Topic: AMD and IBM claim 24 per cent chip speed breakthrough

[F9thDigi]

A new faster version 'strained silicon' has been developed by IBM and AMD. The companies are claiming that the resulting transistors are up to 24 per cent faster than existing processes.
The companies say that the new strained silicon process, named `Dual Stress Liner,` improves the performance of both n-channel and p-channel types of transistors by stretching the silicon atoms in one transistor and compressing them in the other.

As the speed increase occurs without any increase in power, the partners foresee a performance boost without incurring the kind of problems with heat dissipation which dogged Intel's attempts to increase the clock speed of its Pentium designs to 4GHz.

As an added bonus, dual stress liner can be introduced without massive investments in new production techniques and can be integrated into commercial production lines rapidly.
AMD says that it aims to introduce the new strained silicon technology into all of its 90nm processor platforms as soon as the first half of 2005. Among the chips to include dual stress liner technology will be multi-core AMD64 processors.

Further details will be announced at this week's IEEE International Electron Devices Meeting in San Francisco.

[Sirgod]

That is very cool, I'm Glad to hear this esp. after the 10Ghz problem that IBM had earlier this year.

Good Post Digi.

stephen

[F9thDigi]

THE LATEST INTEL ROADMAPS seen by the INQUIRER indicate that the launch of its dual core desktop processors - the X20, the X30 and the X40 - will be launched in the third quarter of 2005, suggesting they won't proliferate in PCs until the fourth quarter.
These chips will two Pentium 4 dies in one package, each with 1MB of L2 cache, and supporting Intel's EM64T 64-32 extensions.

At the same time, it appears that Intel's 3.73MHz Pentium 4 is being pushed back to a February launch, will have 2MB of L2 rather than L3 cache, and will coincide with the introduction of the 6XX Pentium 4 series. These chips are numbered 630 (3GHz), 640 (3.20GHz), 650 (3.40GHz) and 660 (3.60GHz) but the 670 (3.80GHz) won't be introduced until the second quarter of next year.

At launch the 660 will cost about $600, the 650 $400, the the 640 $273 and the 630 $224. The 3.73 1066MHz 2MB processor will cost the re-assuringly less than $1000 sum of $999 when it's launched in February. When Intel rolls out the 670, this is being priced at a not inconsiderable $850. These chips will have 800MHz system buses, 2MB of L2 cache, EM64T and enhanced Speedstep inside.

Intel will perform a difficult juggling act as it hopes to be able to sell the existing 5XX range for much of 2005 as well. They'll be cheaper, but when you buy a machine, you may have a hard time distinguishing whether the chip is a 6XX or a 5XX. Hopefully Intel will introduce a branding scheme to make things easier for consumers. Prices for dual core desktop chips are expected to be announced around the beginning of February.

The dual core chips have now got official Intel names - Glenwood will be called the 955X Express, while Lakeport G and Lakeport P will be called the 945G and the 945P. The 955X Express is due to launch in Q2 next year. That's when the 945G arrives as well.

Intel will lower prices on its Celeron 533MHz chips on the 20th of February, with the 3.06GHz/533/256K thing costing $117, the 2.93GHz/533/256K costing $102, the 2.8GHz/533/256K costing $89, the 2.66GHz $79 and the 2.53GHz $73. The 2.4GHz Celeron, also known as number 320, will disappear then.

Intel will make changes to chipset prices on the 26th of December, with the 915G dropping two bucks, and the same thing for the 915GV, P and 910GL. More chipset cuts are slated for the 3rd of April next year

[F9thDigi]

That is very cool, I'm Glad to hear this esp. after the 10Ghz problem that IBM had earlier this year.

Good Post Digi.

stephen

Thanks Stephen

<rubs hands together>

This is all looking very nice for my new machine I have planned next christmas!!!! Dual Core here I come!!!

[Nemesis]

Thanks Stephen

<rubs hands together>

This is all looking very nice for my new machine I have planned next christmas!!!! Dual Core here I come!!!

Dual core AND Dual CPU Here I come.  I already have 2 dual CPU machines.

Actually this is one advantage that is more in AMDs court than Intels.  Intel has been getting part of the effect of dual cores from Hyperthreading, adding the 2nd core removes the Hyperthreading function so the speed boost of the 2nd core is lessened.  AMD however has not had the hyperthreading feature so when they add the 2nd core the speed boost should be bigger - especially on anything that Intel ruled the roost in because of the Hyperthreading speed boost.

I suspect  that the integrated memory controller in the Athlon64/Opteron lines will give them an advantage as well until Intel duplicates that feature.

One thing that I am waiting to see is a performance comparison between AMD and Intel versions of the AMD 64bit architecture (called EM64T by Intel). 

[F9thDigi]

Dual core AND Dual CPU Here I come.  I already have 2 dual CPU machines.

Of course!!!  These machines will be so fast.

Actually this is one advantage that is more in AMDs court than Intels.  Intel has been getting part of the effect of dual cores from Hyperthreading, adding the 2nd core removes the Hyperthreading function so the speed boost of the 2nd core is lessened.  AMD however has not had the hyperthreading feature so when they add the 2nd core the speed boost should be bigger - especially on anything that Intel ruled the roost in because of the Hyperthreading speed boost.

Very good point.  Although after my original AMD took 10 minutes to draw a simple cube in 3D Max I'm staying well away.  I just can't justify the cost of getting it wrong.......

[FPF-SCM_TraceyG_XC]

What about if each core in a dual package also has hyperthreading? Would that simulate a quad core package?

[Nemesis]

What about if each core in a dual package also has hyperthreading? Would that simulate a quad core package?

It would, but the most current reports that I have read say that the hyperthreading function won't exist in dual core Intel chips.

[Tus-XC]

probably because right now most people wouldn't be able to take adavantage of it.  I think xp home edition (yuck) can only support a max of two cpus and a duel core hyper thread would probably not agree with it (not sure about any others, its the only one i can remember hearing a problem with).  I doubt there would be many similar problems, but its the only one i can think of off the top of me head.

Though i'm going to bet that they will probably have ht in them when they are finally released, or at least a few months after release.  I don't think intel would throw away that just because of duel cores, why give up a good advanatage

[F9thDigi]

Well I know that XP Pro on my work machine reads the 2 Xeon 2.8Ghz w/HT as four processors.

[Nemesis]

Dual Core article on C|Net

Chips with four cores and eight cores will eventually join dual-core chips. The company's research department is also looking at the feasibility of creating chips with hundreds of cores to assist servers and supercomputers with large numbers of relatively repetitive calculations, said Steve Smith, vice president of the desktop platforms group at Intel.

Intel rival Advanced Micro Devices has said it will create chips with two, four and eight cores, too. Its first dual-core chip will also come out in 2005.

Meanwhile, ClearSpeed has developed a 96-core chip for augmenting supercomputers, similar to chips from Japan's RIKEN and others. Nonetheless, Intel is the world's largest chipmaker, so the vast majority of the population will experience dual-core computing through its products.

The first dual-core Intel chip for desktops, code-named Smithfield, will come out in desktops in 2005. Its first dual-core notebook chip, code-named Yonah, will begin shipping late in 2005 and likely won't appear in notebooks in any great numbers until 2006. A dual-core version of Itanium, called Montecito, will also debut in 2005 while Tulsa, a two-headed Xeon for servers, is set to come out in the first quarter of 2006.

Still, while software development may be fairly straightforward, software licensing for dual-core and multicore processors has proven trickier, especially in servers. Despite Intel and AMD recommending that companies charge fees by the number of processors each server contains, regardless of how many cores each processor includes, some companies' policies dictate charging by the number of cores each processor in a given server contains.