Wednesday, December 20, 2006

Post-silicon era in sight?

In recent edition of Technology Review online magazine, a group of MIT researchers are reported to claim that they are succeeded in making and operating a new non-silicon transistor with only 60-nm gate length but with better performance than a 65-nm silicon transistor.
The transistor is made of indium-gallium-arsenide as substitute of silicon and indium-aluminium-arsenide as substitute of silicon-dioxide as gate material. Jesus del Alamo explained that wirh this compound material the Moore's law can be maintained in the future. Electrons can flow faster in the material than in silicon, provides a basis for better shrinking in smaller transistors.
The question is, how much is needed to convert the whole design and production lines which are currently used for silicon-based transistor?

Sunday, November 26, 2006

Bali and Jepara IT-supported industry?

Everyone should have known Bali as it is the world's best island for many straight years. Everyone who is involved in high-class exotic furniture also should have known Jepara, which exports US$ 112 million valued products annually. But doesn't anybody think that both tourism and furniture industry need to modernize themselves with high technology approach?
To keep those industries running, nowadays IT is badly needed. Some good info websites are already available. In tourist area like Bali, a centralized hotel and travel booking system will help tourists and the Balinese themselves a lot. A good online Bali map which guides the tourists everywhere with GPS supported device will also be very helpful for tourists who want to conquer the island conveniently. And of course with a good data system, Indonesian security bodies can recognize as early as possible any unwanted threats.
For Jepara, a good IT system can help them to forecast quickly wood supplies and to collect number and specification of the orders enough time before the deadline. This will result better resources allocation and will avoid either overpricing or underpricing. Many ERP and CRM software products are available in Indonesian market, including some developed-in-Indonesia like ones produced by Zahir Accounting and Mitrais. The latter has even a development site in Bali. There are a lot potentials in developing high tech industry in Indonesia, also because the government has launched Indonesia, Go Open Source! (IGOS) program and in the same time also improves relationship with Microsoft :-)

Tuesday, November 21, 2006

Mobile phone market boom to be expected in Indonesia

According to Yahoo! News, Telkomsel, one of Indonesia's leading mobile service provider, is targetting to get 7 million new subscribers in 2007. Therefore it is expected that Telkomsel will have 41 million subscribers at the end of 2007, whereas in the end of 2006 it is projected that the subscriber number will reach 34 millions. It is roughly 50% Indonesian market. The market growth in 2007 is expected as much as 15 million subscribers. Having its 35% shares owned by SingTel, Telkomsel is ready to invest approximately US$ 326 millions within 3 years for building its 3G network, in which Siemens has won 70 million euros project.
Indonesia is the largest market in South East Asia, and also a huge mobil phone market. Nokia is planning to launch its new TV-mobile product N92 in Indonesia and Vietnam, probably in November. EMTEK will operate DVB-H network needed for N92 probably with TV programs from subsidiary SCTV. But I can imagine this can be a new trend in Indonesia - especially in the big cities where middle class people stay - as many people should work a whole day and it will be small entertainment for them to watch TV in the middle of traffic jam, for example.

Friday, November 17, 2006

Quad core from Intel

Just read this announcement: Intel proudly released two day ago its first quad core architecture, specially designed for servers, high-end gaming machines , and workstations. They boasted that this new architecture offering 50% faster performance than the current dual core architecture and four times than its single core Xeon. The quad core Xeon 5300 server processor series, which was codenamed Clovertown - has speeds in range of 1.6 GHz and 2.66 GHz, power consumption between 80 and 120 watts, and it is priced US$ 455 for 1,000 units order. The other one is designed for high-end client Core 2 Extreme QX 6700 costs US$ 999 and has speed up to 2.66 GHz with power consumption up to 130 watts, which is also approximately the same amount as single and dual core's needs in a chip. Dell, IBM , and HP are among the server makers who are ready with the new Intel release.
Intel has done restructuring last September with layoff of some 10,500 employees worldwide, aimed to save US$ 3 billions a year by 2008. This enables Intel to make a "great leapfrog" from single-core, dual-core, to quad-core architecture in a year.
Critics from AMD say that Intel's architecture with two dual-core chips onto a single piece of silicon "is less elegant" than AMD's "native" four-core design. That should have some impacts with performance and power efficiency. But as Intel is now ahead with the architecture, it has a better chance in the market as it is not likely that the user would wait until 2007 when AMD will release its quad-core chip.
Both designs from Intel and AMD are reported to be made up from 65 nm technology. Both also have a plan to move from 65 nm to 45 nm manufacturing process in the next one or two years. Smaller technology node promises more transistors in the same area of chips which means more functionalities can be built. The problems with smaller node are mainly power dissipation per unit area, signal integrity, and lithography limit.

Tuesday, November 14, 2006

GSM vs CDMA

"What? Why do you still ask who is the winner of the battle", you may ask. But it is not finished yet, even when the manufacturers are considering which standards would be implemented in their products.
After 2G technology fades away then comes 3G from both standards: GSM launches WCDMA (wide code division multiple access) and CDMA tops its CDMA2000. They still compete each other, even though GSM claimed that WCDMA occupies 48% market in Western market, which is traditionally CDMA's stronghold. But CDMA2000 is being implemented also in some traditional GSM providers like in Indonesia. Sometimes CDMA, CDMA2000, and WCDMA terms in Indonesia can be very confusing. But Indonesia is one example, how these standards are competing a huge market.
3G can not be claimed as the most advanced mobile technology. There is already 3.5G technology, popular as HSDPA (high speed downlink packet access), and even 4G technology has been waiting in the front of your door. The HSDPA has been launched last September in Indonesia, at least by the Excelcom (XL) provider, Indonesian provider owned 60% by Telekom Malaysia. This technology promises up to 2Mbit/s wireless data transfer. It is approximately as much as standard DSL connection in Germany (even though some German providers have offered some cheap high speed DSL - 16 Mbit/s).
Meanwhile Telkomsel, another Indonesian mobile phone operator, has awarded German Siemens to help building its 3G WCDMA infrastructure. Telkomsel has already also some contracts with Nokia and Ericsson made before, but now Siemens is getting this 70 million Euros project. Telkomsel is believed to have around 30 million costumers in Indonesia.

Saturday, August 19, 2006

My old article about nanotechnology

It is dated March 5, 2005 in Indonesian newspaper Pikiran Rakyat.

Ini Era Semikonduktor Nanoteknologi

BANYAK orang merujuk pada hukum Moore, dengan jumlah transistor dalam suatu rangkaian terintegrasi (integrated circuit/IC) akan menjadi dua kali lipat dalam 18 bulan. Ramalan ini terbukti dalam produk-produk IC sehari-hari.

Dulu kita kenal prosesor Intel seri 4004, hingga 286, 386, 486, Pentium, dan hingga kini Pentium 4. Jumlah transistor pada Intel 4004 tahun 1971 "hanya" 2.300 transistor, sementara pada Pentium II tahun 1997 kita mendapati 7,5 juta transistor di dalamnya. Pentium 4 malah mengandung 55 juta transistor di dalamnya.

Jangan terkejut jika dalam waktu dekat akan ada mikroprosesor yang mengandung 1 miliar transistor. Tentu saja jumlah transistor sebesar itu menunjukkan, ukuran setiap transistor itu amat kecilnya. Kalau tahun 1970-an transistor berukuran lebih dari 1 mikrometer (1 mikrometer = satupersejuta meter), maka kini perusahaan-perusahaan semikonduktor berpacu di kisaran 100 nanometer (1nano meter = satupersemiliar meter), dengan kemampuan yang makin meningkat pula.

Jika kecepatan proses mikroprosesor di tahun 1970-an hanya 0,1 MHz, kini Pentium mampu bekerja di kisaran 1 GHz. Benar-benar luar biasa hasil rekayasa ini. Meningkatnya unjuk kerja itu amat berkaitan dengan berkurangnya ukuran transistor . Berkurangnya tegangan (voltase) yang dibutuhkan setiap transistor, dan saat yang sama mengurangi waktu tunda (delay time), yaitu beda waktu antara masukan dan keluaran dari sebuah rangkaian transistor, jelas memperbaiki kemampuan dalam bentuk frekuensi yang disebutkan di atas.

Aplikasi kompleks

Mengingat jumlah transistor yang makin banyak pula, dimungkinkan terwujudnya aplikasi-aplikasi kompleks. Telefon genggam yang dilengkapi dengan video adalah salah satu contoh yang tidak terbayangkan tahun 1970-an. Akan tetapi timbul masalah baru karena ukuran transistor yang di bawah 100 nanometer itu. Dengan jumlah transistor yang mencapai puluhan juta dalam ruang yang sempit (hanya beberapa sentimeter persegi), maka jumlah total daya yang dibutuhkan juga meningkat sangat tajam.

Sebagai gambaran, prosesor Intel Pentium membutuhkan daya 100 watt, padahal tahun 1970-an daya yang dibutuhkan sekira 0,1 watt. Diprediksikan dalam tahun-tahun mendatang sebuah mikroprosesor akan membutuhkan 500 watt, bahkan skenario paling buruk bisa menjadi 18 Kilowatt.

Kepadatan daya ini per sentimeter persegi bisa menyamai nosel roket atau reaktor nuklir. Untuk itu, dibutuhkan terobosan baru agar perkiraan itu tidak terjadi. Inilah salah satu tantangan industri semikonduktor modern, yaitu mengurangi disipasi energi transistor. Memang mengecilnya ukuran transistor berefek positif, yaitu meningkatnya arus dari sumber (source) ke penguras (drain) pada saat transistor menyala (on). Pada gilirannya mengakibatkan berkurangnya waktu tunda (delay) yang berbanding terbalik dengan frekuensi. Akibatnya frekuensi meningkat yang berarti unjuk kerja transistor, dan rangkaian terintegrasi secara keseluruhan meningkat. Akan tetapi, dengan makin tipisnya lapisan oksida pada gerbang (gate) transistor, arus “bocor” melalui gerbang bisa jadi meningkat. Hal ini masih ditambah dengan fenomena kapasitansi parasitik, yaitu rugi energi yang diakibatkan kapasitansi yang muncul sebagai parasit dalam rangkaian, baik pada transistor maupun pada kawat (wire) yang menghubungkan satu rangkaian transistor dengan transistor lainnya.

Ada beberapa cara untuk mengatasinya. Desain transistor SOI (silicon on Insulator) meletakkan insulator antara gerbang dan substrat, sehingga mencegah arus bocor. Atau, dengan menjadikan material yang berkonstanta dielektrik tinggi pada gerbang yang juga mencegah mengalirnya arus bocor gerbang yang dapat menjadikan kegagalan operasi transistor.

Desain FinFET, yang menambah permukaan kontak gerbang untuk mengontrol arus nyala (on). Atau, bisa juga dengan memperbaiki material silikon menjadi silikon renggang (strained silicon) sehingga memperkecil tahanan (resistance).

Tantangan lain muncul di bidang litografi. Akibat ukuran transistor yang akan dicetak berukuran di bawah 100 nanometer, maka dibutuhkan teknologi yang bisa mencetak dengan panjang gelombang sekira 100 nm. Ini adalah salah satu titik kritis pembuatan transistor berukuran nano, karena bisa berakibat pencetakan terjadi tidak sempurna.

Kini muncul berbagai teknologi baru seperti teknologi penenggelaman (immersion technology) 193 nm dengan sumber laser tertentu, dan teknik OPC (optical proximity correction). Tantangan-tantangan ini memunculkan harapan akan munculnya generasi transistor Terahertz (satu triliun Hertz). Tinggal kita harus memilih. Akan jadi penonton, pengguna, atau pemain?***

Estananto,
Alumni TF-ITB. Bekerja di perusahaan desain semikonduktor.

Sunday, June 25, 2006

Indonesia was in the World Cup!

If somebody asks if the Indonesians were present in the Soccer World Cup, then the answer is yes. But they were not under red-and-white flag and Indonesia Raya national anthem, but under red-white-blue under the name "Dutch East Indies". It was in 1938, Indonesians were the first Asians playing in the World Cup.
After proclamation of independence, Indonesia still had a good soccer team. In 1958 qualification matches they beat the Chinese but then didn't proceed because Indonesia refused to play with Israeli team. After some political turmoil till now Indonesia never even qualifies itself to come to the first round of the World Cup.
Agusman Effendi, acting chairman of the PSSI (Indonesian Soccer Association) when answering when Indonesia is able to come in the World Cup said that the time is approximately around 2018. From now it is 12 years more to come. Is it not even too optimistic?
We need more investment in open spaces, in standard soccer fields, in talent training camps. Otherwise, let Indonesians dream when they wake up only to watch their television in the late night - to watch the World cup teams compete - which none is theirs.

Monday, January 02, 2006

Bandung High Tech Valley

According to Bandung High Tech Valley (BHTV) blogsite, Synopsys, one of leading EDA supplier, gave ITB rights to use its tools in the value of US$ 155 million. EDA costs is once main problem of semiconductor design industry. Sehat Sutardja, Indonesian-origin Marvell CEO had delivered speech about how important this issue is.
Will Bandung or nearby city Cimahi be the next Bangalore or Penang?
Well, it depends how ITB can build its infrastructure to be able to make good, efficient, and reliable semiconductor design industry. A design center needs also brain power and computing power besides EDA tools. I am sure universities in Bandung and Cimahi can supply the needed engineers, but experienced engineers must be hired up also. ITB must be able to be international enough, and supported by other universities around, can be center of excellence as alternative to Bangalore and Penang. Can ITB catch up the newest technology, e.g. be a partner of Synopsys in 65 nm technology development? Can Bandung or Cimahi local government give some incentives for investors and entrepeneurs in this field?