Clarifications about AMD TLB L3 bug

[chrisv]

In addition to this very valid commercial argument, there is also a
fundamental difference between this AMD "bug" and the Intel FDIV bug:
the AMD bug might slow machines a bit under unusual circumstances.
The Intel FDIV bug gave erronious results in a few cases.
People worry _far_ more about data corruption than machine speed.
The FDIV bug might well have sunk AMD. This will not.

Huh? Were there not work-arounds for the FDIV bug that eliminated the
chance for "data corruption"? Really, from my understanding, the FDIV
bug affected only a TINY percentage of users. The AMD's bug affects
EVERYONE.

Actually, I find the continued harping on the AMD bug interesting
in and of itself. No-one complains of things they don't consider
a threat. AMD must be considered a serious threat.

What silly, saccharin cheerleading. Go AMD!

(snip)

More of the same.

I believe actual performance is increasingly hard to measure.

I believe different. Why would it be "increasing hard" to measure?

MHz hasn't cut it for a long time.

Yeah, we know that.

Intel has focussed on
bandwidth and cache. AMD has worked on latency. As a result,
each processor excels at some tasks and is only mediocre
at others. There is no one "best" CPU.

Sounds like what someone who is getting their ass kicked in the
benchmarks would say.

 

[Sebastian Kaliszewski]

Huh? Were there not work-arounds for the FDIV bug that eliminated the
chance for "data corruption"?

Nope. Nothing except not using FDIV instructions (and emulating it in
software). If every piece of software was not individually patched (with
performance effects of course) users coul'd not belive the results of FP
division of their chips.

Really, from my understanding, the FDIV
bug affected only a TINY percentage of users.

But it affected seriuos users.

The AMD's bug affects
EVERYONE.

EVERYONE of the few who bought the thing. Not too many. And the workaround
works once and for all. Costing performance. But still the thing is faster
than others company offerings (as it's still not enough to counter Core2 is
another matter).

rgds
\SK

 

[Robert Myers]

Robert Myers wrote:

The vast majority of stock is traded without tracking the companies. Ever
heard about algorithmic trading? Then vast majority of human traders base
their opinions on varius indicators not connected directly with company
product. And the reason is simple -- they do not understand what's the product.

You are simply amazing.

Peter Lynch, the founder and long-time manager of the Fidelity
Magellan Fund and widely regarded as one of the most successful
institutional investors of all time, famously taught: "Buy what you
know."

It's true that program trading can produce fast "technical" moves in a
stock; that is to say, moves unrelated to fundamentals. Program
trading can't explain AMD's 2007 performance, and the most recent
stock price slide isn't just a coincidence with the Barcelona fiasco.
Under other circumstances, Barcelona might be a minor fumble, just as
some fumbles in football are of little consequence. When the clock is
running out and you are driving for a touchdown you need and you
fumble the ball, that's like what just happened with Barcelona.

AMD has introduced a product that I really like, something that I've
been wanting a major chipmaker to make for a long time: a GPGPU stream
processor with double precision floating point. It will turn vector
computers into commodities. Unfortunately, it will never be more than
a footnote on AMD's balance sheet. How do I conclude that? The Wall
Street Journal didn't even mention the product introduction.

Then such great manager who does not see the difference between a can od
soft drink and a computer drives the company to the edge of bancrupcy.

CEO's are a different matter, entirely. More and more, CEO's are
given incentives to drive stock price. That often leads to short term
thinking and decisions that are bad for shareholders. The day you see
me quoting a CEO on a company's prospects is the day you will be
justified in speaking to me the way you have.

Even noticed that those most succesfull technical companies are lead to
their biggest success by people who know the technical stuff. Just notice
Intel or Microsoft. Then compare Intel & AMD -- both started as Fairchild
offsprings, Intel was lead by guys who were semiconductor specialist (behind
being good managers) and AMD was run by Jerry Sanders, who was just
management specialist. Compare the performance of the both.

You've just lost the AMD supporters. Jerry Sanders placed a hugely
risky bet on Opteron, and he won. He was a true visionary, a cowboy
almost, in his willingness to take on Intel. I wouldn't have taken
the Opteron bet for any reason, but, then, I'm not Jerry Sanders. I'm
only a techie.

One can look at Intel and see just the opposite. Intel became
obsessed with its stock price and seemed to forget that physics do
matter, even though the company was famously founded and run by
scientists. It mismanaged Itanium horribly, and got away with it only
because it is so wealthy. Stockholders have not been well served by
Intel management, at least not until Core 2.

And great technical companies have trouble if there are taken over by those
management majors you're praising. See Apple, DEC and even Intel at it's
time of little trouble.

Management is a discipline for a reason, and I don't identify classic
management types with the troubles of any of the companies you name.
For a classic management failure, look to Home Depot, where a General
Electric (known for producing sought-after management talent) alumnus
stumbled badly. It happens.

Not for the first time.

You really don't understand what a problem this is. Buyers don't want
to see problems on the balance sheets of vendors. AMD worked long and
hard to get rid of its reputation as a fly-by-night company always
teetering on the edge of collapse, and, almost overnight, it's got
that reputation back.

AMD is now just two pony ride. No one can turn it around without throwing
out much of the company.

That's what turnaround and buyout specialists are about. Without
looking, I think that AMD's current market cap is less than what they
paid for ATI. IBM doesn't want AMD out of the x86 business, but IBM
has to do things at arm's length, because IBM wants to be able to do
business with Intel (which also wants to do business with IBM). There
are people who are pretty good at these things who wouldn't know a
pole from a zero.

This last thing is just a little addition. AMD was worst performer of the
league the entire year. The stock was below 8 before the news wrt that bug
struck anyway. So they "worked" on their current evaluation whole year.

Barcelona was supposed to pull them out of that hole. And, guess
what? Markets tend to find these things out before they make the
news.

Robert.

 

[Sebastian Kaliszewski]

You are simply amazing.

Peter Lynch, the founder and long-time manager of the Fidelity
Magellan Fund and widely regarded as one of the most successful
institutional investors of all time, famously taught: "Buy what you
know."

Yet majority ignores that. It's the 'greater fool' rule which rules here.

It's true that program trading can produce fast "technical" moves in a
stock; that is to say, moves unrelated to fundamentals. Program
trading can't explain AMD's 2007 performance, and the most recent
stock price slide isn't just a coincidence with the Barcelona fiasco.

AMD's slide started aproxx. 1 year ago. It is full year of their poor
finacial performance. The performance is still poor, so what do you expect?

One does not need a technical clue to see that AMD has worst results of it's
league.

Under other circumstances, Barcelona might be a minor fumble, just as
some fumbles in football are of little consequence. When the clock is
running out and you are driving for a touchdown you need and you
fumble the ball, that's like what just happened with Barcelona.

There were similar fumbles with K8. Notice that K8 was delayed multiple
times, and clock was not up to speed initially as well.

AMD has introduced a product that I really like, something that I've
been wanting a major chipmaker to make for a long time: a GPGPU stream
processor with double precision floating point. It will turn vector
computers into commodities. Unfortunately, it will never be more than
a footnote on AMD's balance sheet. How do I conclude that? The Wall
Street Journal didn't even mention the product introduction.

Wall Street Journal is clueless about technical matters. The might be
interested if the thing either ships en masse (so there are many customers
buying) or at least AMD signed a deal with some big, important party.

Wall Street Journal and likes notice such new things when they are not new
anymore. They are ntriously bad at predicting things

And wether it will be just a footnote on AMD's balance sheet or not depends
on AMD's execution (both in dhat area as well as others), not on some
technically clueless media.

CEO's are a different matter, entirely. More and more, CEO's are
given incentives to drive stock price.

Got you! How they could 'drive stock price' if market is so smart as you
say?! Market was stupied enough to not see that technically they were up to
a disaster! They could because market is clueless about all the details.
MArket understand things at the level where there is no difference between a
box of bananas and box of electronics.

That often leads to short term
thinking and decisions that are bad for shareholders.

Yet, shareholdres are stupid enough to buy that BS.

The day you see
me quoting a CEO on a company's prospects is the day you will be
justified in speaking to me the way you have.

Those CEO's are outsmatring the marked manyfold.

You've just lost the AMD supporters.

I don't care. It's reality what matters.

Jerry Sanders placed a hugely
risky bet on Opteron, and he won.

He placed many, many such bets with mixed success. His biggest bet the
company stunt was K7 -- it was "save or die" for the company, as AMD was
never closer to going belly up than in June 1999. They had money just for
few months. K7 was on time and it was superior to what Intel then produced
and could produce.

K8 fast just a logical next step. K8 success is in majort part due to
Intel's poor execution caused by the takeover by marketing and their loose
grip with the reality.

He was a true visionary, a cowboy
almost, in his willingness to take on Intel. I wouldn't have taken
the Opteron bet for any reason, but, then, I'm not Jerry Sanders. I'm
only a techie.

One can look at Intel and see just the opposite.

That's why I wrote about Intel. They were by the time not run by technically
clueful with all the bad results of that.

Intel became
obsessed with its stock price and seemed to forget that physics do
matter, even though the company was famously founded and run by
scientists.

The scientists where then already replaced by standard stuff not seeing
difference between a box of bananas and box of electronics known to general
public as computer.

It mismanaged Itanium horribly, and got away with it only
because it is so wealthy. Stockholders have not been well served by
Intel management, at least not until Core 2.
True.


Management is a discipline for a reason, and I don't identify classic
management types with the troubles of any of the companies you name.

DEC is a typical case. Company was sold into pieces removing all the
sterngths it had. But financial results were good until the shit has hit the
fan.

For a classic management failure, look to Home Depot, where a General
Electric (known for producing sought-after management talent) alumnus
stumbled badly. It happens.


You really don't understand what a problem this is. Buyers don't want
to see problems on the balance sheets of vendors. AMD worked long and
hard to get rid of its reputation as a fly-by-night company always
teetering on the edge of collapse, and, almost overnight, it's got
that reputation back.

AMD was late with K8, with Athlon64, with 180=>130 nm transition then
130=>90 then with 90=>65 and seems to be late with 65=>45.

That's what turnaround and buyout specialists are about. Without
looking, I think that AMD's current market cap is less than what they
paid for ATI. IBM doesn't want AMD out of the x86 business, but IBM
has to do things at arm's length, because IBM wants to be able to do
business with Intel (which also wants to do business with IBM). There
are people who are pretty good at these things who wouldn't know a
pole from a zero.

I just think they are still too expensive for buyout.

Barcelona was supposed to pull them out of that hole.

So they'll have to wait for 3 monhts more.

And, guess
what? Markets tend to find these things out before they make the
news.

Markets are clueless. AMD peroformance this year is an obvious case so the
stock plummeted.

rgds
\SK