, a good teacher has in-depth knowledge of what he teaches.

Re:100 books?

[neumayr]

You think so?

A comparison between university professors and school teachers comes to mind. The former surely has a lot more knowledge on his subject than the latter, he has a PhD and everything. But the school teacher has a lot more training in how to convey that information efficiently, how to best reach the people he teaches.
In addition to being an example for the GP's argument, to which you agreed, this is also a case where broader, albeit less in-depth knowledge makes the school teacher a better educator.

No need for the lesson

[Anonymous Coward]

We know perfectly well that when we boot from a Ubuntu CD that we will end up with Ubuntu and don't think a second of why Windows 7 didn't install.

Simpler Solution

[Anonymous Coward]

If crapware won't install on your computer, don't install the crapware. I'm sure there's plenty of other choices out there.

Re:Simpler Solution

[keeboo]

I'm all for FOSS, Linux etc.
But this approach of yours won't convince any Windows user to switch. Instead, it's likely more people will get convinced that FOSS users are assholes.

Re:Simpler Solution

[Anonymous Coward]

Fuck off little windows bitch.

Re:Simpler Solution

[davidla]

And all of those that you mention are just as crappy if the CPU is malfunctioning.

Please don't turn topics into 'Windows vs Linux' when it is obviously not the issue.

ES CPU

[tftp]

If this hardware geek spends days solving a CPU-meets-Windows 7 problem, what chance do mere mortals have?"

You need first to show me a "mere mortal" who has, and uses, an engineering sample CPU. There is a very good reason why -ES parts are marked as such - because they have bugs. And those bugs will be a problem sooner or later.

So the whole sob story can be reduced to this. The guy runs software on a prototype hardware, and the software crashes. In other breaking news, dog bites man.

Re:ES CPU

[Mashiki]

Well I thought it was CPU bites man. Everyone knows that unless the CPU Blood God gets his fill when you're doing an upgrade, you're going to have problems.

Hardware issue with unoffical hardware.

[bv728]

Also, he was using an unsupported engineering sample CPU which was, in fact, the culprit for his issues. In other words, he ran into a Hardware Issue NO end user will, or most power users.

Top 7 problems with Windows 7

[tomhudson]

We like to imagine that every Microsoft OS installation will work just as well as the company promises.

Actually around here people like to imagine that every MS OS installation will miserably crash, because then they strut around feeling good about using Linux.

[X] Psst ... it's not your imagination, honey :-)
[X] I use BSD, you insensitive clod!
[X] In Soviet Russia, Windows crashes YOU! ... oh, wait a sec ...
[X] CowboyNeal is my runtime environment (Ewww!)
[X] ... but at least it blends ...
[X] If someone says "There's an app for that" one more time I'll throw a chair at them!
[X] Steve Ballmer posts on slashdot!!!

Re:Top 7 problems with Windows 7

[Anonymous Coward]

[X] CowboyNeal is my runtime environment (Ewww!)

From the pictures we've seen in the past, it seems unlikely that CowboyNeil does any running.

Re:Top 7 problems with Windows 7

[caspper69]

He was my professor a couple of semesters ago. I can vouch for that!

Hell of a nice guy, and pretty talented to boot though.

Re:Top 7 problems with Windows 7

[techno-vampire]

[X] If someone says "There's an app for that" one more time I'll throw a chair at them!

Is there an app now that throws chairs for you?

Re:Top 7 problems with Windows 7

[Abstrackt]

[X] If someone says "There's an app for that" one more time I'll throw a chair at them!

Is there an app now that throws chairs for you?

With a little ingenuity, there is!

Harware issue? Welcome to Linux

[ben_kelley]

If you have never had a hardware issue when installing Linux on a machine you must be very lucky.

"Most things work fine" people tell me, which is true. The trouble is that the chances of you owning something that doesn't work is relatively high. (There's probably something from my statistics course that explains why that is, but I have so far managed to suppress that memory.)

After having rebuilt a Mac with OS X, and rebuilt a laptop with Ubuntu 9.04, I was surprised at how smooth and the Ubuntu install was. Of course that was until I wanted to use my webcam with Ubuntu. These kinds of problems get very difficult very fast in Linux. When 9.04 first came out there was a dependency problem that meant that you couldn't easily get some webcams working.

To be fair, that problem is most likely sorted out now, and a non-Apple webcam would have needed a (very easy to install) driver on OS X as well. The point is, Windows and hardware generally work very well.

Re:Harware issue? Welcome to Linux

[jedidiah]

There was another fellow that mentioned the idea of staying away from the top and the bottom.

Avoid the dregs and the bleeding edge.

That middle will probably me much more reliable under Windows and more likely to be supported on Linux (or even MacOS).

No one cares enough about the dregs to support them under Linux or MacOS and the bleeding edge stuff is just too new.

That approach does pretty well regardless of OS today and did pretty well 16 years ago too.

The problem with "statistics" is that any give PC isn't really random. It's a reflection of it's owner. It may be a dreg, a poster boy for bleeding edge gamer conspicuous consumption or something that's more moderate.

"When 9.04 first came out" is covered by this rule actually.

Re:Harware issue? Welcome to Linux

[Anonymous Coward]

TFA's about defective engineering sample hardware and windows, whose arse did you pull linux hardware issues out of?
just a case of pre-release hardware bites man, slow news day i guess.

Re:Harware issue? Welcome to Linux

[Kitkoan]

If you have never had a hardware issue when installing Linux on a machine you must be very lucky. "Most things work fine" people tell me, which is true. The trouble is that the chances of you owning something that doesn't work is relatively high. (There's probably something from my statistics course that explains why that is, but I have so far managed to suppress that memory.) After having rebuilt a Mac with OS X, and rebuilt a laptop with Ubuntu 9.04, I was surprised at how smooth and the Ubuntu install was. Of course that was until I wanted to use my webcam with Ubuntu. These kinds of problems get very difficult very fast in Linux. When 9.04 first came out there was a dependency problem that meant that you couldn't easily get some webcams working. To be fair, that problem is most likely sorted out now, and a non-Apple webcam would have needed a (very easy to install) driver on OS X as well. The point is, Windows and hardware generally work very well.

Linux works very well with PC hardware too. Apple's hardware isn't an exact copy of normal PC hardware if I recall correctly (thats why you need to buy Mac versions of things like graphic cads). This means that you will need slightly different drivers (and your comment of a non-apple webcam tells me its a MacBook you installed Linux on). And while many (though not all) pc hardware vendors release Linux drivers, Apple on that other hand doesn't which makes it that much more harder.

Re:Harware issue? Welcome to Linux

[Anonymous Coward]

If you have never had a hardware issue when installing Linux on a machine you must be very lucky.

And yet, I never win the lottery.

Re:Harware issue? Welcome to Linux

[poopdeville]

(There's probably something from my statistics course that explains why that is, but I have so far managed to suppress that memory.)

Birthday paradox. More of a combinatorics issue than statistics though.

Re:Harware issue? Welcome to Linux

[Angst Badger]

"Most things work fine" people tell me, which is true. The trouble is that the chances of you owning something that doesn't work is relatively high. (There's probably something from my statistics course that explains why that is, but I have so far managed to suppress that memory.)

It's simple probability. If (let's say) 95% of all current devices are supported under Linux, that means there's a 5% chance -- 1 in 20 -- that any randomly selected device won't work under Linux. Now, how many devices -- i.e., components requiring drivers -- are there in or attached to your computer? Odds are that it's at least twenty, and on many plain vanilla boxes, probably more. Linux works amazingly well considering the odds. Most of the problems I've had have been with graphics cards whose manufacturers are disinclined to care about their non-Windows-using customers, which is a far cry from the situation in the late 90's when hardware support for Linux was much worse.

That said, it's been a very long time since I had any hardware problems with Windows other than getting post-XP hardware to work with XP, as all or nearly all hardware manufacturers have a very strong incentive to make sure their hardware plays nice with the overwhelming majority of machines. But getting back to the original post, if there was a situation where I expected problems with hardware under Windows (or anything else), it would be with unsupported engineering samples. And in such a case, I wouldn't shout and wave my arms around and generalize from the worst-case scenario that the sky was falling.

Re:Harware issue? Welcome to Linux

[techno-vampire]

Of course that was until I wanted to use my webcam with Ubuntu.

I recently bought a new laptop and installed Fedora 12 on it. I was quite pleased to see that Anaconda had detected my webcam and installed cheese, and the first time I ran it, It Just Worked. Considering that Fedora's a bleeding-edge test-bed for RedHat, I'd expect Ubuntu to have the procedure down pat. Did you add the webcam after installation?

Simple Answer really.

[MrShaggy]

"If this hardware geek spends days solving a CPU-meets-Windows 7 problem, what chance do mere mortals have?"

Simple really. WE ARE SPARTANS!

400 installs w/o a problem

[Bryan-10021]

I probably installed Windows 7 over 200 times [..]. Since then, I’ve installed the OS at least another 200 times [...]Until a couple of weeks ago, I never encountered a single problem that stopped me from installing Windows 7 itself.

400 installs w/o a problem and then a problem that has nothing to do with Windows 7. Now read the summary of the article, 'If this hardware geek spends days solving a CPU-meets-Windows 7 problem, what chance do mere mortals have?'

Really? That summary blurb is what you got from the article.

Gotta love /. people.

Re:400 installs w/o a problem

[Barny]

Despite my contrary experience with this unsupported and unwarranted engineering sample CPU from Intel, I’d agree with that remark. In fact, of the 30-plus Intel processors I’ve installed Windows 7 on, this one is not only the single solitary item I’ve had any problems with at all, it’s also the only freebie engineering sample I used.

It appears, he either uses a LOT of AMD or VIA chips, or he really really likes reinstalling windows over and over again on the same hardware ;)

'Mere Mortals' use OEM

[Kitkoan]

And those work because they have the drivers built in without the need to swap hardware or anything.

Summary.

[Anonymous Coward]

Man installs Windows 400 times on 30 different processors without a hitch. On the 401st time, it fails when trying to install on a machine with an old engineering sample CPU. What on Earth is a mortal to do?

Re:Summary.

[socceroos]

I think it was the 404th time actually.

Summary: successful install on new system

[Anonymous Coward]

So, after swapping out every component of his PC for a new one, the installation was successful. Did he not notice that it was no longer his old PC? That it was a now new PC in the old case?

This is not unlike the story of the guy with an axe that he claimed used to belong to Abraham Lincoln. The handle had been replaced three times and the head replaced twice, but it was still Abe's old axe.

Hardly worthy of slashdot

[zorog]

Ultimately this is in Microsofts best interestes to ensure people can install Windows SHINYNEW(tm). There scattered offerings are the cause of this problem, they should offer one OS install disk with maybe a few kernels. Disclaimer I've never installed windows 7, its on mums cheap laptop and works fine for her much improved on previous offerings and for once MS makes XP look like the dated and cumbersome giant it is.

Re:Hardly worthy of slashdot

[Anonymous Coward]

WTF?? This has nothing to do with MS's scattered offerings, This article is about faulty Engineering sample hardware that was not performing to spec and hence causing the OS to crash and be generally unstable. Whether MS had one 64 bit version or 343535345 versions is completely irrelevant.

Assigning blame doesn't alway help

[iYk6]

I didn't read the article, but judging by the summary, I think it is more about troubleshooting than assigning blame.

Re:Assigning blame doesn't alway help

[PopeRatzo]

I didn't read the article or the summary. The title was plenty.

Re:Assigning blame doesn't alway help

[PNutts]

I didn't read the article or the summary. The title was plenty.

I didn't read the article, the summary, or the title and you're both wrong.

Re:Assigning blame doesn't alway help

[EricX2]

I haven't read the title, summary, article or any posts, and I think it's time to get new glasses... I can't read a damn thing!

Re:Assigning blame doesn't alway help

[Runaway1956]

"I can't read a damn thing!"

Coming soon, to a galaxy near you, will be a COMMUNITY COLLEGE, complete with a REMEDIAL READING class! Enroll early, and avoid the rush!

Re:Assigning blame doesn't alway help

[PNutts]

Enroll early, and avoid the rush!

You insensitive clod! http://yro.slashdot.org/story/10/04/12/225248/Red-Light-Camera-Ticket-Revenue-and-Short-Yellows?art_pos=3

bass-ackwards

[spywhere]

He had a running PC, but he couldn't figure out how to install a different OS on it (using the barely-supported bootable flash drive method)... so he threw parts at it until it worked, essentially installing Windows on a completely different PC than the one he started with.
I fix computers for a living, and I often reinstall Windows for my customers... the difference is that if I fail, I don't get paid.

He should have tried installing Windows from the DVD, or from the hard drive (using WIndows PE to kick off the installation).
I am unimpressed. I award him no points, and may Dog have mercy on his credit card bills.

Re:bass-ackwards

[Aranykai]

The flash drive method is very much supported and works great in my experience. I do about 10 installs a week and I use a USB flash drive for all of them. I have yet to run into a single instance where installing from flash hasn't worked. Its not only faster, its a heck of a lot quieter too.

When I return the machine to the customer their Windows disc is all nice and intact and I never had to touch it.

This happened to me in a production CPU.

[UnifiedTechs]

Actually I had the exact same problem a few months ago upgrading a Dell server from Win2003 x86 to Win2008 x64, I suspected the CPU from the beginning, but I spent a few hours before the Dell Tech agreed with me. They sent a replacement and it worked like a champ.

This proves it has happened to a production Intel Core2Duo CPU at least once, I can't believe I was the only one.

No kidding

[Sycraft-fu]

Dear self important guy who isn't near as good at computers as he thinks he is:

This may surprise you to learn, but all those defaults out these, all those specified values, all that kind of stuff, that isn't just arbitrary. See many smart engineers and other folks worked on designing and creating all the hardware for your computer. A lot of extremely complex stuff went in to it, modern computers are quite a marvel of engineering. As such, they discovered that certain tolerances, certain ranges work well. Outside of that, there can be problems. Thus the defaults because, well, default. They set them so that things are very likely to work in all cases.

As with most things, they aren't absolutes. They aren't things you can never exceed. In various circumstances you can go outside those normal ranges, sometimes by a little, sometimes by a lot. However, problems can potentially result. What problems those are and when they happen is not predictable. A system can appear stable but only crash on one app, or it can be stable for awhile then develop an instability.

Regardless, the first step to troubleshooting should be to USE THE FUCKING DEFAULTS, you idiot!

Seriously, I'm supposed to take someone seriously who is running overclocked settings of some sort or another (RAM timings, FSB, etc) and an engineering sample CPU and has problems? Ummm, duh. That right there is asking for problems. When you OC, you go in to it knowing you may have some difficulties. You understand this is the tradeoff for something that runs faster than spec. If you start having problems, the first step is to back off the OCing and see if that fixes it.

This is true even of OC'd systems that were fine but aren't now. I had a Celeron 300 that I OC'd to 450 back in the day and it worked well for about a year, then started to burn out. System started crashing randomly, and so on.

To me, it sounds like he's being whiny because he didn't bother to troubleshoot his setup properly. Come talk to me when you've got a retail CPU running at stock spec and FSB, RAM running per it's JEDEC spec at standard voltage and so on. Oh, what's that? You did that and it stopped having problems? Well there you go then. Don't bitch that your i7 920 "should" run at 3.8GHz. I don't care if others have done it, doesn't mean it'll work in your case. If it does, wonderful. It if doesn't well tough shit. Don't get mad at the software. It has pretty much no way to know if the CPU is going crazy as it runs on the CPU. About the only way software can indicate a CPU problem is by inducing a problem and thus a crash.

Re:No kidding

[Gadget_Guy]

Regardless, the first step to troubleshooting should be to USE THE FUCKING DEFAULTS, you idiot!

The only problem with your rant is that the Asus AI Tweaker is turned on by "fucking" default (or at least set to "auto"). It is how Asus tries to score higher in benchmark tests. As he found out, these auto settings can get confused and push the speeds too high.

I don't bother with any overclocking these days, so I alway turn off these so-called intelligent settings. The slight improvement in speed can be completely offset by random crashes. It has been a long time since I had a computer that wasn't more than fast enough for my needs. (Mind you, I don't play Crysis.) But I can understand how someone might think that the default setting would be the conservative one.

Re:No kidding

[Colonel Korn]

Regardless, the first step to troubleshooting should be to USE THE FUCKING DEFAULTS, you idiot!

The only problem with your rant is that the Asus AI Tweaker is turned on by "fucking" default (or at least set to "auto"). It is how Asus tries to score higher in benchmark tests. As he found out, these auto settings can get confused and push the speeds too high.

I don't bother with any overclocking these days, so I alway turn off these so-called intelligent settings. The slight improvement in speed can be completely offset by random crashes. It has been a long time since I had a computer that wasn't more than fast enough for my needs. (Mind you, I don't play Crysis.) But I can understand how someone might think that the default setting would be the conservative one.

The author of TFA claimed it's on by default, but it wasn't on when I received my Asus P5E3 Pro, which is the same motherboard he used. I suspect he accidentally turned it on without realizing what it was while mucking around in BIOS before the OS install.

Re:No kidding

[Anonymous Coward]

Why is this modded up? What's sad is that you read the article to see that it was overclocked, but couldn't comprehend what you read because the article states that those settings were the default values.

Quite useful infact

[Anonymous Coward]

I found this very useful. I have been having the exact same problem, Asus Mobo w/ AI Tweaker enabled. Swapped out some parts, came to the conclusion it must be the mobo and decided to install ubuntu instead.

In all fairness

[tpstigers]

Not all Windows 7 installs are painless. When I installed mine, I found out the hard way just how dependent a Win7 installation is on a network connection. My network connection wasn't supported so the install failed miserably (seriously - I haven't seen a desktop like that since Windows 95). I eventually worked my way around it by installing a wireless card and connecting through it. Great fun.

Re:In all fairness

[Anonymous Coward]

windows 7 is not dependent on a network connection at all. I do standalone installs all the time.

Re:In all fairness

[harryjohnston]

Windows 7 will install perfectly well without a network connection. (Just to make sure, I postponed sending this until my test VM, sans network interface, had completed installing.)

My guess would be either you were using badly OEM'd install media (a practice I do wish MS would prohibit) or you don't know how to manually install device drivers.

Re:In all fairness

[Barny]

Anonymous Coward iz credit to team!

You do not require a network connection at all to install, it is a good idea, because windows will get any installer updates on the fly, but certainly not required.

In fact, you can turn off the minimal need it has for one via a setting in the setup XML files :)

Re:In all fairness

[mlts]

That is odd. I install my Windows 7 boxes not connected to the network in any way until I make sure the firewall has the "Block all incoming connections" checkbox on. I've never had a problem with it having to get to the network during the install process.

Re:In all fairness

[tpstigers]

Actually, in the end I blame HP (the manufacturer of the box in question). Very few of the devices in the box were supported out-of-the-box. When I first tried the install, I ended up with bad video, no sound and no peripherals. The later install (with the wireless card) installed beautifully, I assume because it had access to Windows Update.

Re:In all fairness

[harryjohnston]

You could probably have obtained the drivers from HP's website, even if they didn't provide a driver CD. It's not exactly rocket science.

Re:In all fairness

[tpstigers]

Hey - thanks for the response. Especially, thanks for being a dick about it. While you're at it, explain to me how I was to obtain drivers from the website when my network connection wasn't working. I look forward to your wisdom with great anticipation.

Re:In all fairness

[harryjohnston]

Use a different computer? Or a dial-up connection? Or download the network driver before performing the re-installation? Or use the drivers on the media which HP presumably provided with the computer? Or add a different network card temporarily ... oh, wait, that's what you did ... you didn't really need to reinstall after adding the wireless card, though. (On the other hand, Windows installs are so painless nowadays that may have been the easiest solution. We can call that one a tie if you like.)

The point is, this is a perfectly normal situation - one people deal with all the time - which has nothing to do with either the specific HP model or Windows 7. Heck, I recall a time when operating systems hardly ever shipped with drivers built in, so this was a process you went through essentially every single time you installed or reinstalled an OS [1]. There's no need to go blaming either Windows 7 or HP just because you actually had to do some of the work this time!

Sheesh. Young folk today, eh? Expect everything handed to them on a silver platter.

    Harry.

[1] And the driver download was uphill both ways, in the snow.

So, a 0.5% faillure rate....

[rueger]

The guy does 400+ successful installs, then runs into a decidedly obscure hardware problem, and people flame him? And Windows 7?

Yee Gods. Get a life folks. I read this as a success story, both for the author and for Microsoft.

Re:So, a 0.5% faillure rate....

[Anonymous Coward]

obscure? he was running an engineering sample of a CPU.

That being said, it scares me to think someone who does benchmarks for Toms Hardware switched the case before the CPU...

Re:So, a 0.5% faillure rate....

[Kaboom13]

It's more that he did 400+ installs, clearly considers himself an "expert" yet is completely incompetent as a technician. I don't mean to be purposely insulting, but it is the truth. He has a problem installing Windows, and instead of first verifying the integrity of his hardware (using any of a wide variety of tools available for that express purpose) or he buys a new motherboard. When that doesn't fix it, he replaces everything else and buys a new power supply. When there is nothing else left, he finally thinks to replace the CPU, which all along has been an unsupported (by definition) engineering sample CPU. If he had been doing this for a 3rd party, it would have been many times cheaper for them to throw it into the trash and buy a new one then follow his diagnostic method. At no point does he use his knowledge of how a computer works, how Windows works, and specifically how 32-bit and 64-bit Windows differ, to assist his diagnosis. For comparison, Heres how I would have approached the same situation:
1. Check the setup logs for info. Perform some quick googling for known problems with this chipset and this processor in 64-bit mode.
2. Try a DVD based installation (this saves you time, you can rule out a problem with your thumb drive and a problem with your systems USB booting support at the same time)
3. Disconnect all peripherals. Anything not absolutely needed for the install goes. If you have multiple sticks of ram all but one goes too. BIOS settings are set to the most stable possible and to spec for components. (in a custom built machine, this is often NOT the mobo defaults). Disable any onboard sound/lan/usb etc.
4. Try install again
5. Run Memtest
6. Verify CPU is supported by the motherboard, check for BIOS updates that may contain microcode updates for that CPU (at this time if I had the knowledge the CPU was an engineering sample, I would already consider it highly suspect)
7. Test the system for stability by load testing it under 32-bit windows/pe/linux. I like the tool Prime95 for this, there are many others. Although it worked fine in 32-bit Windows before, that is no guarantee that it was in fact perfectly stable. Working from a known good slate with minimal variables is important.
8. Test the system by installing a 64-bit OS other then Windows, or booting to a boot disc that is 64-bit. (Is there a distro that makes a good boot disk that uses a 64-bit kernel? I've never needed one before now but I would be interested if anyone has suggestions. If it is possible to make a 64-bit PE disk, I would probably try that too.)
9. If the 64-bit linux works without issue, I would probably proceed to start looking for obscure windows compatibility related causes. Presumably the 64-bit linux would have had stability problems also in this case, I would have to assume the problem lies in the cpu or motherboard, or the combination of the two. If I knew the CPU was an engineering sample, I would replace it first. Otherwise, I would contact Intel, the chipset maker (generally Intel for Dual Cores) and the motherboard makers tech support to see if they knew of similar issues. Tech support often has access to beta firmwares or by request only hotfixes they don't publish. If they had no solutions, I would replace motherboard or CPU, whichever is cheapest, one followed by the other.

This list is pretty long, because it is a fairly obscure problem (99% of similar problems would be found by #1-6) , but it would be faster and cheaper then throwing hardware at it (and does not rely on having a large stockpile of spare components to swap out with, or waiting on shipping or trips to the store), If you went to a mechanic because your car was making a noise, and he solved it by replacing every component one by one until you had a new car, you'd hardly call him a master mechanic. The same thing here. He throws hardware at a pc problem the way a complete amateur would. Our industry is FULL of self-proclaimed experts, who know little but claim to know everything. I

Re:So, a 0.5% faillure rate....

[tibit]

You forgot one step: run gimps. I've been to a place with a bunch of rather crusty early P4 PCs. One in three (n=50) would fail gimps stress tests within 24 hours. Naturally enough, there were complaints of "Windows crashes" etc. on those very machines.

Re:So, a 0.5% faillure rate....

[Kaboom13]

No, I didnt. See #7.
From wiki:
"Prime95 is the name of the Microsoft Windows-based software application written by George Woltman that is used by GIMPS, a distributed computing project dedicated to finding new Mersenne prime numbers. As of September 2009[ref], 13 new Mersenne prime numbers have been found by the network of participants, and, on average, a new Mersenne prime is discovered approximately every year. The Linux-based version is called MPrime."

Lost weekend

[Anonymous Coward]

may help you troubleshoot a problem of your own, and save you from a Lost Weekend

I thought people really liked that show. What's wrong with watching the episodes on friday night, saturday, and sunday?

It's interesting...

[The Spoonman]

Nowhere in the original article did I get the sense that the author was blaming Windows for his issues. In fact, he starts out by stating that he's installed Windows 7 hundreds of times without a single incident, but this was a "problem PC". So, how did this turn into an anti-Windows rant? Oh, right, it's Slashdot...

who's written over 100 books

Michael Behe's written dozens of books trying to debunk evolution. It does not make him an expert in evolution. He installs Windows, copies down what he sees on the screen and writes it down. That does NOT translate into "he knows what he's doing". I'm not saying he's not an expert, just that it's not a valid qualification.

If this hardware geek spends days solving a CPU-meets-Windows 7 problem, what chance do mere mortals have?"

They wouldn't be installing an OS. Very few non-geeks do so. They buy a computer from a vendor like Dell, it comes with an OS. When it's time to upgrade, they buy a new PC and give the old one to their kids or grandparents. They also, as has been stated numerous times in the comments, wouldn't be installing on machines that had an engineering sample for a CPU. Actually, this debunks the claim that because he's written books, he's an expert. He knew he had a machine with an unsupported processor in it and still replaced everything in the machine first. Um....duh!

Re:It's interesting...

[Barny]

Well, first of all he might not be an expert but he is a professional, which is mildly disturbing.

Although he claims 400 installs (personally I accomplish about 6-10 a day, not counting pre-install/slipstreamed systems installed from my installer disk by another person) he admits at the end he has only installed about 30 intel based CPU systems with Windows Seven, this is not a large amount when you think about it, although I guess this system counts about 10 times according to his description about his final install.

As other people point out, he replaced the PSU and case!

He may be a professional, but I would worry about his time spent on useless tasks if I was his boss :)

All I have to say is . . . .

[bogidu]

60 to 90 minutes to replace a motherboard?

Ed Tittel...

[actionbastard]

The Gregory House M.D. of OS installation.

Re:Ed Tittel...

[HBI]

You have to admit, it wasn't lupus...

Has his geek card been taken yet?

[MrCrassic]

Standard approach for solving installation issues:

• Check installation media.
• Check cables.
• Remove components; add one at a time.
• Swap memory/move to empty bays.
• Change RAM completely. (Or test it with MemTest86+ et al, i.e. the REALLY FUCKING LONG way)
• Change hard disk (or test it with manufacturer-provided tools et al)
• Test with different CPU, if available
• THEN test motherboard.
• Replace power supply.
• At this point, it's a complicated problem.

This isn't even geeky; it's rote procedure for anyone that's been in the tech support business long enough to see enough weird crap happen. Glad he learned, though; you need to start somewhere!

For those that didn't read the article, he apparently fiddled with BIOS settings and replaced the mobo before even looking at the components.

(Obligatory nostalgic moment: I remember when Windows was much younger and error messages were oh-so-much crueler. Those were the times. Then again, blue screens nowadays are still pretty cruel, though way more informative. [STOP codes, memdumps, and even culprit FILES sometimes are great] Remember the completely useless VxD blue screen errors from W9x?)

Re:Has his geek card been taken yet?

[Anonymous Coward]

Generally speaking if you have an actual memory problem, MemTest86+ or any other BootCD memtester will detect it within the first 2 tests.

Re:Has his geek card been taken yet?

[Anonymous Coward]

I remember when Windows was much younger and error messages were oh-so-much crueler. Those were the times.

If only those times were totally gone. BSODs are informative, but the number of apps which give some horrifically cryptic error when trying to update without an internet connection is annoying (i'm looking at you MBAM).

In response to your list, I mostly agree, though pulling all nonessentials (PCI cards, etc.) should include front panel headers.

Re:Has his geek card been taken yet?

[MrCrassic]

Yeah, but that's a program-specific error though. And they still give you an error code; way better than this. Of course I know that memory region; I write all my bits to RAM manually! :(

new lesson; any trouble, just start shooting

[Anonymous Coward]

that's how the corepirate nazis do it. just fire randomly/blindly into the 'audience', then demand applause. anything to keep your eye off the 'ball'.

as bs is not usually a fatally effective weapon, there's lots of repeat (daily) performances.

Buggy CPU, clueless dude

[Hells Ranger]

I did read the article. The buggy part he found after swapping everything is the CPU... An engineering sample. The kind of part that should be the first on the potentially defective part.

Most component called engineering sample are called that for a reason. Usually it really mean : Wow I can't believe it kinda work!

If he's supposed to be good debugger, I fear for everyone who need is PC fixed.

My lesson.

[w0mprat]

Just last night I fixed my parents computer in one of those long fixes that turns out to be the most fundamentally trivial things. This is why this is not my main occupation.

Basicly they had a reccently built custom Windows 7 + Ubuntu PC that had begun randomly shutting down, often minutes after it had been powered up.

Ok first thing, any obvious errors or cicumstances? No, it would just randomly power off. Windows event logs showed kernel power events, no specific driver, service or app crashing anywhere. Linux was the same. Not a thermal issue cpu + gpu temps nominal and stress test din't immediatley cause a crash.

Suspecting a power or a motherboard issue, first checked and re-seated things internally. It still occured.

Removed extraneous cards, connectors and drives. No result. It would even happen sitting in BIOS setup. Have ruled out a number of problems.

Checked for electrical shorts, poor voltage etc.

Dying power supply? Overloading or shorting? Nope, all voltages nominal, and it was brand new.

I was about to try a spare power supply and a thought occured to me..

It's almost as if the reset switch was being hit, but it wasn't even close to being knocked at any point and the switch otherwise worked fine. Then I knocked the case and the system reset. Yep, the reset switch was faulty, jolting it even slightly would reset. Who needs a reset switch since Vista anyway? Unplugged it from mainboard. Solved.

I decided not to even joke about charging my Dad for two hours of my time.

Chances are if he paid someone to do it they wouldn't necessarily have found the fault that quickly, and he'd be hundreds of dollars out of pocket.

The lesson in troubleshooting? Um... I'm not sure.

Re:My lesson.

[Anonymous Coward]

Er, how exactly did you get "kernel power events" when the reset switch was hit? Reset switches are dumb enough not to do any cute ACPI tricks, but just hard and directly.

Re:My lesson.

[girlintraining]

The lesson in troubleshooting? Um... I'm not sure.

You did exactly what any computer tech should: Check the most common reasons for failure, then move to the edge cases. A faulty switch is rare. Swapping individual components out would have eventually narrowed it down to the case itself. Two hours sounds about right for a competent technician to run down the list to get to the point where that would be a likely cause of failure.

Re:My lesson.

[Anonymous Coward]

I once had exactly the same problem with a brand new case (an all-perspex one which I thought was cool at the time but which in reality is a nightmare to cool).

Was randomly restarting, so I suspected pretty much everything from power supplies down.

In retrospect I was kicking myself. After all, with a symptom like random restart it is only sensible to suspect the button responsible for, well, random restarts. But at the time I spent a good day and a half trying ever more desparate random shit (replace power supply, systematically jiggle cables, re-seat cards for the 20th time) before the lightbulb moment.

Re:My lesson.

[Anonymous Coward]

While your comment has already convinced me that you're better at this than the guy from TFA, you fuckup one part:

Removed extraneous cards, connectors and drives.

Thats a great idea but you did it wrong. That step includes pulling all the front panel headers (power button, reset, HDD led, etc) and starting the machine with a flathead. You described the machine as "custom" so I'm guessing it didn't have some proprietary stupid Dell front panel connector. And like everyone else said, why someone wouldn't suspect an engineering sample earlier is beyond me.

Re:My lesson.

[Anonymous Coward]

A faulty switch isn't as rare as you think.
It's the only problem I ever had to fix on my parent's PC, but that I had to fix twice (Power Button and Reset Button).

Re:My lesson.

[British]

Funny you should mention that. I too had a system that would just randomly power off, as if someone pulled the plug. After replacing the power supply and just about everything else, I just replaced the mainboard & processor & called it a day. It was getting old anyway.

But I thought the reset switch would leave the system on & just reboot, not power down. Not even the soft power switch does an instant shutoff, unless you hold it down & it goes through the motions.

Obligatory

[Anonymous Coward]

...

Chaos Manor?

[dpbsmith]

How many people had the same impression I had: "Why, this sounds exactly like one of the 'Chaos Manor' columns Jerry Pournelle used to write in BYTE!"

All it needs is a few of Jerry Pournelle's favorite stock phrases. "The disk trundled for a while..." "I tried swapping out the hard disk, but no joy..." "I called up Bill Godbout..."

Re:Chaos Manor?

[Anonymous Coward]

Same here. Those were the days.

Someone needs a helmet...

[Anonymous Coward]

The OP needs to be careful, there are sharp corners present.

Wow! He's written over 100 books!

[hdparm]

That's exactly why I wouldn't trust him much. Takes time to write a book. More time to write it well.

Fixing Windows is a lucrative gig...

[Anonymous Coward]

...and it's a good thing, because then I can afford a Mac and don't have to do shit like that on my personal time.

Obligatory blah blah blah

[Colonel Korn]

http://xkcd.com/627/

Engineering Sample ... thats just plain obviousnes

[Anonymous Coward]

Nice article but good god it was an "engineering sample". Thats the equiv. of "Beta" in the chip world. They are unfinished chips, or those that are not yet properly characterized so you should expect it to fail. Not really a storey here, nothing is at fault other than someone should have warned you that things may be very unstable.

Swap and Pray

[Cylix]

That was not a strong article in troubleshooting.

He basically used the parts blast technique to isolate the problem. No where in the article did I read any actual troubleshooting steps.

In situations like this there are two methodologies that can be applied. In a system with many components its usually easier to use a reduction technique. It's just a play on the isolation technique, but removing a large number of components.

For home troubleshooting you can still get fairly far without a lab full of equipment or spare parts. However, you don't need to rush to the store to replace every component. Memory, hard drives, processors and even some testing can be targeted against the motherboard in a limited extent.

If you are very lucky you may even have some debug information available from the motherboard itself.

A good linux rescue cd with the right tools for testing and diagnostics can do a world of good. Unfortunately, I don't have time to write a really good article and apparently the TFA didn't either.

Happened to me...

[EgNagRah]

but I wasn't using m$ winblows, I was installing Ubuntu from a 16gb flash drive on to an asus eee. That might have worked if I knew what I was doing...

Wow, what a dope

[frist]

What were those 100 books on? Think about that - how many years has be been writing books. Say it's been 10 years. 10 books a year? A book every 5.2 weeks? WTF

Re:Wow, what a dope

[frist]

Forgot to mention why I initially thought of him as a dope. Instead of doing a normal install, where you burn the DVD ISO to a DVD, he fucks around with a USB drive. Some mobos require driver updates before things like the USB controllers work right. Whenever you make things more complicated, you're just asking for trouble. Take the simplest approach when installing an OS, especially if it is a new (to you) hardware configuration. While this turned out not to be his problem, continued reading only further reinforced my initial opinion - he's a dope.

Re:Wow, what a dope

[Jedi Alec]

Some mobos require driver updates before things like the USB controllers work right.

Some mobo's also require driver updates before the ata controllers work right. In either case the BIOS should provide sufficient support for the install the complete after which all the bells and whistles can be activated through the drivers.

Just get the right BIOS!

[Acting Ordinant]

Here's what I posted as a reply to this "expert's" article. It's now awaiting admin approval to appear as a comment. We'll see if it makes it.

==================
While reading, I was thinking this was a well-written detective story. Then I got to the end and found out it's a story about a massive waste of time because you didn't follow standard procedures.

Here's how to save a few days next time: go to the motherboard manufacturer's website, get the list of supported CPUs for the motherboard you're trying to install. Then download and install the BIOS that supports that CPU. It really is that simple.

Asus is particularly good at providing a CPU support list for their motherboards. It took me entire minutes to find the lists for the P5Q3 and P5E3 Deluxe (not P5E3 Pro, as you wrote). The QX9650 is listed for both motherboards -- and in both cases, it is supported only as of a recent BIOS revision.

So all you had to do was download and install BIOS version 0204 or later for the first motherboard, the P5Q3, and I bet Win 7 would have installed correctly the first time.

As for the motherboard automatically making BIOS changes to match the fast DIMMs you installed, Asus motherboards do NOT do this by default. You must have left the BIOS in some sort of overclocker's mode.

Next time, look up and download the BIOS that supports the CPU you're trying to use. After installing it, use the BIOS setting that restores all other BIOS settings to their defaults. Then install the OS. THEN and only then, can you start tweaking BIOS settings.

Once again, the article was well written. But it's also an inadvertent confession.
==================

Re:Just get the right BIOS!

[Anonymous Coward]

This is from my experience, but any PC builder's first rule when building a computer is to make sure the mobo, RAM, and CPU are compatible. Of course, it is assumed the case is decent, and the PSU is a quality one (for noise reasons). Compatible as in explicitly supported on the motherboard. Yes, the new chip might be pin compatible, but if the motherboard doesn't explicitly support it, it might not even boot to a BIOS flash utility.

I'm glad he's an author and not an engineer

[Anonymous Coward]

A test CPU would be the first culprit to come into my mind on a new OS, but then again I guess that's why he gets paid for writing books and not solving real world problems with a boss or customer breathing down your neck.

Clueless

[Animats]

The gjy is using an engineering sample CPU, he has a motherboard set up for overclocking, and he doesn't have a set of CPU diagnostics. Clueless.

A wasted weekend?

[dogzdik]

Ummm by "week" and "end" - were you referring to the 2 days of the week between Friday and Monday; or where you referring to the days between Sunday and Saturday?

.

With Microsoft and all of it's bullshit, I can hardly remember a weekend spent fixing it - didn't run from Sunday to Saturday - that being Sunday the first of the month and Saturday being the first day of the next month.

Interesting idea, but not the same...

[Mathinker]

One of the reasons I use Linux is that, currently, it is much more secure than Windows, given my personal use scenario.

Yes, if I were a specialist in securing Windows that might not be the case, but I'm not. Yes, if equivalent amount of effort was invested to break the security of casual users of Linux compared to that invested in breaking Windows, again, Linux might not be any more secure than Windows (well, with Linux, there are distros where I can always boot off of USB and then not save any changes, so until Microsoft offers me the same functionality there's little chance that I could use it in as secure a fashion as I can use Linux).

Running Linux in a VM under Windows just wouldn't "cut it" for me. Sorry.

Re:Interesting idea, but not the same...

[elnyka]

One of the reasons I use Linux is that, currently, it is much more secure than Windows, given my personal use scenario .

If I could mod up you more, I would. Very few people in /. explicitly make the caveat of their own specific work needs. Kudos to you sir.

Re:Interesting idea, but not the same...

[badkarmadayaccount]

Actually XP embedded offers ro-union mount type functionality with a special driver. Though it's not as easy as with linux.

Wrong question

[Mathinker]

what chance do mere mortals have?

Yes, what chance do mere mortals have of being in the particular use case this super-geek is in, using an engineering sample CPU?

Answer: 0. Well, not exactly 0 --- you never know exactly what you're buying nowadays!

And even if not

[Sycraft-fu]

What kind of expert is he if he can't figure this out? I mean I'll give normal users a pass on this kind of stuff. It is complicated, no question about that. To the extent I'll razz normal users it is to not mess with it. However, if you are a supposed expert, you should understand it. While complicated, it isn't really all that hard. Learn about what the different kinds of settings are, learn the defaults (or know where to look them up) etc. This is the kind of thing to check when you are troubleshooting a system.

This is also why OEMs turn them off in their BIOSes, because they don't want you screwing with them and then lying and saying it is all default.

At any rate I don't cut this guy any slack. He wants to pretend to be an expert, yet clearly knows fuck-all about the BIOS. Anyone who has mucked around with computers ought to understand it.

Heck, you run in to this stuff even if you don't mess around from time to time. When I upgraded my desktop to 8GB of RAM, it suddenly started having some random reboot problems very occasionally. Going back to 4GB fixed it. Odd, board supports 8GB. So I had a look see at the settings. The RAM was "high end" RAM that had it's SPD set above JEDEC timings, however it wanted 1.9v to do that. My board supplies 1.8v only. Ok, so I went in to the BIOS and set the RAM down to the JEDEC timings (1 more CL). All problems cleared up.

At work we had a system that we got 8GB of RAM for. Older system that uses 533 or 667MHz RAM. We got 800MHz since that's what they sell these days and it functions fine at lower speeds. Well, turns out the board will try to support 800MHz, however with all 8GB in there, again reboots are a problem. The memory controller just can't handle it. Unfortunately, you can't change the timing in the BIOS. So instead put in a set of the 800MHz and a set of 667MHz, which forces the system to 667MHz. All works good then.

While I won't call this basic knowledge, the sort of thing a user ought to know, it is sure as hell part of doing hardware support.

Stupid

[ledow]

Well, for some kind of "expert" I find his component-swapping "diagnosis" a bit dubious to be honest. Yeah, a lot of the time you can just swap out a suspect component but swapping a hard drive for a installation-hang? It's towards the bottom of my list, especially if that drive has handled other OS installations without any problem and doesn't have SMART errors (Did he check? Did he clean-format in between? Even a corrupt NTFS can exhibit those same symptoms... nary a mention of things like that). I'd actually be suspecting a Windows bug before most of the stuff he replaces, considering that he suggests he was already using that hardware fine on the 32-bit editions of Windows - I'd be making an up-to-the-minute Windows 7 installation CD with every update I could find.

And the graphics card replacement is just completely illogical unless there was some hint that the graphics card was the fault - bad imagery, hangs happening on a resolution change or reboot, etc. The graphics cards use doesn't change during the installation except at those points and so unless there was a temperature problem or something related, it would just be *weird* to suspect that. Power supplies can present odd problems certainly but it makes me wonder how reliable his testing really is normally if he's powering things up on the wrong wattage of power supply without even noticing and then possibly publishing benchmarks / stability reports etc.

By the time you get to the casing, he's just clutching at straws rather than thinking. I'm not saying that I'd suspect the CPU immediately, or even at all, but the logical processes he's using are just dubious. I mean, he changes the memory before he remembers that he has some stupid BIOS option for upping the memory speeds enabled. And doesn't think to do a simple memtest at *ANY* stage? Surely a memtest would have picked up the same memory errors before all the testing, and that would lead him to check the BIOS and then replace the RAM way before he does? It saves a lot of hurt to do *checks* like that, rather than blindly assuming, especially if you're replacing one set of RAM with another that is "older, and well-tried and trusted" (with memory, there's no such thing as tried and trusted... they can die overnight, don't take kindly to handling, etc.). He was using a non-standard CPU cooler, too, which doesn't get mentioned until later in the article. I'd be suspecting temperature problems with his setup which he doesn't even consider.

But at no point does he actually suspect the *only* engineering *sample* that he's ever used and that's been in the machine all along. He just blindly replaces it eventually but he never suspects it until he "finds" the problem.

Hell, must be nice to just be able to order random components until your problem goes away. Other people have to do the same job, with cheaper, shittier hardware, no budget, no diagnostic tools, useless vendor support, while under pressure, to earn their living. But then according to the Wiki he's a "freelance writer", "trainer", "Internet consultant" (Yeurk...), "Director of Technical Marketing" (for a year), "Technical Evangelist", "Director of Training", "Senior Researcher", "series editor", "currently writes for... TechTarget.com Web sites... Tom's Hardware and Tom's Guide... the American Institute of Certified Public Accountants (AICPA)... InformIT.com., "content development and book publishing projects". To be absolutely honest, the glaring thing that sticks out like a sore thumb in that list is writing for Tom's Hardware - everything else is book/article-writing and sales. This guy isn't a tech - he's a marketer. Why his "diagnosis" should be interesting at all is beyond me, except possibly as a humour article.

Oh, and the article seems to just be a way to show off brandnames, model names, and his hardware. I expect as much from a marketing guy.

RAM, it is nearly always the RAM.

[the_raptor]

Real trouble shooters pull RAM first* because mis-seated and bad RAM is the cause of most hardware errors. After that you pull the PSU and then the GPU. If that still doesn't find the culprit you test those parts in known good machines and then swap out the CPU to isolate it or the MB. This guy didn't even have a serious problem, a real hardware gremlin is when you have combination issues and can't isolate a single faulty component. A month in the trenches of front line support teaches you these lessons that you won't learn from a lifetime of building the occasional machine for yourself.

That said if I had a frigging ENGINEERING SAMPLE anything I would pull that first. Anyone who knows anything about hardware production knows engineering samples are for troubleshooting and not freebies for journalists. It is like over clocking, far too many people expect that OCing is guaranteed and stable because most gear these days has some head room. I have seen people whining and blaming the manufacturers because their hardware didn't get the OC that everyone else got.

*And if the user has been OCing or fiddling with BIOS you set that to default as well.

However quad core cpu.

[leuk_he]

If you have a quad core processor you need a 257 bit power supply.

That extra bit is for the power switch.

Re:However quad core cpu.

[FreonTrip]

No, you taffer, 258-bit! One bit for the power switch, and another for parity.

Also quantity and quality are often exclusive

[Sycraft-fu]

When you crank out a lot of stuff, it is extremely hard to make all of that stuff be high quality. Quality usually takes time, it takes research, it takes refinement. It is possible, in some rare cases, to have someone that produces a vast quantity of work, all of which is top quality. However it is far more common to see someone produce a vast amount of mediocre to bad quality work.

As an example: Dr. Mark Russinovich has written a grand total of three technical books to date. So, clearly a man who doesn't know what he's talking about right? Wrong. Those three books are "Inside Microsoft Windows 2000," "Windows Internals Fourth Edition," and "Windows Internals Fifth Edition." He has, literally, written the book (along with David Solomon) on the recent versions of Windows, published by MS themselves. These are extremely accurate, comprehensive, technical documents of Windows down to its very fundamental levels. He also has written a suite of tools, the Sysinternals tools, so good that MS bought them, and hired him on as a technical fellow.

So while he's produced only three books, they are all of the highest quality of technical information. There haven't been more because he hasn't had the time to write hundreds of books, nor the need to issue revisions to correct problems with the ones he has (each new edition covers a new version of Windows).

Thus when I hear someone talk about how good they are because of the quantity of they works, I am skeptical. The only way you get a vast quantity of high quality work is either laboring an entire lifetime (and even then often not), being a prodigy, or both.

The lesson here is that people still don't have...

[sarkeizen]

..a clue about how computers work. Even experienced windows professionals.

I mean this guy has 32 bit OS working and moves to 64 bit OS...am I following this ok. The 32 bit install presumably went well on the hardware and the 64 bit install fails.

So I grok his first attempts which are replacing the install media once. Seems like a reasonable assumption (some bit out of the billions on the DVD image just happened to be flipped the wrong way). From there though he starts to lose me. The motherboard is perhaps plausible but you would have to be assuming some rather significant difference in hardware support between the 64bit and 32bit systems. From there? RAM is 64bit how? Or even my HD?

I think the most significant thing to learn here is twofold.

i) People - even experienced computer professionals - treat computers like they are magic. Like there is no real science behind how they work. Clearly this guy was replacing parts based on some "experiential weighted average" with regard to how likely they are to cause a "weird" problem.

ii) When A. C. Doyle said "When you have excluded the impossible" he neglected to state that the *order* in which one does so is significant. Eliminating things in order of their apparent relation to the problem (i.e. all the things for which 64 bits makes a difference) and (in a business environment) with respect to cost (i.e. Replacing a CPU is often a cheaper test than replacing a motherboard wrt labour) will likely fix your problem sooner than just going for the "usual suspects".

Aside: I've had two cases where I found a CPU issue. One was very similar to this - crashing during a Windows 2000 install - often at the same place. The problem I had was actually thermal - the heatsink was reversed leaving the thermal patch making minimal contact with the heat spreader. Somehow I figured that out without replacing everything else first.

Apple

[Anonymous Coward]

Once again I am reminded why I buy Apple. Back in the day I used to love tinkering with this sort of thing. Then I got bored of that. Now I buy Apple and it just works.

I'd do it the other way around

[Viol8]

INstall linux and run Windows in a VM. When your windows install gets infected/hosed with a virus/malware/whatever it could well mess up your linux VM machine and make it inrecoverable but if you install Windows in a VM and run on top of linux the worst that can happen is the VM gets hosed.

Re:I'd do it the other way around

[Anonymous Coward]

You missed his point for putting Linux in the VM. He'd have to reboot if he did it your way.

Drop to Windows if you need a Windows app (say, a recent version of Photoshop or real MSOffice) or to play games.

Re: 100 books

[TaoPhoenix]

I'll go a little easier on the fellow and his 100 books. It's hard to not learn at least one nifty trick per book you write. So I'm sure the guy knows a few hundred nifty tricks.

I like your last sentence better. The article tone is wrong. Instead of "the world's hardest install", it should have been a "Friday-At-7PM-at-the-bar" story over a triple beer.

"Okay guys, I know, I know, but when my special spec system crashed on install, I was kinda fried and my method was like going to New York via Alaska..."

Re:No kidding? Read the article...

[Anonymous Coward]

The overclocking WAS in default!!! The BIOS mucked around with FSB, memory and clocking due to a DEFAULT "optimize" setting. It only worked AFTER he changed the settings AWAY from default!

On top of which, he had been running the CPU with stock specs and 32-bit Win7.

Meanwhile we're supposed to take someone seriously who either doesn't read or misreads the article then launches into a tirade. Sheesh.

So, when is it going to be

[Compaqt]

the Year of the Windows Desktop?

Windows 7 Upgrade Advisor?

[freefallden]

So this guy is writing a book about Windows 7 and in his blog at least makes no mention of the Windows 7 Upgrade Advisor? http://www.microsoft.com/downloads/details.aspx?FamilyID=1b544e90-7659-4bd9-9e51-2497c146af15&displaylang=en Granted he is installing a different version (x86 vs. x64) and edition (ultimate vs. pro) of the same OS, it is definitely not a given that driver support is the same with different versions of Windows.

Bad practice not going with a 24 hour burn-in...

[KennyP]

Epic FAIL - not bothering to bench burn-in any change in system hardware.

I've been building PC's for well over 25 years, starting with my empty case IBM PC 5150 (no drives, cards - nothing but 64k soldered on the system board).

First test - TEST! Then, test it again.

I currently use PC-Check to run a diag for a minimum of 24 hours. This does nothing more than to check if you're going to release any "magic smoke" because of an assembly failure.

After that, I load the OS.

I had an old Novell-brand 386 server (yes - they rebranded hardware back in the 2.12 SFT days) that had a bad CPU, but, only intermittently. Worked great when cold, would lock up dead after 10 - 15 minutes, but not all of the time. We replaced the server, I took the old one home, popped in another 386-16MHz (yes - the days of unbridled power!) and it worked flawlessly 24x7 for 8 years until it got too old to bother with and chucked into the shitcan.

Test and test again. That helps ensure your 5 9's of availability!

Re:It goes both ways, actually

[lien_meat]

People that prefer linux will like to believe that windows will not ever be reliable, and people who prefer windows tend to believe that linux won't be.

I've seen this time and time again, both through myself, and my friends/acquaintances.

Just two recent examples of this:

A friend of mine bought a really nice Asus i5 gaming laptop with win7 home premium preloaded on Thur last week. I envy his laptop. It's awesome. Anyway, on Friday, after only running the laptop 2 times, he turned it on, only to find that win7 wouldn't boot. Luckily Asus provided a recovery disk (some OEMS require you to burn your own, he wouldn't have done that that soon). My very first reaction (as I do prefer linux, and I myself happen to have many issues just like this one with any and all windows oses, my friends will confirm that I do for some reason, whereas they do not usually...) was to start laughing, and I remember saying "When will MS get their ^(&# straight?". Now, we all know that it's possible it was his fault, not win7, but it wasn't my first reaction.

Likewise, a different friend of mine has a win7/ubuntu dual boot setup on his desktop, and at one time had xp and ubuntu via wubi on his laptop. He very very rarely uses ubuntu, and I honestly don't even know why he insists on having it installed. His antivirus program thought the wubi virtual disk of ubuntu was a virus, and deleted/broke it. He instantly blamed ubuntu for being unstable (who wouldn't). It was only later, after it happened again and we caught his antivirus in the act that we figured out what was happening. This same friend had a sata cable fail on his desktop which was hooked up to the drive with ubuntu on it. Because of this, his computer wouldn't boot, although oddly enough the drive did show up in bios. He blamed ubuntu instantly for the grub error that appeared when grub couldn't find that drive, and had me come over and help him "fix ubuntu". Well, we figured out pretty quick that he just had a bad sata cable, and after replacing it, everything was perfect.

In both cases, the operating system might not have been at fault. (ok, so ubuntu wasn't in these cases, but I've seen it become unbootable on a netbook after a regular "apt-get upgrade" too..., and the win7 issue might have been my friends fault too, I don't know.) However, it just goes to show how our personal feelings about our favorite operating systems get in the way of factual information about them, and many times people say things and bash them before they think about it. The same goes for both windows and linux.