BAD HARDWARE WEEK

I went "unprepared" to my own wedding !.You know, when you don't provide "that one" necessary condition for macho man, your honeymoon can't be properly consumed.
The moral for the future is obvious: for successful Iran invasion we will need much more troops. But, lack of money will force US for more borrowing from Red China, and that will off shore even much more of our jobs in that direction, than it is already the case. That will in return make our budget even more tiny, and so on. But, after final acomplishing it (like in Iraq), new troubles will suddenly rise, in alas, nuclear equipped Pakistan. We will then again need more troops !. What about abandon of Xeon Israel only politics, it seems a beat cheaper? And might avoid final collapse like that one of ancient Egypt.Nobody ever thought about it !?.

This post link

Moreover, the light of the moon shall be as the light of the sun and the light of the sun shall be sevenfold as the light of seven days. We have then, temperature of heaven, 525°C (977°F). Temperature of hell, less than (445°F). Therefore, heaven is hotter than hell.

But, on the Doom Day chosen will go to much comfortable heaven, not to hell? Or, all the physical laws will be invalidated by then? But if so, what is effect of fire in hell, when it would stop functioning without the physical laws?

Burn in hell curse seems, isn't so bad at all?
At least, sounds better than : Burn in Heaven?
Anyway, we shouldn't forget the following (sort of the ultimate Theory of Everything):

"Those things which are impossible with men, are possible with God" - Luke 18 27, and "...with God all things are possible." Matthew 19 26.

Probably, in my humble opinion, provided there are enough faithfuls Matthew 5:3 Blessed are the poor in spirit, for theirs is the kingdom of heaven.
This post link

Finally, a truth on bad hardware.

Benchmarks show VIA C7 faster than comaparable Dothan and than Merom (when emerge).

Opterons push Intel out from top of scale, and now VIA will from bottom. As badhardware found last month.
This post link

Bush lost more than 50% of approval in 5 years !. Lets spill bean. If Israel initiated attack on Iran bring oil to $100+ price level (and no fast victory like in Iraq), this diagram shows Bush approval will fall below 20%. Lowest ever in US history, among the all Presidents. But not only in the US. Any Communistic regime (including Cuban's) had approval rating bigger than 20% !!!. Actually any regime in history had better approval. But, once (before Bush) the US were democratic country. Well don't judge the whole nation after only one man. Lets look for the other people in power there.

http://www.chron.com/cs/CDA/printstory.mpl/nation/3692292
Bush's approval rating is far below those of three of the last four two-term presidents in February of their sixth year:

• Dwight Eisenhower (64 percent);
• Ronald Reagan (63.50 percent); and
• Bill Clinton (57 percent).

Only Nixon, at 27.5 percent in February 1974 — six months before he resigned — was less popular than Bush is now.

About Bush (as the first face of US) popularity world wide, you will have to use much smaller numbers than 20%. No bought main newspapers and TV nets there, or what? Isn't that a bit weird for a victim of terroristic attacks 9/11?

Perhaps something is deeply rotten in Danish Kingdom? (Shakespear), as Badhardware correctly predicted last month. I mean not only in Microsoft financials? And how could be with such a high gas prices?

I mean people have to drive, industry have to burn oil and less money will be left for luxury like a computer and software. But, industry still support those to whom Bush is only a puppet. Long nose wodden Pinochio.

This post link

Intel in troubles with Conroe, mobo manufacturers with its chipsets.
But, AMD is not in any better performance position with AM2. More on AM2 and Intel's
new processors look at badhardware archives.

Intel’s move to push its 65nm dual-core Woodcrest, Conroe and Merom processors (all three utilize the new Intel Core microarchitecture, but target different applications) ahead of previous schedules has not much impressed industry players in Taiwan, according to industry sources. However, since AMD is considered by the industry as being a step behind in the performance race, Intel still has a good chance to regain market share recently lost to its main competitor.

On Thursday, Intel announced that Woodcrest shipments will start in June, followed by Conroe in July and Merom in August – about a couple of months ahead of previous schedules. This will definitely put additional pressure to motherboard makers, the sources indicated, even though a lot of players are expected to come to Computex showcasing new desktop models based on Intel’s new P965 chipset (which supports Conroe).

The sources expect the chipset, which is known as a part of this year’s Intel entertainment PC platforms for the digital home, to be formally launched on June 7, the second day of Computex, or even earlier.

In an attempt to catch up with the competition, AMD will probably try to move the launch of its AM2 dual-core CPUs from June 6, as previously planned, to about May 23, according to the sources. So far, these processors appear to have only minor chances to take a lead over Conroe, and this raises concerns within the industry about AMD’s competitive power, at least, in the second half of this year.

On the other hand, AMD may benefit from claims that Intel may be experiencing difficulties with the Conroe platform, with the sources mentioning possible compatibility issues with the current Intel 975X Express chipset and the unclear situation with the DDR2-800 memory standard, which is still not approved by JEDEC.

The AM2 platform seems more stable at the moment, but the lack of performance should prevent AMD from looking forward with optimism.

Motherboard makers do not believe that the new Intel schedules for Woodcrest, Conroe and Merom will heavily boost shipments. Even before the update, Intel was in a good position, targeting the new CPUs to hit the market in a hot season, the sources indicated. This quarter, according to Intel, motherboard supply will again exceed actual demand, so the sources expressed some worries about how the inventory pile-up will play out.

This post link

Electron is mistery, the biggest ever, as Einstein said once. But after the following story, who can now say that scientists aren't mad? I mean without electron there is no matter.

Bayard Peakes submitted a self-published 33-page photo-offset publication on the non-existence of the electron to present it at the next meeting of the American Physical Society (APS) for mass publication entitled So You Love Physics. Bayard's publication was reject by Karl K. Darrow, head of Bell Labs and the secretary of APS. Other scientists reviewed the publication and described it "as a pointless and elementary discussion of various aspects of physics." Bayard was outraged and planned his revenge.

Near the beginning of July, he went all the way to Bangor, Maine, and purchased a .22-caliber automatic pistol and ammunition for the gun. Eileen Fahey, an 18-year-old secretary who had worked for the APS for the past two years, arrived for work at 9:05 a.m. today with three unopened letters from her fiancé, Ronald Leo, a Marine serving in Korea. The wonderful girl with reddish-blonde hair and blue eyes sat down at her desk, opened one of the letters and began to read it. Dressed in a dark gray suit, Bayard arrived at Columbia University this morning with the intention of killing Karl and any other scientist he could find. He entered the APS office on the ninth floor of the Pupin Building on the campus and saw that only Eileen was in the office.

Frustrated that his target wasn't there, he lashed out at the only person he could. At 9:20 a.m. Bayard pulled out his pistol and emptied into Eileen's body. Five of the six shots struck her, three of them in her chest, one in her right hand and one in her right forearm. The remaining bullet lodged in the desk. He left the office and encountered another secretary for the APS, Mrs. J. V. Lumley, 32, in the hallway by the elevators. He told her, "You better call the police because I just shot somebody. You better call a doctor. I just shot a female." Bayard then walked down one of the staircases and left the building. Mrs. Lumley raced into the office and found Eileen's body on the floor, one of the letters from Ronald had been open, but the other two were still sealed. An autopsy on Eileen's body found that her heart, lungs, liver, spleen, left kidney and aorta had all been punctured by the bullets. Police initially had no leads in Eileen's murder, until they suspected that her murderer could be a disgruntled former member of the APS.

The APS supplied police with a list of members and the investigators went to work. Early Thursday morning, July 17, 1952, police arrested Bayard, 29-years-old, in Boston's Back Bay section as he returned home from a dance. He confessed, at length and without remorse, to killing Eileen on the train ride back to New York City. Bayard was wounded in Belgium during action in World War II and was discharged from the Army in 1945 as mentally ill but not in need of hospitalization. At a hearing on Friday, experts said that "establishing this thesis, and upsetting the conclusions of the great physicists of the last half century, would be a typical objective for a person with schizophrenic tendencies attributed by military psychiatrists to Bayard." He was tried and sentenced to the Rockland County Asylum for the Criminally Insane.
This post link

Chinese eat Wintel's eggs. That is the best way to say No to Wintel. Please note that Chinese offer complete Linux PC for $150 with 800Mhz Godson 2C processor. Please note that Intel's striking Merom processor will cost alone $350 -$650 when emerge. But, we can't compare those processors !! Why? Well, Merom is dual core (needed for DRM) that could run sometimes over 2Ghz, though bigger part of time it runs actually below 1Ghz, to spare some power consumption, due to its big leakage. But, Chinese don't need at this moment DRM ! Hardly Intel - Microsoft will be able to solve this problem in the future. Perhaps to announce all those who don't want to pay for it a very dangerous terrorists, und push them in Gunatanamo until they reveal its (open) sources? I mean almost half of Intel's future production will come from Israel (currently $3,5B is invested in that hughe fab) and dirty cheap PC is obvious dirty anti-semitic idea. The only problem I see is how to put among the wires some 1,5B Chinese?
This post link

Seems naturally for AMD to make before that, a safe transition to new HT 3.0 standard and FB-DIMM servers based on DDR3. On the other side, Intel will switch to 45nm by the beginning of 2008, the whole year earlier. Thus, Intel will not have enough time to make transition to a new architecture until then ( I don't count current NGMA as a new architecture). What will AMD do until 45nm emerge? It would produce 100M processors anually, some 33% of world's x86 production in 2008.

Thus, 45nm could further increase AMD's market share!? .
Or not?. Intel will introduce new architecture every 2 years. AMD will introduce high-K technology at the 45nm node, earlier than Intel. Seems like fight to death, both companies are relying on its own weapon.
What about Intel's Next generation Gesher microarchitecture?
Gesher = current project M(it)osis. (That led Jews out of oppressive Egypt to Promised Land?). My confident source? Deep throat from Intel's multicore development center at Braunschweig , Germany. Please don't miss the fact that Gesher's party David Magan was a mayor of Quiryat Gat, where is currently in construction Intel's gargantuan 45nm fab, that will too produce 32nm Geshers (bridges) in the year 2010 and after. Lets go back to Nehalem. In Hebrew nehal means a stream.
Nehelam means multi streams. Now, Gesher+Nehalem in Hebrew literally means bridge+multistream.
With transfered meaning: BRIDGE OVER TROUBLED WATERS ! . Err, link is here.

I am joking, but probably we have here not only one, but two new, competitive Intel's microarchitectures. Nehalem might be streaming architecture like Stanford's Merimac ( stream in native Indian language, but no problem for us multilinguals, hehe). Gesher? Speculative multithreading architecture, sort of functional bridge over few cores (4 at least). And who will win in the end? Well, being Gesher is from the Holy Jewish Land promised by Jahve, what do you think who will be favorized by God (Jahve)? Lets wait some 3,5 years for this hypothesis to be tested in reality. Not too far.
But, don't forget. Nobody was before Badhardware.




This post link

Stupid, idiotic question, Jesus was a Jew, thus any Jew has to understand Jews even when they use modern Hebrew. We all know that! . Well, we are all wrong this time. For the right answer we should know only something about language. Jesus has spoken ancient Arameic language.That language is still alive and some 400 000 people are still speaking it in a different dialects world wide. The following story is from 1986, the Chernobyl disaster year, curiously the same year when everyone expected to be the "end of the world" from all those time counting method for 6000 years from the genesis first day.

The lady was every afternoon sitting on the last row, at the very end of that jewish temple.
At that time she was in her fifties and nobody, among the those coming from time to time, knew her at all. Nobody, nor the rabbin himself.
Finally, three weeks after the first days she was saw the first time, the rabbin decided to approach her and very gentle he ask her:
- Good afternoon my lady, we have seen you for several days to come in the temple. Are you jew or interested about our belief? I can help you in some manner?
She was a little confused and a lot nervous, for this moment she expected and fear at once, and answered:
- No Sir, I am not, but I came here because I thought you can probably help me on a particular matter.
- What could be this? - answered the rabbin.
- I need to learn some arameic - answered her, not less confused than before.
The rabin was not nervous now, but a little bit confused due to the inexpected way the conversation turned to, and wanted to know some more details about:
- I beg your pardon mylady, but coul I know why are you interested in learning SOME arameic? - the arameic is a very ancient language, probably not spoken anymore today or in some very reduced areas.
She felt at her utmost shyness at this moment, but she replicated:
- To say Jesus "Hello" in his language when I'll be there.

Thus , if she was sure that Hebrew is more understendable to Jesus, she wouldn't bother to learn much harder available Arameic.

This post link

#5000 visits in 3 months
Not too much, but not too bad for beginning. 5000 visits in a quarter means 20K visits per year. Thus, multi person blog could reach up to 100K visits per year. Not bad result for a niche topic like hardware is. Personally, how do I feel? Is it easy to make it?


That is relative, I couldn't believe I would make it, but personally I feel now like Jesus when he had to feed 5000 men with 2 fishes and 5 barley loaves. However, I used only my 5 fingers and a pair of my eyes. And still without disciples help. : )


This post link

Nothing new at the latest AMD Tokyo presentation. Literally, except 65nm low power consumption production starting early in 2007. Nothing radically will change at AMD until mid of 2008. Then it is expected Rev2 of Core G and HT 3.0 . But, hmm I never heard for revision of revision. :)
Beside low power, we could expect low prices too. Not appealing to you? But what is more important than that, in situation when further performance rise is limited? Could AMD go beyond 3Ghz effectively?. No not at all.

Then, Intel will brake AMD? How? By its much bigger production capabilities and smaller cores, due to lower applied geometries, as all we know. Well, let check first what we actually know about that. AMD has now in reach 3 fabs, 2 of them are 300mm and 65nm. Thus, Intel has no any competition smashing production powers like it had in 2001. But, it has 65nm production and AMD has not, thus its core will be smaller and consequently its production bigger. Wrong again. AMD will start 65nm by the end of year, and no more any advantage of that type for Intel. Quite contrary. AMD 65nm SOI is much better in power consumption terms than Intel's 65nm bulk. Thus, all Intel myths are busted. Add to that possible negative effects of forthcoming anti monopoly case and dying Itanium low blows. Than, think about again.
This post link

Merom supports SSE4, as badhardware found long time ago. But do you think that mPGA 479 socket will drive announced 64bit until Santa Rosa platform chipset (that will support 800Mhz FSB speed) when arrive in 2007? I am not so sure. Beside, Merom T5600 at 1,8Ghz will use only 2MB L2. But, please note Merom's core speed: 997Mhz. Thus, there is no such a thing like a fixed clock at Merom. Its 2Ghz clock is something it hits only sporadically. So its max performance. Like Super PI25 for model T7200. Add to that Bus Speed of only 667 MHz. Doesn't seem like monster when announced in September. Of course, it will be launched earlier, but alas on paper only.
This post link

Amazing.

Here we go. My guide for another friend who queried on the same topic. Decided to add so that it didn't seem incomplete and therefore leading to wrong conclusions.

Screen Size and Resolution:
LCDs are measured from one edge of the screen to the other. CRTs are measured from the it's physical edge to the other. As a result, while LCDs are smaller, its screen size is a few notches higher.
17" CRT - 15" LCD
19" CRT - 17 to 18.1" LCD
21" CRT - 19 to 20" LCD

Do remember a 19" CRT with viewable size of 18" is of course larger than a 17" LCD.
15" LCD 1024X768
17-18.1" LCD 1280X1024
19" 1280X1024 - 1600X1200
20" 1600X1200

Take into consideration the screensize and ratio if you intend to game. The lower you get from the native resolution, the worst the game will look, even if you do have a hardware calibration unit. Even worst is when the aspect ratios are different.
If your videocard plays at 1024X768 and no more, you'll be seriously dissapointed when you game on a 18.1" monitor. Or it can be a excuse to upgrade once again.

Contrast Ratio:
Light intensity from the brightest white to the darkest black. Higher contrast ratio means better colour reproduction. The better the colours will look and be displayed, the more detail and saturation. LCD range is from 150:1 to 600:1. They average around 350:1 and 450:1. 350:1 should be the minimum.
Contrary to popular belief, while CRTs are capable of displaying more colours than a LCD, many do not. You get what you pay for.

Colour Depth:
Most LCDs can display 15.2 million colours. That's 8 bits per primary. Remember, the alpha channel doesn't really exist. Merely a blend of the 3 primaries. To decrease the response rates, some manufactors start at 262K colours then use varying voltages to output 14.9 million colours.
As far as that goes, it's not much different from what most CRTs actually display. Again, even so, it is still preferred that you stick with a high-end CRT if you demand 100% colour accuracy. Tried and true.

Response Rate:
Mostly been covered by original poster. Just to add that some manufactors will usually state their tr time. You have to add tr and tf together to get the full response time.

In addition, they will commonly state those response time given several conditions. Say 80% contrast and 50% brightness. If you want full colours, you will bring those up and response will suffer. This varies upon monitors.
Being that there are so many ways mislead, a 30ms spec'd monitor may very well ghost LESS than a 20ms spec'd monitor. When contrast et brightness are turned up, same applies.

Refresh Rate:
Ignore this. LCDs don't flicker at 60Hz. Most glass screens only do 60Hz. Few will do 75Hz. Models that state they do 75+ are 90% of the time limited by the glass screen and circuitry anyways. Either way, you won't see any flicker.

Analogue or Digital:
Digital is clearly better to the keen. On LCDs will hardware calibration, it's subtle but still there. There are problems with using analogue connection to a LCD and those are usually sync problems. This leads to "shimmering" lines and irregular sized text. You can adjust it by eye or have hardware calibration do it for you.

All LCDs have adjustments by hand. If you don't know what it is, it's fustrating and usually leads to no gain.
Hardware is harder to come by in that they come on $1000+ models generally. But you get other features that are well worth it. Most are passable. EIZO's hardware unit is a whole in a league of its own.
As far as quality.

Cables:
Not all cables are created equal. There is no such thing as a high quality bundled cable. While most of us would say it's very well good enough, it never hurts to get a new one if you experience excessive ghosting blurriness or dull colours. Do not mix and match cables as some meet the requirements for their monitor only. In that I mean, bandwidth. Higher resolution monitors require higher bandwidth. Using a cable from a 15" unit will be far from great on a 18" unit. Leads to horrid picture quality, shimmering, shaking, ghosting, et-c.

Brightness:
LCDs are all bloody bright. If you look at a unit over 250cdm^2 for too long, you're asking for trouble. Since LCDs do depend on the backlight for colour, take this into consideration. There are always gamma adjustments to make if dark areas are too dark. No different from a CRT.

Backlight:
While many manufactors use the same screen in their display, they will use different backlights. Quality will vary. Which is why, for example, NEC and Hitachi, even though they use the same panel in their 16ms monitors, the NEC has better colour.

Backlight are of very high quality nowadays. Many have a life of >40,000 hours. That's 4.5 years of 24/7/356. 9 years at 12/7/356. How many of you even do 12 hours a day?

Burn-in & Screen Fade:
Unlike CRTs, the LCD screen never degrades over time. It will be just as crisp 4 years from the day you opened it up for the first time. This makes it a very good long term investment. You can keep it att 100% contrast and brightness and it will never degrade. Do that on a CRT, and you've cut its life in half.

It is still possible to burn in an image on a LCD screen. Say, you left your desktop on for 12 hours. To rid of that, just run a screen saver the next time. If you actually did get a burn in image, again run something with motion for the same amount of time and the image will soon disapear.

Dead Pixels:
CHECK to see what the manufactor policy is on this. Some will let you return after 3 dead pixels. Some will be 7-8 sub pixels. Some will be 5 but only if they're noticable spot and/or in a cluster. If this bothers you, check before you buy.

Technology on dead pixels have improved by quite a bit from even last year. It's rare to have a fully dead pixel now. A dead sub-pixel is one that displays one colour and only that one no matter what the others are doing.
Not to be confused with a lazy pixel. These change a bit slower than the others. Usually, this is normal on a new or display that hasn't been used in a while. Use the screen for 5-10 minutes in some fast movement game and you'll find it has vanished.

Keep the screen clear of dust particles as those may block a pixel and you may see it as being dead till you wipe it.
This post link

Dell loses market share to HP, says IDC
AMD and HP ready corporate desktop push
Intel & Dell down.
PC shipments increased by 13.1 percent in the first quarter, according to research firm Gartner, thanks in part to sales in emerging markets and to consumers. Overall, 57 million PCs, notebooks and servers based on so-called x86 chips were sent to customers. Rival IDC came up with similar results and rankings although the exact numbers are slightly different.
The Round Rock, Texas-based PC maker saw shipments rise worldwide by 10.2 percent from the first quarter of 2005, resulting in a decline in market share from 16.9 percent to 16.5 percent.

Gartner further added that, in processors, Advanced Micro Devices gained market share over Intel in the last seven quarters. Before Intel's next generation paper launch: The whole 2 years.

This post link

Finally, a truth on bad hardware. Intel's cache cow gets mad.

Intel's cash is down 7.1 billion from Q105, and down $3.7 billion from Q405.
Intel went down: 8.9 billion sales (-5% on YtoY), only 1.3b net income (-38%!!!). Profit margin is extremly low, it is only 55.1%.
Expectation for Q2 is even worse: revenue is between 8.0b and 8.6, it was 9.2 Q2 2005 and it means 11% decreasing! Expected gross margin is 49%(!!), down from 56.4% (Q2 last year).
The interesting thing is that AMD sales balanced the negative impact of Intel sales, so the market was more or less flat compared to Q1 2005. But in Q2 the situation will be different, the market will go down.

Seems that no one wants Intel's beef? Nasty AMD, you Intel's market eater ! Interesting thing is that somehow $3,7B of cache has dissapeared this year, since design of new "Intel's " fab in Qiryat Gat, Israel (worth JUST THE SAME "Intel's" $3,7B) began ??? No more monopoly, no more cache? Or that "small amount" of $3,7B should be laundered first in the Middle East before the ongoing antitrust case investigation?
Is that the cause of MAD CACHE COW DISEASE at Intel's farms? Anyway, seems that sudden outbrake of CACHE COW DISEASE was quite justified reason for Intel's CEO recent fast and ferociously wild rides.

This post link

AMD requests documentation from Microsoft for its antitrust suit against Intel.

We hope that if AMD can get it, the revealing documents and email would not be ciphered and their key accidentally lost somewhere. Alas, if that happen, further chances of AMD at the court are pretty slim (see cartoon on the left). But, if Microsoft officials use the same strongness ciphering, as the ordinary Windows users have to deploy, than AMD could easily solve that pretty small hypothetical obstacle. Right?


This post link

IBM hits 8Ghz in 90nm using Limited switch dynamic logic circuits for high-speed, low-power processor designs.
That high performance logic will be used in IBM's forthcoming high clocked (up to 4,4Ghz) Power6 processor. Intel recently demonstrated 10Ghz circuits in 65nm. Thus, generally there is promising improvement space for higher clocks , but the question is would be that the most optimal choice?

But, as you can see from the graphs, there is no simulataneously low power consumption and high clock (high performance). At 5Ghz linearity between those stops, and higher system clocks are simply not economical in term of performance/power-consumption.

Interestingly, top diagram shows linearity of voltage scaling with frequency. That imply robustness LSDL for 65nm manufacturing technology and covering the whole spectrum of processors (from mobile to server ones) with one unified manufacturing technology family in the range of of 4,4 to 7Ghz.

However, all clock values beyond 5Ghz will be suitable only for high power consumption (and consequently heating) workstation or server roles. But, isn't IBM after Lenovo and Apple spinoff, just that: a server manufacturing company?

Please note that the thoughts above are quite simplified analysis, based on ring oscillator measurements, but processor itself is something much more complex.
This post link

100M AMD's 64 bit processors in year 2008 alone, through continuous transistor improvement.




This post link

Mitosis - Cell division principle in biology is applied for Intel's new multicore microarchitecture in 2009.
Replicating small cores with its own register file and speculative L1 cache, and using Global Register file+ Register Versioning Table, and version control Logic to glue multicore, boost a lot of gains in performance that might be obtained inside single die. Intel is pretty good in large L2 so, it is promising new architecture. Drawback is however, the lack of memory controller on this picture, that means still no CSI I/O for such a fast multicore !?.

Will AMD be competable by 2009?
AMD will do almost the same with its future K10 processor , that should be at disposal about the same time or somewhat earlier. However, K10 will use large integrated L3 (due to recent Z-RAM deal) that naturally fits into its SOI technology, but not into Intel's bulk. By the year 2008 Intel will use 45nm technology, that will leak a lot at higher temperatures and clock, but that will not significantly hit speculative multithreading directed, naturally lower clocked processors. And finally, Mitosis processors will be produced at Intel's 45nm fab in Qiryat Gat in Israel, currently under construction, less than 50 kilometers from troubled Gaza strip in Palestine.

If you can't wait so much, you have an option to join opened Sun T1(Niagara) multicore project that with Rock implementation, a year or two earlier, will offer very similar core collaboration concept.
This post link

Intel Server Roadmap Update (on left, click for higher resolution) given to Badhardware and less than month old, shows ambitious multicore plans, that will be supported by Longhorn R2 OS in 2009. Thus, no CSI supported by Microsoft before the year 2009 ! AMD has the whole 2,5 years to prepare K10 and HT 3.0

Microsoft Windows Server roadmap reveals Intel server roadmap too. Among the other things, Intel's New Generation Architecture is not so knew, as if it supposed supposed at badhardware. Actually, Woodrest is not Opteron killer, but only a low power blade server. Its bad I/O is not so restrictive in blade role, but benefits that offers low power processor consumption are much bigger. As you see from picture, Woodcrest will start in 2H and by the end of year finishs its DP top server role. By the end of 2008, Woodcrest will be replaced with a New Xeon Architecture.
What REALLY NEW microarchitecture will use both companies, AMD and Intel, after 2008?
Speculative multithreading.

More details will follow!

This post link

We have two big global deception going on simulateneously:
1. Collapse of US economy
2. Global warming and life extinction

Though something might be done about the first, the second one is in function of covering the first one.
Covering US economy collapse under its leadership, current US administration has simulataneously organized a big deception campaign on Global warming and its tragic consequences for life on Earth.

By the way, do you see more intelligence in the eyes of this one subject, from the top of the endangered list, more then in the eyes of some Presidents?
How we can expect our further prosperity when we destroy all we didn't create? And sometimes even that to make space for a new prosperity.
This post link

AMD64 is obviously performance broken, without any doubt !
The Three Kings (Th'Inq, Anand, Geek) follows a new performance shining star, traveling from Haifa to Jerusalem.
This one fallen, BROKEN star is actually 754 pin A64 3200+. Sorry, Intel fans, if you thought about the headline in the more general terms. You will however, have to wait to see that.


Heat spreader on chip is obviously still only a piece of metal, literally glued on to A64 die. Quite simple.
This post link

What, someones have already forgotten Tejas at 10Ghz? Well this time Intel demoed actually a register file running at 10,1Ghz in 65nm. Beside they are able to run at 4Ghz clock and VRM11's bottom voltage of 0,85V at only 42mW/Ghz. That shows actually Merom ultimate goals in 65nm (in Gilo form perhaps?). Will we see Intel's processor running at 10Ghz? No, of course not. Devil is again in the details. Intel's 65nm leaks and heat like a hell. Lower clock means lower temp. and leakage.
Thus, nothing new again, except yet another David Copperfield style show for the ignorants.
This post link

AMD has managed to obtain enough Opteron processors in the 2xx and 8xx configurations that it now has a 3GHz version, and these 256 and 856 parts are yielding some really desirable performance! We find that some recently submitted SPEC2000 scores were submitted using a Tyan Thunder K8SD Pro (S2882-D), and the results are significant. For the first time ever, a single-CPU system has been able to break the 2,000 barrier on SPECint, scoring an impressive 2,043. The SPECfp score was even better at 2,365.

To put this in perspective, Intel's highest 3.8GHz Pentium 4 670 numbers are 1,863 and 2,091 from a Precise Workstation 380, with the highest P4 955 numbers being 1,713 and 2,006.

Any comment?
What would happen with headlines of many hardware sites if Intel would be able to do it? However, at this moment I haven't noticed even single one.
But,Intel's future chips will win for sure! I swear they will, Intel told me so !!!!

This post link