April
12 2004:
Welcome Back,
Its been quite a while since my last report. ( I have
written a 3 part article for Mixdown in the down time , which you can
read here)
. The original plan was to hold off until Intel got their shite together
and officially release Prescott, so I could get a better grounding on
how the battlefield was shaping up. After originally expecting Prescott
to land late last year, I wasn’t planning on such a long hiatus
between drinks, but alas, not everything goes to plan when we are dealing
with the technology circus.
So here we are !!
The first section of this article covering Prescott
has been published in Mixdown –March Edition. I have updated and
extended the article to include our friends from Cheapzilla and Fruitzilla..
Enjoy :-)
Taking a leaf out of AMD's book, Intel finally paper
launched the .09 Micron Prescott Series of PIV on February 1st. Well
lets just say that after all of the hoopla that the tech community had
stirred up, I can only say a collective yawn was all that most of us
could muster. Sparking memories of the first Willamette PIV’s,
Prescott is proving slower, and hotter than the current Northwood PIV’s
(and minus the rumored 64 Bit extensions) Snore!! It looked like Chipzilla
had struck out, big time. but alas, all was not as it seemed.
On Feb. 16 Intel officially announced their CT (Clakamass Technology,
sorry, what was that, “ Crack In My Ass “?) 64 bit extensions
for both the new Series of Xeon and Prescott PIV ranges.
What About Cheapzilla? Well despite having the performance advantage
for the last 3-4 months, AMD have idled their way thru the New Year
without really being able to consolidate the momentum that they had
garnered with the 64-bit Athlon 64/ Opteron range. Not to be out done,
our old favorite Fruitzilla (Apple) have also been feverishly banging
the 64bit drum, mind you, without really having anything to show for
all of the noise they have been making.
Now with the Evil Empire also on the 64 bit desktop wagon, it’s
going to be very interesting to see what spin the 3 Tribes are going
to try on the unsuspecting public.
Ah Yes, the Circus is definitely back, .. Buckle Up…
Intel Prescott: Is It Getting Hot In Here?
Is been a long road for the PIV from the Dog .18 nm “Willamette“
Series to the latest revision .09 Prescott. Many thought the transition
to the latest offering would be the typical scenario of a die shrink,
pumping the cache, and bumping up the clock speed. This in the past
has netted Chipzilla with a smaller, cooler and faster chip than the
predecessor. Well, not this time.
So what did we get? On Paper Launch Day, Prescott was released at speeds
of 2.8 to 3.4 Ghz, along side 3.4 GHz Northwood versions, both Standard
and Extreme Editions. The last of the line!
*Update * “ Its been over 8 weeks since Paper launch, and
the only Prescott to hit the showroom floor has been the 2.8 Ghz slug
“
In regards to Prescott, Intel promised next to nothing with this chip,
and on the surface, it looks like they managed to deliver. :-) Sure
we’ve got the die shrink from .13 to, 09 nm, we have double the
L1 and L2 cache, 13 new instructions (whhooo-eee SSE3 .. Yawn), improved
branch prediction, etc,etc. All good I hear you say, unfortunately we
also have an Instruction pipeline that has increased from an already
huge 20 stages, to an elaphantile (is that even a word?) 31 stages.
Say What!! Man that’s going to do wonders to the IPC (Instruction
Per Clock) NOT ! That coupled with increased latency in the caches,
and a thermal heat dissipation over 103 Watts, and we have on our sleeves
a possible tail wagger of the highest order. Yep this puppy is slower
clock for clock than the current Northwood, and runs a good 5 –10
degrees hotter. Whoooa, hold me back. Not really much to entice the
tech community to hop aboard.
Something isn’t quite right here though. Despite
the 50% extra pipeline length, performance is not that far behind the
Northwood, thermal performance aside. The other interesting thing is
that as the clock speed goes up, the clock for clock performance delta
decreases. There is definitely some technical wizardry hiding under
the surface, and as the clock frequency passes the Northwood’s
3.4 Ghz, we will see Prescott draw away dramatically. There is a bit
of “ smoke and mirrors” surrounding this chip at the moment.
The current 478 pin paper launch pimp is going to have the shortest
life span of any chip Intel have ever released, with the 775 pin LGA
socket version expected in about 2-3 months. Why was this dog was even
let out of its cage? * Update * “ 775 release now shifted
back to June –July”
Its obvious that for Intel to keep scaling their Processor
lines to the expected 5 Ghz +, that some major architectural changes
would have to be made, but these clock speeds are planned for the next
series of processor (Tejas). Another interesting question is, what are
all of those millions of extra transistors that have been crammed into
the new die, that mysteriously don’t seem to be doing anything
at the moment ? Hmmmm. And by the way, where the hell are those rumored
64-bit extension? Hold the phone. Bling !!!! Intel is obviously testing
the water with some newer technologies that they plan on introducing
as they see fit. Lets not forget Hyperthreading was implemented all
along, but was dormant until almost 18 months into the PIV’s life
cycle.
And Then There Where 3:
Enter CT, Clakamass Technology, formally known as Yamill,
Intel’s 64 bit extensions for the X86 platform.
* Update * Now official known as EM64 T * . On Feb. 16th Intel
finally confirmed that they were introducing 64 bit extensions on the
new series of Xeon (Nacona) processors that are due next quarter, with
the Prescott getting the instructions shortly after. This coincides
with the introduction of the 775 pin LGA version of the chip and the
new Grantsdale and Alderswood chipsets. Surprise!!
The question is still raging whether the instructions are 100 % compatible
with AMD’s “ AMD64 “ instructions, with Intel being
pretty coy whether they have actually used the core x86-64 code developed
by AMD, or whether they have an extended code that is at least somewhat
compatible. Most analysts agree that the CT code is in most respects
identical to AMD’s, but due to an agreement signed years ago by
the 2 companies to share technologies, Intel has no obligation to admit
or announce that they are in fact using the competitors code, which
frustratingly for Cheapzilla, isn’t allowing them to gain any
brownie points in the hype department. In the end, all that is important
is that the 2 technologies are compatible with Windows XP-64/2003 -64,
which from all reports, they are.
I covered the 64 Bit scenario last month, so I won’t
repeat myself here, suffice to say that by the time a 64 bit O.S, driver
support and applications are considered mainstream, all of these shenanigans
will be a distant memory. We are still talking years, not months, so
all of this 64-bit nonsense is little more than posturing for the foreseeable
future. Hopefully Chipzilla can reign in the thermal problems they are
having with the current Dog 478 Prescott, and re launch the 775 LGA
version with some decent clock speeds and thermal spec’s, or we
could be seeing our future systems doubling as microwaves. * Update
* With the reported delay of Windows 64 for both AMD64 and EM64T, now
not being expected until very late 04 or early 05, It is no surprise
that Intel have announced the 64 bit extensions for Prescott will not
see the light of day until 05. Interestingly Intel announced the delay
weeks before Microsoft ??, So who is actually driving this Bus ?? Certainly
not poor old Cheapzilla, with their AMD 64 campaign not really meaning
diddly until the O.S see’s the light of day…conspiracy theorist
can line up to the left .. :-)
You can read thru some review fodder here,
here, and here
|
AMD 64 :Lets play the Socket
Game
Now despite all of the hot air and hyperbole generated by the oh yeh
faithful in regards to the mighty Intel capitulating and using the
x86-64 code developed by AMD, Cheapzilla have not really managed to
ride the ensuing wave. They have made some major inroads into the
Server markets, and rightfully so, with the Opteron being a formidable
force , especially in those markets that can utilize the benefits
of a true 64 bit O.S , i.e. Linux . On the Desktop however, AMD are
still battling in the 32 bit arena, so the gains are not as easily
garnered.
Also like Intel, AMD are introducing a Socket change for the Athlon
64 from 754 to 939 pin. Not to be confused with the 940 Opteron and
Athlon FX units. There is some confusion on whether the new slugs
will have 1 MB of L2 Cache, or whether they will have only 512, like
the lower end Athlon64 3000+ that was introduced late December 2003.
The new chips have been upgraded to dual channel memory like the Opteron
and FX variants, but will not need to run ECC registered Ram. With
the added Dual Channel memory support, its going to be a lot harder
to justify the” rebadged Opteron “ Athlon FX Range, as
punters will have similar performance with out the extra expense.
I’d suggest the possible shift to smaller L2 Cache would be
to more to do with justifying the continued existence of Athlon FX
range , than any architectural change. With the frustrating delay
of Windows 64, AMD have really no choice but to bide their time, and
hope to hell Chipzilla don’t sort their Prescott woes out too
quickly.
Apple : Life & Times in the RDF !!
Ah Fruitzilla, Our friends over at Apple have not really been having
a good run of late, not that you would know, judging being their over
active hype department. Since the release of the G5 , the avalanche
of shite piled on the unsuspecting public in regards to the alleged
performance of the new Super computer, has unceremoniously landed
back in the fruity boys lap, not that anyone in the tech community
is too surprised. If getting their advertising banned in certain sectors
of the globe for ”false advertising “ wasn’t enough,
Apple have also been dragged thru the mess for the alleged
64 bitness of the their beloved OSX –Panther . That mixed with
the growing reports of quality problems with the new baby, lawsuits
in regards to iPods and we have some very , very interesting fodder
to sort thru.
Shall we ?!! :-)
Its been a good 9 Months since the official launch of the G5, and
in that time the original avalanche of debate regarding Apples performance
claims has slowed to a whimper in Tech circles. This is due to the
claims being repeatedly proved as being nothing more than Wishful
thinking on Apples part . The continued claims of being the “Worlds
fastest DeskTop Computer “, and the related benchmarks to prove
that claim , is proving not only to be a constant source of amusement
to the tech community, but has also landed Steve and his boys in a
bit of bother.
In November , the U.K’s Independent Television Commission decided
that Apple's claim that the G5 is "the world's fastest, most
powerful personal computer" did not meet their standards for
truth in advertising., and demanded that Apple cease and desist in
all future advertising. As a general rule, advertising claims in the
UK have to be provable, and Apples work of fiction in regards to proof
didn’t quite cut it. Although no official statement was made
in regards to the ads in Australia, Apple also tempered their claims
here as well. The Aussie website now boasting “ One of the Worlds
fastest Desktop computers “.
Hmmm doesn’t quite have the same punch :-)
To Ad insult to injury, Even in the good ol USA , where you can claim
just about anything without any repercussion, Now, after receiving
a "tip" from Dell, the National Advertising Division (NAD)
of the Council of Better Business Bureaus has recommended that Apple
cease making such claims, and they also targeted Apple's claim to
offer the "world's first 64-bit processor for personal computers."
Ah it was fun while it lasted. Apples response ?
Apple has said that the ad campaigns in question have effectively
run their course, but they promise to keep the NAD's concerns close
to their heart. Hmmm, lets see here, lets put out some absolute bullshite
to misrepresent the actual real world performance of a product, and
then without retracting any of the claims, promise to be more mindful
in the future… LOL… Blind Monkeys !!! Thankfully, I’
haven't met anyone still falling for this shite in quite a while..,
but keep banging away there Steve, I’m sure they're out there
:-)
The G5’s have also been plagued by quality problems , ranging
from outright hardware failures , to noise problems with Pro Audio
equipment. The Noise problem , which is described as “ introducing
a hum and a rhythmic chirp to an audio signal path resulting in these
noises coming out of studio speakers “, is one that is causing
the most grief in the Pro audio circles for obvious reasons. The “Fix
“ is not as clear cut as Apple would have us believe. A replacement
PSU has resolved the problem for some, but others are still being
affected, and the issue just seems to be being pushed further and
further afield by our fruit loving friends. More here
The other question being raised is in regards to the scaling capabilities
of the the G5. There has not been a speed bump since its release,
with only a Dual 1.8 Ghz Model being introduced late last year to
replace the original Single Processor 1.8 Model. ( Probably due to
the fact that the Single Processor G5’s are being pummeled by
the X86 variants and even by the G4 DP offerings ) Now lets not forget
that there is in actuality only one G5 processor despite the 3 speed
offerings.. Say What..?? O.K , let me spell it out 1.6 GHz –800
FSB, 1.8 Ghz –900 FSB, 2.0 Ghz – 1 Ghz FSB. All of these
systems feature an IBM 970 CPU with an identical clock multiplier,
that is then multiplied by the FSB to achieve our final speed rating.
Get It ? Now Stevie boy was beating his chest about 3 Ghz variants
within 12 months,, hmm wishful thinking again ? There are some new
models expected shortly, but they have already been delayed for unknown
reasons, but the whisper is that heat could be playing a part. However,
just to prove how deep the RDF has gone, the next series of IBM 970,
the FX ( .09 Micron- 2.5+ Ghz ) , which technically is yet unannounced
,has already won the Microprocessor Report Analysts' Choice Award
for Best Desktop Processor. Now there's one for credibility LOL..
Following on from my 64 Bit article , I apparently struck a few nerves
with some of the Fruit Fraternity, and have subsequently suffered
some fallout for my “ so called” claims that OSX is not
a 64 bit O.S. Well, just for those so inclined, read it and weep..
I don’t have the energy to even argue the “ fact ",
I’ll let some of the Monkeys at Fruitzilla do the talking for
me ..here,
here,
oh and here..
In regards to the iPod fallout, seems the battery life on the new
slimline models are proving to be dramatically less than Apples claims,
surprise, surprise. This has incensed some hapless end users to launch
a lawsuit claiming unfair competition, false advertising, fraudulent
concealment and breach of warranty. Who Apple ???? Nah, who would
have believed they would be capable of such things !!?
O.K , that’s it , I think I’ll go have a lie down now.
Till next time :-)

|