Tuesday, July 31, 2007

AMD shows Phenom X4 CPU operating at 3GHz

AMD shows Phenom X4 CPU operating at 3GHz, running with standard cooling. To be available only at fall 2008. Just as Bad Hardware Week deduced two weeks ago. By the end of this year X4 Phenom will be available mostly at 2,4 Ghz.

AMD's 16 cores Bulldozer

Well have to be be pretty "communicative" to keep them bussy all. Croatian magazine Vidi found
Intel's 8 cores MCM bussy only 20% of processor time.

Friday, July 13, 2007

Intel Nehalem in H2 2008

Nehalem offers 300% performance per watt increase. Then, Intel's current core seems pretty pale compared to Nehalem envisioned in 2001. Because it though didn't reach 10,2 Ghz in 2005. And Nehalem push into H2 2008 (it was scheduled for 2009) might be a sign of forthcoming AMD rise.
However, Intel started with Nehalem development in 2002 and 6 years after that seems like a bunch of problems not completely resolved. Lets wait how Intel will do it. Intel's last all in one solution to all problems was well known under the name Prescott in 2004.
Intel has revealed some information on their Nehalem processor in the latest roadmap updates. The 45nm Hi-K Nehalem-EP is based on 4-issue Intel Core microarchitecture technology, supports Simultaneous multi-threading and Multi-level shared cache architecture (L2 and L3). Nehalem-EP processor will arrive earlier than expected in H2 2008 and will be pairing up with Tylerburg chipsets. EP stands for Efficient Performance, one of the new naming for the server platforms for energy efficency on 1-2 processor sockets. There are also Mission Critical (MC) for best RAS from 2-512 processor sockets, Expandable (EX) for RAS from 2-32 processor sockets as well as Entry (EN) on 1-2 processor sockets to better align with future IT usage. LGA1366 and LGA715 are used by Nehalem

Microsoft renames XBox trade mark to FBox

Microsoft renames XBox trade mark to FBox. Is this Xbox return queue?
Fbox is probably abbreviation after Faulty Box, because all of 11,6M delivered X boxes are seems faulty.
I like this story part the best:
Microsoft estimated that the repairs will cost the company US$1.05bn to US$1.15bn
Thanks The Inq for the link
. However, seems that The Inquirer revealed previous finance
securing operation here.
I am asking myself what about some hidden bugs in Intel's processors? In much more pieces than in Xboxes? Intel is not obliged towards its customers, or what? Find some quietly dumped patches and thats all. What after possible common sue of its customers after constant delivery of F serie Duo cores?

Monday, July 09, 2007

AMD K10h will hit 3Ghz by fall 2008

But, Intel tells employees there’s a bigger fish to fry than AMD.
Independent analyst Brookwood said he believes AMD will be able to raise the chip's frequency by 10 percent a quarter, to about 3 GHz by fall 2008. Some 200 Mhz per quarter? Please note some fuses at picture above. If some cores don't pass test, their fuses will be blow out, and product will be still sold as a low performance 2 corer. What do you think, could Intel do that? Even in 45nm? Intel has no HT implemented, and it now uses only 2 separately tested core dies, so have to glue them together on MCM. And it will stay so until Intel's CSI emerge.
Beside, Northbridge is incorporated in K10h, so its power consumption is much lower than Intel's one.
And that kind approach makes 4 core chip manufacturing prices lower than at Intel. Thus, lets prices go down.
By the way, I remember well when Northbridge chip alone costed $31. Today you can purchase Sempron 3200+ for that money. Add some Linux atop and you will get dirty cheap firewall, router or whatever you want. Especially if you are from China. If China is strategic market for Microsoft, why shouldn't be for Intel and AMD? Ie. todays tantalum capacitor 3300 microF costs $250. 8 times more than the cheapest single core Sempron. !
Some say AMD is doomed. Looking up to the end of 2008, I am not so convinced. And in 2009 low priced Fusion chips will follow. Seems Intel will be in bigger troubles , after constant margins eroding. I mean its products are after one year worth only 33%, just like AMD's. AMD is in financial troubles, Intel will be soon.
Nieztche once said: What doesn't kill me, will make me stronger. Currently, seems than Intel only cures AMD, not kills it. Beside, why AMD can now successfully lower its prices? Because now, low end processors Chartered fab started its volume production. Previously AMD had no general solution to Intel's strategy. But, lets wait and see the final price war outcome.

Seems quite good before AMD 45nm appearance under TMSC agreement in 2Q08. The 65-nanometer Barcelona will have to compete with Intel's upcoming 45-nm Penryn, which will also ship this year.

Wednesday, July 04, 2007

What Intel's customers actually need?

This. Let take a look.
However, if they want low priced buggy $300- quad core, this way , please.
I ask myself how will Intel feed its shareholders with lower prices and lower market
share ,when Barcelona start?
But, road towards success starts from garage work below, up to the stars.

Monday, July 02, 2007

Intel's Core Duo processors are buggy as hell !

Date: Wed, 27 Jun 2007 11:08:16 -0600

Various developers are busy implementing workarounds for serious bugs
in Intel's Core 2 cpu.

These processors are buggy as hell,
and some of these bugs don't just
cause development/debugging problems, but will *ASSUREDLY* be
exploitable from userland code.

The most interesting quote (after the latest Microsoft QUIET patch) is this:
Open Source operating systems are
largely left in the cold.

Full (current) errata from Intel: here.

In a statement from Intel Global Communications, Nick Knuppfer writes:

“Months ago, we addressed a processor issue by providing a BIOS update
for our customers that in no way affects system performance. We
publicly documented this as an erratum in April. All processors from
all companies have errata, and Intel has a well-known errata
communication process to inform our customers and the public. Keep in
mind the probability of encountering this issue is extremely low."

OMG, I wondered why Intel's X86-64 (contrary to AMD's) has extremely low
support to RAS (reliability, .. servicebility)
Now, I know why. And why AMD was late, contrary to Intel.
AMD simply has lower bug threshold, compared to buggy Intel.
Ie. Intel should have at least some 134 bugs NOW and AMD perhaps comparable
number in the first incarnation of Barcelona. However, AMD will make
things better in Q4/07 or 1Q08 with new K10h revision.
But, Intel will make response 45nm PERFORMANCE based response by then, that would
rise their bug menagerie number to respectable 200+ uniqe, exotic bugs.
And all of it you will have INSIDE your new INTEL beast processor.
Enjoy, but without me.

Scientists find drug to banish bad memories

EDIT: However, those drug helpless, and those who would like to sue after SCANDAL being
by this Intel's official claim, please let comment this article.
Millions of people gives
at court neccessary, hmm WEIGHT to their arguments.

AMD supercomputer World 2nd and 3rd, Intel's at 12th place

From the latest TOP 500 list, June 27th. AMD supercomputer is World 2nd and 3rd, Intels starts only from the 12th place.
But, with 420 Tflops Sun Constellation supercomputer based on Barcelona, AMD will attack IBM's first place in November TOP 500 list.
Intel at TOP 500? It is good in lower performance systems area. That should be enough verdict on RAS and FP performance for both companies.

AMD Barcelona CONFIDENTIAL 2,3 Ghz projected performance

AMD Phenom X2 GE at 2,3 Ghz; 45W TDP, socket AM2+, HT3 3200 Mhz?
Its main rival is Athlon X2 at 2,7 Ghz. Seems that Phenom is in
average 14% faster on the clock to clock basis. However, with 2MB L3 enlarged core, 14% is not some breakthrough to me.
Though 45W TDP definitely is. Using single core it will not consume more than 25W.
Enough for mobile use, that is on unstoppable rise.

AMD Barcelona CONFIDENTAL 2,3 Ghz projected performance, click on to enlarge.
They claim FP and DB superiority.
Howerver, AMD will need faster versions 2+ Ghz of Opteron, due out only in the fourth quarter, to do damage to Intel , the lastest channel register recon finds.

But, isn't fat ass 6MB L3 in 45nm the sign of diminishing processor returns, like at the Intel, or a neat signature of the stellar high performance?

This page is powered by Blogger. Isn't yours?