All, Intel manipulates media hiding slow progress in technology.
A few days ago I run into the following message on Intel thread:
Message 5772657
"Article..Intel readies Katmai onslaught for 1999"
The post contained some interesting revelations but did not
contain the original URL. Today I tried to find the URL,
and found the following:
InfoWorld Electric:
idg.net
Intel outlines future strategy - By James Niccolai
98% of the article is word by word identical to the posted one.
However, big surprise! The phrase I was looking for:
"Intel won't say yet how fast the Katmai chips due late next
year will run, but in a demonstration here, engineers cranked
a Katmai chip up to 804 MHz -- at which point the online
banking application it was running crashed."
no longer exists in the current InfoWorld article!
It is hard to suspect
the Intel stalwart (who first posted the the whole story
on Intel thread, see above) in fabrication of these data.
The only explanation would be that Intel forced the
InfoWorld Electric to withdraw the phrase.
Apparently Intel did not like the phrase, but not because
it mentioned unpleasant crash. I think the most damaging
information is in the number itself: 800MHz. As all of
us remember, one and half year ago Intel conducted a
demo of Klamot at 700MHz. I would not speculate on what
technology it was built on, but rather will assume
that in both cases - today and 1.5yr ago, they have
demoed the best technology they managed. From this
perspective, their advances in process technology
must be pathetic - 15% during 1.5 years! I just wonder
where their denial of copper technology is leading to?
There are other pearls of marketing wisdom:
"Intel will release the successor to Merced, another 64-bit
chip code-named McKinley, which will offer twice the
performance of Merced, according to Yu."
People all over the Net are gloating about this phrase:
"twice as fast as non-existent processor, how fast it
would be?"
I take it as an acknowledge that the Merced project contains
major architectural and conceptual flaws and will never be
in mass production but only as a slow hardware prlatform
for software developers, somewhat better than their current
achievment - RTL emulation. Or just as an expensive
academic exercise :)
Another revelation:
"After his presentation, Yu said it is "too early to say"
if the Katmai New Instructions will be used to beef up
multimedia performance in Intel's Celeron chips."
The meaning of this is not clear. It may mean that they
are actively working on this problem but are not very
successfull - I guess the KNI/later take silicon space,
L2 cache takes space - not a good combination for their
current low-density process. Sorry Yousef for your
sloppy job: "make Idsat so", eh?
|
