Plague of dodgy capacitors starts to bite

Hi Mike,

A friend brought her PC for me to see why it powered itself off and why the modem quit working.

Then the video adaptor quit working, prompting me to take it to the neighborhood clone shop.

As Dirty Harry said: "A man's got to know his limitations"

They noticed that some of the capacitors on the Gigabyte GA-7VKML motherboard were leaking, as described in your article:

Dodgy components threaten mobos, modems here.

I researched the problem so that my friend would know the clone shop wasn't just trying to sell her a new PC:

LIGHTSPEED 2004 BASIC PC

"Complete system only ...$288.88...with Lifetime Warranty..."

I found several articles such as this one:

Leaking Capacitors Muck up Motherboards

This site listed the symptoms and brand names: MotherBoardRepair.com :: Mother Board Repair At A Component Level. "Most motherboard related failures are due to the "On-board" regulated supplies and component failure within those circuits".

This site lists capacitor failures on almost every brand and model of motherboard manufactured between 1998 to around 2001.

It says the most common symptoms of capacitor failure are: "1. Failure to boot; 2. Must attempt booting several times before machine will start; 3. Instability , especially when graphics are complex; 4. Machine boots with a pre-Coppermine Celeron but will not boot with a PIII Coppermine; 5. Bad odor wafts about the room and this time it wasn't you or the dog; 6. BIOS health alarm (hi-low siren) at boot but PC health screen shows no reason for the alarm; 7. Fans spin up, power indicator lights up and nothing else happens..."

I feel sorry that my friend has to buy a new PC because the repairs to her PC are too expensive. My soldering techniques aren't good enough to replace capacitors on a motherboard.

Too bad these companies can't be "encouraged" to do the right thing for their customers, the way Intel finally did for the FDIV bug ~10 years ago.

Similar sentiments are expressed in the IEEE article:

"Carey Holzman, as a builder of custom PCs, has been trying to raise awareness about the defects since last spring. He thinks manufacturers should be more public about the problem and issue a recall. "Main board replacement is a big job. It's a huge amount of downtime for the user," he says. Failures can also occur after the warranty has expired, he points out. "The manufacturers should do the right thing."

Jerry Leslie

Microsoft Update will ask to check your software licence

Hello again:

I think with this latest maneuver Microsoft may be going the final step too far. They certainly would like us to think that they have an eternally improving product, worthy not only of subscription licensing (or at least they tried that for big business) but also we should have no qualms in letting them make sure we're legit just so they don't lose any potential profit.

Trouble is Windows XP is showing its age. It's a 2001 operating system that was cutting edge for 1999-2000 when it was being developed. All we've been doing since then is patching one security hole after another, with scant little improvement in the OS itself. (But wait! It's perfect! And notice the wireless networking thingie in SP2! And--and--Media Player! Just don't try to uninstall it!).

XP is nominally a good OS, but that damnable product activation and all the spyware makes it a pain the posterior. How about the novel idea of doing away with the product activation and just giving us a dongle? My copy of Lightwave uses one, and I've lost track of how many computers it's been on. But I can only USE it on one computer, period. But no! Microsoft can't live without getting our hardware specs and checking up on us for purposes of quality assurance and superior experience. Have they ever thought I'd buy more hardware if Windows wasn't so difficult?

Here's an even more radical idea: Reduce the price of what is essentially three-year old software. The full retail price of XP Professional is still $300 from a lot of places. But no, they need the extra money for endless bug fixes!

My latest quandary was about three weeks after installing SP2 my system totally packed up, most likely the registry. Rather than spending hours trying to fix it, it would be simpler to just re-install it. Except I'm using the Home upgrade, which requires the Win98 CD I gave away some time ago. (Silly me! XP is perfect, right!) So after finding another Win98 CD I'd then have to ask the Microsoft serf to let me reactivate for the 7th or 8th time, and finally I'd have to walk the DRM nightmare of reauthorizing all my music and other activated software.

Or I could just re-install Windows 2000 and put on only the most essntial software. Easy choice to make. The only real alternative is Apple (!!!!YIKES!!!!) which isn't really a choice as you also have to be a bubble headed Kerry fan liberal to use one. But it's the only other OS that has substantial commercial software for it. Sure, there's Linux, but none of the SOFTWARE I USE comes for it.

So things get grim and grimmer. Microsoft has lost our hearts and if there was a real choice they'd soon lose our wallets. They can patent until the end of time, sue everyone who has any idea even remotely like theirs, and trademark Webster's dictionary. (Can I still say 'Windows?') It won't make us like them any better. I'll only use XP or whatever-horn OS that follows because I have to, with the minimal outlay of money. Want me to jump for joy? How about something NEW! Something that doesn't phone home and come with DRM restrictions and ten bazillion bugs.

Instead the marvelous choice Microsoft has made is to place crossing guards at every intersection, "Are you legitimate? Wouldn't lie to me, would you? XP is great, isn't it? That's a good boy now, move along."

Stop the Planet of the Apes, I want to get off.

Sincerely,
Scott Peterson
(No, not the one in California)

Offshoring, out saucing

Although I've never been laid off directly due to offshoring, I've definitely seen its effects in my job search. Wells Fargo comes right out and admits it isn't hiring U.S. programmers to program, only to support their offshore programmers. Verizon doesn't even supply hardware; they expect you to bring your own laptop. There's job security for you!

Having said that, I have to wonder what effect the Open Source community has had on programmer employability. I mean, if a bunch of programmers are willing to work for "free" -- although I expect many of them are using taxpayer or even corporate funding or equipment somewhere along the line -- why should a company pay me to write software? I should be so "passionate" about my work that I'm willing to do it for free, too, right?

The last two companies I've been contracted to use almost nothing but Open Source. Aside from Linux, most of it is much lower quality than COTS I've used in the past -- and often a PIA -- but my employers don't care. So what if the stuff is crap? It's the programmers' problem to work around, and we'd better not complain about it.

I've even heard comments like "It's not rocket science!" and "A monkey could do your jobs!" from both technical and business management. Makes me want to go buy a monkey, bring him into work and say, "Here's Bonzo, my replacement. You can direct deposit his paychecks to my account."

Linda Wilson

Intel way behind on EM64T performance claim

hi guys,

i guess numerica optima is skewing their claim a bit. what is typically used for comparison are things like dgemv (double general purpose matrix vector multiplication) dgemm (double general purpose matrix matrix multiplication)

it'd be interesting to see whether their algorithms are better also for that range. it's just a very special application where numerica optima are better, probably due to a better algorithm. this has precious little to do with amd64 vs 32 bit code but all with the actual algorithm used.

and no, i'm not an intel fanboy. quite the opposite, the amd64 cpus are pretty darn good, but intel doesn't deserve blame for this one.

just my 2 cents,

alex

How Intel could deliver dual core chip prototypes

About the article "How Intel could deliver dual core chip prototypes".

The author of this article has come up with a dumb idea.

As a way to get 2 or even 4 cores on the same piece of silicom of course it would work, but the resulting piece of silicom would be totally useless because it could not function.

Why? Because there would be no way to interconnect the cores to each other, to the arbitration logic (non existing) and to the external bus.

The only way this could work would be for the cores to be designed as complementary adjacent pairs, allready interconnected to each other, to an included arbitration logic and with a global external interface compatible with some CPU socket.

But then we would be in the presence of a true dual core solution. Wouldnt we?

Regards
Santos Costa

I don't know if this is still the case, but back in the 'old days' a die was connected to the pins of a chip package by a series of thin wire interconnects. The connections were made on the edges of all four sides of the die. If that is still the case, the technically skilled employee of Insight 64's plan won't work - in simple terms, a quarter of the interconnects wouldn't be made, leaving an unusually dysfunctional P4.

I think the more likely solution is via the "Pentium Pro" route - two individual dies bunged together by the die interconnects at the chip packaging level - the way Intel did with a separate Pentium Pro die and 256/512k cache die on the old Pentium Pros. This is probably the best quick-to-market solution, and Intel has the added advantage that they have experience with the required packaging tech. I doubt it will be cheep for them, though. The 'two dies on one package' solution certainly counts as 'dual core' in my book.

As always guys, and Eva, keep up the great work!

Ben Tarver

Itanic man revises Intel graph

You posted a chart that showed the relationship between mhz, thermal dissipation, and spec int scores.

What boggles my mind is the centrino @ 2Ghz is on par with a 3.2-2.4Ghz P4 on spec_int and is using only 20 watts.

Can you please tell me why intel is not using the centrino architecture for there entire line of PC's????

I would think that if they clock it high enough they could whip AMD out of the water. And if one core is making 20 watts, worst case scenario 2 cores makes 40 watts, why don't they go duel core with that?

I love my AMD but holy crap if I was the CTO and CEO of intel I would have a major product shift in the centrino architecture direction.

Thank you,

Stefan Constantinescu