AMD vs Intel: - Tes 50 CPU Part5

• ISTILAH PENTING

64-Bit: Standar 32 Bit lambat laun akan diganti oleh 64 Bit. CPU baru ini mampu memproses 8 Byte (64 Bit) dalam satu cycle (putaran). Dengan demikian, kalkulasi yang rumit dapat dikerjakan lebih cepat lagi.

Dual-Channel: Istilah ini menunjukkan komunikasi antara CPU dan RAM. Dua modul RAM dikontak secara paralel oleh CPU sehingga memungkinkan bandwidth 2 kali lebih besar.

Dual-Core: CPU terdiri dari 2 inti prosesor, sehingga dapat memproses 2 langkah kerja (threads) secara paralel.

Proses produksi: Pengembangan lebih lanjut dari teknologi produksi memungkinkan penempatan transistor yang lebih banyak pada bidang yang sama. Keterangan ukuran seperti 90 nm menunjukkan lebar struktur CPU.

FSB/Systembus: FSB menghubungkan CPU dengan chipset dan komponen lainnya. Pada CPU AMD, systembus berfungsi dengan bandwidth 2 kali FSB, sementara pada Intel 4 kali FSB. 

Hyperthreading: Logika kontrol CPU digandakan. Dengan 2 prosesor virtual, 2 buah thread juga dapat diproses secara paralel. 

L1-/L2-/L3 Cache: Memori sementara prosesor untuk data, perintah, dan hasil. Biasanya, semakin besar jumlah chache, semakin cepat pula kinerjanya.

Stepping: Berbagai stepping menunjukkan perbaikan-perbaikan kecil dalam teknologi produksi prosesor.

Mekanisme hemat energi: Konsumsi daya dan clockspeed CPU diturunkan. Teknologi AMD disebut Cool n' Quiet, sedangkan teknologi Intel adalah Enhanced SpeedStep (ESS)

Non-Executable-Bit: Pemisahan RAM menjadi bagian data dan bagian program code. Dengan ini, eksekusi code berbahaya dalam bagian data dapat dihindari.

• KESIMPULAN

AMD dan Intel berpacu ketat dalam teknologi CPU Dual-Core. Intel menang tipis dengan memperkenalkan model pertama untuk kelas Desktop, beberapa hari lebih cepat dari AMD. Namun dari sisi kinerja, AMD lebih unggul. Athlon X2 menduduki kedua tempat pertama dalam tes CHIP kali ini. Produk unggulan FX-57 mencapai posisi ke 3. Kinerja yang tinggi dengan konsumsi daya yang rendah merupakan platform tepat untuk PC tenang. Selamat AMD! Tampaknya Intel masih harus memperbaiki diri.

Para profesional sebaiknya menggunakan prosesor terbaik Athlon X2 4800+ (No.1). CPU ini tidak hanya gemilang pada rendering 32 bit, tetapi juga pada aplikasi 64 Bit yang mencapai peningkatan kinerja sebanyak 30%.

Untuk PC Game, Athlon 64 FX-57 (No.3) adalah pilihan terbaik: Kinerjanya 35% lebih cepat dibanding Pentium D 840 EE. CPU ini juga mampu memenuhi tuntutan game 3D terbaru.

Harga PC Pribadi yang terutama digunakan untuk selancar, berkirim e-mail, chatting atau memutar MP3 akan jauh lebih murah. Dengan harga sekitar US$270, AMD menawarkan Athlon 64 3500+ (No.17) dan Intel menyediakan Pentium D 820 (No.18). Seri untuk soket lama ini setara dalam kinerja maupun harga. Sayangnya, motherboard yang tersedia tidak mendukung PCI-Express.

Murah untuk PC Ruang Duduk. Pemenang Tip-Value, Sempron 3000+, (No.38) sudah bisa diperoleh dengan US$85 dan menawarkan cukup tenaga untuk semua aplikasi seputar TV & HiFi. Sistem yang tenang juga dimungkinkan berkat teknologi Cool n' Quiet

Di ambil dari CHIP 11/2005

AMD vs Intel: - Tes 50 CPU Part1
AMD vs Intel: - Tes 50 CPU Part2
AMD vs Intel: - Tes 50 CPU Part3
AMD vs Intel: - Tes 50 CPU Part4

AMD vs Intel: - Tes 50 CPU Part4

• PENINGKATAN KINERJA HINGGA 30% DENGAN 64 BIT

Software 64 Bit Pertama Akhirnya Siap

Sistem operasi Microsoft kini tersedia dalam versi 64 Bit. Dipastikan selanjutnya akan segera tersedia driver dan software 64 Bit. Aplikasi pertama untuk 3D profesional sudah tersedia, Cinema 4D (aplikasi 3D rendering). Software ini memanfaatkan RAM besar yang disediakan Windows XP x64, registry serta perintah 64 Bit CPU.

Peningkatan kinerja lewat teknologi 64 Bit dapat dilihat pada hasil uji Cinebench 2003 dari Maxon Cinema 4D. Dengan Athlon X2, aplikasi 64 Bit ini mencapai kinerja 30% lebih tinggi dibanding aplikasi 32 Bit. Sementara dengan Pentium 4/670 kinerjanya lebih cepat 20%.

 

Aplikasi 64 Bit memang masih jarang, tetapi sudah banyak yang berada dalam stadium Beta, antara lain program burner Nero dan browser Firefox. Untuk game, ada patch 64 Bit untuk Far Cry di http://amd64andfarcry.gamedaemons.net.

Aplikasi 32 Bit juga berfungsi di Windows XP x64, kecuali yang sangat terkait hardware seperti virus scanner atau tool system. Untuk itu, Symantec Antivirus Corporate Edition V10.0 dan Avast 4.5 sudah mendukung 64 Bit.

Windows XP x64 saat ini tersedia dalam versi System-Builder seharga US$130, artinya belum ada support melalui telepon dari Microsoft. Apabila Anda telah membeli Windows XP Professional antara tanggal 31-03-2003 dan 31-07-2005, Anda bisa menukarnya dengan biaya kirim US$20. Keterangan lebih lanjut bisa disimak di https://microsoft. productorder.com/clientx64/default.aspx.

Apabila Anda ingin pindah ke Windows XP baru, Anda harus memastikan bahwa semua driver yang dibutuhkan sudah tersedia dalam versi 64 Bit. Pilihan lain adalah dengan membeli PC baru dengan sistem operasi 64 Bit, misalnya dari Dell.

• Intel Pentium M

Mobile CPU untuk PC Desktop

Berbeda dengan prosesor Intel lainnya, prosesor notebook Intel Pentium M sangat hemat energi. Pada prosesor Desktop, semakin tinggi clockspeed semakin banyak panas yang dihasilkan. Sudah lama Intel menghadapi masalah ini. Bahkan model Desktop seri 6xx yang dilengkapi mekanisme hemat energi SpeedStep, masih tetap panas.

 

CPU notebook memiliki instalasi hemat energi yang canggih dan arsitektur prosesor yang lebih efisien, menjamin power-loss yang rendah. Dengan demikian durasi baterai lebih lama. CPU tersebut juga dapat digunakan pada Desktop untuk mendapatkan PC yang tenang.

Banyak prosesor, sedikit motherboard
Intel juga menawarkan solusi lengkap untuk segmen notebook. Penomorannya meniru sistem desktop: Pentium M dinamai 7xx, P4 mobile dengan 5xx dan Celeron M dengan 3xx. Boxed CPU hanya tersedia Pentium M dan Celeron M (7xx dan 3xx). Dengan demikian terdapat kurang dari 20 prosesor Mobile dengan kisaran harga US$80-675.

Untuk motherboard dengan soket 749 M, pilihannya lebih sedikit lagi. Saat ini hanya ada 2 platform di pasaran dan keduanya dari Aopen. Board pertama untuk CPU mobile ini adalah i855 GMEm-LFS. Seri selanjutnya adalah i915 GMM-HFS dengan chipset baru Intel Sonoma. Board terakhir ini sudah mendukung PCIe dan prosesor FSB 533 yang lebih cepat.

Produsen MSI menyusul awal tahun depan dengan mainboard soket 479M. Untuk solusi praktis, saat ini sudah beredar modul board adaptor dari Asus. Modul ini memungkinkan penggunaan prosesor soket 479M pada mainboard Asus dengan soket 478 dan BIOS khusus. Kelemahannya, mekanisme hemat energi SpeedStep tidak berfungsi. Akan tetapi dibanding P4 biasa, Anda masih dapat menikmati penghematan daya sebesar 80 W.

Bagus tetapi tidak murah
Pengujian dengan Pentium M 770/2,13 GHz menunjukkan bahwa kemampuan kinerja platform Mobile sebanding dengan salah satu sistem PC standar masa kini (prosesor 3 GHz). Namun konsumsi dayanya jauh lebih rendah, hanya 61 Watt (onboard graphics) posisi standby dan 78 W saat beban penuh. Sistem desktop P4 rata-rata membutuhkan daya 2 kali lebih besar.

Mekanisme penghematan Intel dan kendali kipas yang aktif memungkinkan sistem PC yang tenang. Kelemahannya, CPU ini harganya cukup mahal, sekitar US$675, motherboardnya US$250. Dengan setengah dari jumlah dana tersebut Anda bisa mendapatkan sebuah sistem AMD dengan kinerja setara yang juga tenang.

Gusdiharto.Pratomo@chip.co.idThis e-mail address is being protected from spam bots, you need JavaScript enabled to view it (NO,TL)

AMD vs Intel: - Tes 50 CPU Part1
AMD vs Intel: - Tes 50 CPU Part2
AMD vs Intel: - Tes 50 CPU Part3
AMD vs Intel: - Tes 50 CPU Part5 

AMD vs Intel: - Tes 50 CPU Part3

Gaming PC: Kinerja Optimal

Untuk urusan gaming, ukuran FPS (Frames Per Second) adalah segalanya. Semakin besar angkanya semakin baik. Agar tampilan grafik nan menawan tidak terhambat oleh prosesor, Anda harus menggunakan CPU yang kencang. Dari sisi kinerja, terutama pada game 3D, saat ini Athlon FX-57 yang terbaik: dengan hasil 86 FPS pada uji Unreal Tournament, CPU top terbaru AMD ini unggul 30 frames (53%) lebih cepat dibanding model high-end Intel Pentium Extreme Edition 840.

Tidak hanya kinerja CPU-nya saja yang meyakinkan, sebuah PC berbasis FX-57 lebih hemat energi 120 Watt dibandingkan sistem Intel dengan konfigurasi yang sama. Dengan demikian pada pertarungan "terpanas" pun PC berbasis AMD FX-57 tetap "cool" dan bekerja lebih tenang.

Teknologi Dual-Core maupun Hyperthreading tidak terlalu dibutuhkan dalam segmen ini. Gamer jarang menjalankan aplikasi lain ketika bermain. Tenaga CPU sedapat mungkin dikerahkan seluruhnya untuk game 3D.

Platform ideal untuk AMD FX-57 adalah mainboard dengan chipset NVIDIA nForce 4 SLI. Dua graphics card yang tergabung dalam SLI menjamin tenaga 3D yang sangat besar. Untuk sebuah PC game high-end, Anda harus rela merogoh dana cukup besar. Harga CPU-nya saja mencapai US$1.000, belum lagi graphics card dan periferal lainnya.

PC Pribadi: Kinerja dan Harga merata 

PC dalam segmen ini biasanya digunakan untuk berselancar di Internet, e-mail, memproses foto-foto, meng-edit video keluarga, dan mungkin juga sesekali bermain game. Anda tidak perlu mengeluarkan banyak uang untuk teknologi yang mahal. Di segmen ini, rasio antara harga dan kinerja yang seimbang adalah hal terpenting.

CPU "umum" yang CHIP sarankan untuk jenis soket keluaran terbaru adalah Athlon 64 3500+ dan Intel Pentium D 820 (kisaran harga US$250). Prosesor di atas kedua tipe ini harganya sekitar US$400 dan kinerjanya hanya sedikit lebih tinggi.

Pada kelas ini, CPU AMD 64 Bit cemerlang dengan kinerja game yang bagus dan konsumsi daya yang rendah, dua elemen penting untuk sebuah PC yang tenang. Prosesor Intel dengan Dual-Core berada di depan dalam aplikasi standar, multithreading dan multitasking, menarik bagi pengguna yang sering menjalankan beberapa aplikasi sekaligus.

Platform ideal untuk segmen ini berbasis chipset NVIDIA nForce 4 Ultra untuk CPU AMD. Beberapa di antaranya sudah menawarkan FireWall terintegrasi, solusi keamanan praktis untuk pengguna rumahan. Sementara untuk CPU Intel, motherboard yang tepat adalah yang berbasis chipset Intel terbaru 945P atau  945G bagi Anda yang menginginkan solusi onboard graphics.

Dana yang Anda keluarkan bisa menjadi lebih hemat bila PC tersebut hanya digunakan untuk mengetik atau mengerjakan hal-hal yang sederhana. CPU hemat seperti AMD Sempron 2600+ atau Intel Celeron D 325 merupakan prosesor-prosesor murah (sekitar US$70), tetapi kinerjanya relatif lamban. Untuk video encoding misalnya, dibutuhkan waktu sepertiga lebih lama dibanding prosesor-prosesor yang disebutkan sebelumnya.

PC Ruang Duduk: Bukan yang tercepat, tapi Paling Tenang 

Perkembangan teknologi yang begitu cepat dewasa ini telah mendorong penyebaran PC dari ruang kerja ke ruang duduk. Sayangnya, kebisingan PC masih menjadi kendala. Oleh karena itu, pilihan CPU yang tepat bisa menjadi solusi. Untuk merekam dan memutar format video dan MP3, prosesor high-end tentunya terlalu berlebihan. Penurunan clockspeed pada beban rendah memungkinkan PC bekerja lebih tenang.

Untuk segmen ini, lebih baik gunakan CPU AMD Sempron 3000+ dengan mekanisme hemat energi Cool n' Quiet dibanding CPU Pentium 4. Alasannya, mekanisme Enhanced SpeedStep Intel yang terintegrasi dalam seri 6xx tidak mampu menyaingi teknologi yang sama dari AMD. Idealnya, PC ruang duduk menggunakan motherboard dengan chipset NVIDIA nForce 4 (tanpa Ultra atau SLI) dan sebuah graphics card berpendingin pasif, seperti ATI Radeon X300SE atau NVIDIA GeForce 6200TC. 

Tabel Perbandingan 50 CPU 

AMD vs Intel: - Tes 50 CPU Part1
AMD vs Intel: - Tes 50 CPU Part2
AMD vs Intel: - Tes 50 CPU Part4
AMD vs Intel: - Tes 50 CPU Part5 

AMD vs Intel: - Tes 50 CPU Part2

Dalam tes yang dilakukan CHIP, arsitektur AMD terbukti lebih unggul. Namun, Intel juga berencana membuat CPU Dual-Core mendatang dengan konsep yang sama dengan AMD.

64 Bit: Inovasi yang paling menggebrak dari pengembangan prosesor adalah implementasi 64 Bit. Dari sisi hardware, sejak dua tahun lalu, AMD sudah siap dengan solusi CPU 64 Bit pertamanya. Sayangnya, CPU ini belum didukung Windows di segmen Desktop. Baru setelah Intel juga menawarkan CPU 64 Bit dengan seri 6xx, industri prosesor mulai berubah. Sejak Mei lalu, Windows XP versi 64 Bit sudah tersedia dan dapat digunakan pada CPU AMD maupun Intel.

Dari sisi software, Extended Memory 64 Technique (EM64T) Intel ini sama dengan yang dipakai AMD. Teknologi ini adalah perluasan konsep 32 Bit dengan fungsi-fungsi tambahan sehingga software 32 Bit masih dapat digunakan.

Registry dan perintah-perintah 64 Bit antara lain dapat mempercepat penanganan bilangan yang membutuhkan tempat lebih dari 32 Bit, termasuk format "Long-Integer" dan "Double-Floating-Point" yang digunakan pada aplikasi ilmiah.

Dengan teknologi 64 Bit, memory address yang dapat dikontak juga meningkat 16 Exabyte (praktiknya hanya sekitar 1 Terabyte). Sebanyak 32 memory address maksimal hanya mencakup 4 GB. Kelebihan konsep teknologi 64 Bit dapat Anda baca pada halaman berikut.   

Hemat energi: Untuk PC yang tenang, prosesor harus hemat energi. Pada CPU high-end terbaru, daya panas yang dihasilkan tiap bidang yang setara dengan uang logam Rp 25, sama dengan daya bola lampu 100 Watt! Pendinginan yang dilakukan dengan kipas menjadikan PC bersuara bising.  

Mulai Sempron 3000+, AMD menangkalnya dengan mekanisme hemat energi Cool n' Quiet yang menurunkan clockspeed CPU pada beban rendah. Permukaan prosesor menjadi tetap dingin dan sistem lebih tenang. Mulai seri 6xx, Intel melengkapi CPU-nya dengan mekanisme serupa, Enhanced SpeedStep (ESS). Namun, solusi ini kurang efektif. 

• PROSESOR DAN SOKET

Apa Arti Kode Penomoran?

 

Keluarga prosesor AMD meliputi Athlon 64 FX, Dual-Core X2, Athlon 64, dan Sempron. Anda bisa membandingkannya berdasarkan peringkat dalam nama (misalnya 4000+). Nomor 2 digit pada model FX juga menandakan kecepatan, semakin besar semakin cepat.

Perbedaan teknis antarkeluarga produk kini tidak lagi terlalu jelas. Dulu, dukungan dual-channel hanya untuk seri FX dan Cool n' Quiet hanya untuk Athlon 64. Kini keduanya memiliki semua feature tersebut. Spesifikasi tepatnya bisa Anda lihat pada nomor OPN yang ada pada CPU (atas). Anda juga dapat melihatnya di alamat www.oc-inside.de/index_e.html bagian "Workshop" atau di www.amdcompare.com/ USB-stick-encoder/desktop.

 

Sejak sekitar setahun ini, CPU keluaran Intel tidak lagi menyandang clockspeed sebagai nama produk, melainkan nomor model 3 digit. Misalnya "Pentium D 840", disamping keluarga produk ("8xx" berarti Dual-Core) juga menjelaskan clockspeed, frekuensi FSB dan ukuran L2 cache CPU tersebut. Pentium 4 Hyperthreading dengan L2 cache 2 MB dinamai "6xx", dengan L2 cache 1 MB "5xx". Prosesor desktop Celeron disebut "3xx", sementara CPU high-end mendapat atribut "Extreme Edition".

Pada prosesor Intel, spesifikasi terpenting terpatri di permukaan prosesor (lihat atas). Selebihnya, keterangan lain bisa Anda peroleh dari nomor sSpec di alamat http://processorfinder.intel.com/scripts/default.asp.

Soket prosesor: Untuk CPU keluaran terbaru mereka, AMD dan Intel menawarkan CPU desktop untuk 2 tipe soket: Athlon X2, 64 dan FX memakai soket 939 (gambar 1), Sempron terbaru dan model 64 generasi pertama menggunakan soket 754 (gambar 2). CPU Dual-Core Intel, P4 dan Extreme Edition memakai soket 775 (gambar 3), sementara P4 lama dan beberapa model Celeron masih memakai soket 478 (gambar 4). 

 

• Saran CHIP

Pemenang tes CHIP tidak harus sama dengan pilihan Anda. Tentukan pilihan berdasarkan kebutuhan Anda masing-masing. Untuk membantu pilihan, CHIP akan menunjukkan prosesor mana yang ideal untuk para profesional, gamer, workstation, dan ruang duduk.

Workstation: Tenaga Besar untuk para Profesional 

AMD Athlon X2 48000+ adalah prosesor tercepat yang sempurna bagi pengguna profesional. CPU ini mencapai kinerja tertinggi pada benchmark aplikasi dan 3D-rendering. CPU kelas atas AMD ini menawarkan kinerja 5% lebih banyak dibanding produk unggulan Intel Pentium D 840 Extreme Edition pada benchmark Cinebench dan bahkan unggul 15% di PCMark04.  

AMD benar-benar menunjukkan kekuatannya di sisi teknologi multi-prosesor. Pengguna profesional sering menjalankan beberapa program dan proses sekaligus. Urutan kerja CPU (threads) dibagi kepada 2 core prosesor dan dikerjakan secara paralel. Pada single CPU, tugas dikerjakan berurutan sesuai antrian.

Software yang mampu mengelola beberapa threads ini sudah tersedia di pasaran. Antara lain program rendering 3D Studio Max atau Maxon Cinema 4D. Judul terakhir merupakan salah satu aplikasi pertama yang tersedia dalam versi 64 Bit. Platform yang tepat untuk CPU kencang ini adalah mainboard dengan chipset NVIDIA nForce 4 Ultra atau nForce 4 SLI (untuk penggunaan 2 graphics card).

AMD vs Intel: - Tes 50 CPU Part1
AMD vs Intel: - Tes 50 CPU Part3
AMD vs Intel: - Tes 50 CPU Part4
AMD vs Intel: - Tes 50 CPU Part5 

AMD vs Intel: - Tes 50 CPU Part1

Bingung memilih prosesor yang tepat buat Anda? Teknologi apa yang ada di dalamnya? Prosesor aktual dari AMD dan Intel akan diulas pada artikel ini!

Membeli prosesor sekarang tidak semudah dulu. Dulu, Anda tinggal menyebutkan angka GHz. Sekarang, clockspeed tidak lagi digunakan sebagai bagian nama prosesor. Belum lagi, produsen menambahkan jargon-jargon teknologi canggih seperti Dual-Core, 64 Bit, Cool n' Quiet, Enhanced SpeedStep (ESS) dan Non-Executable-Bit. Jika tidak ingin salah beli, Anda harus memahami makna istilah-istilah teknis tersebut. Asal tahu saja, singkatan "D" dalam nama CPU Intel yang baru bukan berarti “Dual-Core”, melainkan “Desktop”. Penamaan gaya baru ini semakin memperumit pilihan. Prosesor Model Numbering seperti pada Intel atau AMD Rating memang tidak mudah dicerna. Apalagi menurut daftar harga resmi saat ini, produsen menawarkan sekitar 50 prosesor yang berbeda! Angka ini di luar prosesor untuk server dan notebook.

Saran Belanja sesuai Penggunaan
CHIP tidak hanya mengungkap arti nama prosesor, tetapi juga menyarankan yang sesuai untuk tujuan penggunaan. Pada artikel ini, kami menguji kinerja dan konsumsi daya 50 CPU terbaru. Chip mana yang tercepat, baik untuk Intel dan AMD, akan ditunjukkan dalam tabel pada halaman akhir.

Tidak semua pengguna membutuhkan tenaga processing yang tinggi. Oleh karena itu, CHIP merancang 4 profil pengguna: Workstation berkinerja tinggi, PC game yang cepat, PC pribadi dengan rasio harga/kinerja yang seimbang, dan PC multimedia untuk ruang duduk yang tenang.

Dalam artikel ini Anda juga akan membaca saran-saran kami untuk CPU dan platform yang tepat. Dalam bagian berikut akan kami jelaskan terlebih dulu teknologi yang ada di balik Dual-Core, 64 Bit dan mekanisme hemat energi CPU Intel dan AMD.

• Teknologi Baru

Dual-Core: Adalah pencapaian teknologi mutakhir pada CPU AMD dan Intel. Dua core (inti prosesor) ditempatkan pada sebuah CPU untuk meningkatkan kinerjanya. Setiap core ini sebenarnya tidak lebih cepat dibanding CPU biasa dengan clockspeed yang sama, tetapi semua proses perhitungan dibagi kepada 2 inti prosesor tersebut.

Istilah yang kerap dipakai adalah multithreading. Aplikasi yang dirancang untuk ini menugaskan langkah-langkah kerja (threads) tertentu kepada setiap core prosesor yang dikerjakan secara paralel. Peningkatan kecepatan yang bisa diperoleh mencapai 70%.

Pada aplikasi yang tidak dirancang untuk multithreading, peningkatan kinerja baru terasa bila beberapa aplikasi dijalankan sekaligus (multitasking). Misalnya ketika Anda bekerja dengan program word processor, sebuah tool AntiVirus bekerja di latar belakang untuk melindungi pekerjaan Anda dari serangan virus. Hal ini serupa dengan Hyperthreading pada Intel. Bedanya, konsep Hyperthreading menggunakan 2 prosesor virtual.

Walaupun prinsip Dual-Core pada AMD dan Intel sama, realisasinya berbeda. Saat ini Intel masih memotong 2 core prosesor (Die) tunggal dari wafer semikonduktor, dua CPU Pentium dengan masing-masing L2 cache. Kedua cache baru disatukan dalam paket Dual-Core.

Sebaliknya, AMD sudah menggabungkan kedua core tidak terpisahkan dalam sebuah Die. Melalui Hypertransport Protocol, keduanya dapat langsung berkomunikasi seperti komponen pada chip Northbridge. Sebuah protokol khusus digunakan untuk menjamin pengelolaan L2 cahe secara lebih efisien. Selain itu, cache-controller terintegrasi memungkinkan akses data langsung dalam RAM.

AMD vs Intel: - Tes 50 CPU Part2
AMD vs Intel: - Tes 50 CPU Part3
AMD vs Intel: - Tes 50 CPU Part4
AMD vs Intel: - Tes 50 CPU Part5 

CPU Overclocking Basics

by Vince Freeman: January 1, 2001

[Note: There are inherent risks when performing any overclocking exercise, such as long term CPU degradation, serious heat issues and a chance that peripheral cards can be damaged. Also, many manufacturer's warranties may be voided if you attempt to overclock your system. Users need to understand and accept the risks before attempting this procedure to avoid damage to your PC and its components. - ED]

Introduction

Over the past few years, the practice of CPU overclocking has become more and more popular and what was once the domain of the technically adept has graduated to a more mainstream audience. Overclocking your CPU brings inherent risks, as many users will attempt to overclock their processors without the proper information and techniques necessary to do the job safely and proficiently.

CPU overclocking is actually a very simple concept, and is basically an attempt to get a higher level of performance from the CPU than was originally intended. Just like finding a nice suit on sale for half price, overclocking a Celeron 566 to 850 MHz offers a whole lot more bang for your buck. Since the CPU speed affects all aspects of computer performance, this added performance can drastically improve game framerates as well as application speeds.

By perusing the many CPU overclocking articles, it is not difficult to pick out the prime CPU contenders. Knowing which CPU to buy is only half of the equation, with the actual implementation and procedure being the integral step to CPU overclocking.

To that end, this guide will explain the basic information regarding processor and motherboard features, as well as examining the basic process and techniques of overclocking. Since individual systems can be very different not only in terms of features, but also how these features are presented, hard and fast overclocking steps are difficult to set. Instead, this guide will present a basic overview of CPU overclocking, consistent with different platforms and processors, and is intended for the novice or intermediate user. 

Know Your Motherboard

The first step to overclocking success is getting to know the features and capabilities of your motherboard. The easiest way to do this is to read the motherboard manual and see if the board supports overclocking, and if so, which method it employs. To make sure the feature list is up to date, check the manufacturer's website for any new manual or BIOS revisions. If in doubt, spend some time in some of the USENET overclocking newsgroups, or those areas specific to your motherboard manufacturer. Even reading a product review may shed some light on motherboard features you may be unfamiliar with. Here is a list of the basic motherboard features that need to be identified:

1. Front-side bus speed manipulation:

This feature allows the user to increase or decrease the FSB (front-side bus) speed from the CPU default. Many ABIT, ASUS and MSI motherboards offer FSB increases in 1 MHz increments, while standard boards may only offer a small selection of speeds. Some motherboards may not even allow any type of FSB tweaking, and even for those that do, the number of available FSB speeds can differ dramatically between models. The FSB speeds may be selectable through the system BIOS, or physical jumpers may need to be changed to enable the higher speeds. Some manufacturers, such as Gigabyte, offer a software overclocking program that allows FSB manipulation through Windows. If in doubt check the manufacturer's website for any specific software applications.

  

2. AGP and PCI Ratio or Dividers:

This option can be described many different ways, but it all boils down to the setting of the AGP and PCI dividers. When using a 66 MHz CPU like the Celeron 500, the AGP and PCI dividers are quite different than with a 133 MHz CPU like the Pentium III 833. Also, the available FSB speeds may be limited only to those that apply to a specific AGP/PCI divider setting. Like the FSB speeds, these AGP/PCI divider settings may be present in the system BIOS or through a set of on-board jumpers. The following images represent the most common formats, from jumpered designs to a full BIOS setup.

Here is a chart outlining the basic divider settings at each of the default settings:

By using basic math, you can see that at each of the default FSB speeds, the dividers each arrive at a speed of 66 MHz for the AGP bus and 33 MHz for the PCI bus. When using FSB speeds between the 66, 100 and 133 MHz settings, simply multiply the appropriate divider against the new FSB speed and the AGP and PCI speeds can be determined.

With motherboards based on newer chipsets such as the Intel 815, 810, 820 or VIA Apollo Pro 133A, there should be options for 66, 100 and 133 MHz operation, though the actual user manual may refer to them using different terminology. Older chipsets like the Intel BX only support the 66 and 100 MHz divider settings. Traditionally, AMD motherboards have only used the 100 MHz setting, though upcoming chipsets such as the KT133A will add the 133 MHz option. Some basic motherboards opt for a pure CPU Detect operation (thereby deriving the FSB from the CPU itself), and not even allow users to set these options manually.

 3. CPU Core Voltage Selection
The ability to select a specific core voltage for the CPU could be the factor that determines overclocking success or failure. Many performance motherboards will offer a full selection of core voltages, most likely through the system BIOS, but also through on-board jumpers. Other motherboards do allow voltage increases, but only using a set percentage boost such as 5% or 10%. This is standard operating procedure on many VIA motherboards, but may also be found with other chipsets as well. There are also motherboards that simply use a “CPU Detect” format that limits the core voltage to the default level of the CPU.

 

4. CPU Multiplier Adjustment  

Other than extremely old Intel CPUs, the clock multiplier can 

only be adjusted on AMD processors. The multiplier can be changed using select AMD Thunderbird/Duron motherboards or with an Athlon Goldfinger overclocking device. This useful feature allows the default CPU multiplier to be overridden and is one of the best methods of getting the highest core speed from an AMD Athlon, Duron or Thunderbird processors.
5. Halted System Procedure
When trying a particularly difficult overclock, sometimes the system may halt and refuse to start up. This is especially true when using the FSB selections in the system BIOS. Look in the motherboard manual for any reference to resetting an overclocked PC and write it down for future reference. Usually this safety feature will consist of a keyboard button being depressed while the system boots. 


Motherboard Analysis
After the above information has been compiled, it should be quite easy to rate the ability of your motherboard to handle the overclocking job ahead. The best scenario is having a motherboard that allows a full complement of voltage tweaks, along with a wide selection of FSB speeds. Even if the FSB options aren't as robust as you would like, the voltage control is the key element. Motherboards that do not offer any form of voltage tweaking at all are likely to offer the lowest overclocking return. The only way around this limitation is by using a Socket CPU/Slocket adapter combo with a Slot 1 motherboard, or an Athlon with a Goldfinger overclocking device. Socketed processors will always require some form of multiplier and voltage control directly from the motherboard itself.
The above analysis is important both in terms of knowing the motherboard's capabilities as well as having realistic overclocking expectations. If you own, or intend to purchase, a motherboard offering full core voltage control and FSB increases in 1 MHz increments then you will have an excellent chance of reaching the CPU's maximum speed. On the other hand, a motherboard with little or no voltage control and a small set of FSB options will have less overclocking success, and the target should be lowered to a more realistic level. Regardless of the overclocking potential of the CPU, the actual overclocked results will either be limited or enhanced by the motherboard itself. The only exception to this is with the AMD Athlon, where the use of an external Goldfinger overclocking device is the key element.
The FSB Riddle
When increasing the FSB beyond the CPU's default speed, it is important to understand how this can affect your system. Depending on the motherboard chipset, some or all of the 66, 100 or 133 MHz settings will be available using standard AGP and PCI dividers. With the Intel 815E or Apollo Pro 133A, this would mean that when running a CPU on the 66, 100 or 133 MHz FSB, that the system would maintain standard 66 MHz AGP and 33 MHz PCI speeds. On older chipsets such as the Intel BX, AMD 750/751 or VIA KX/KT133, this would mean the 100 MHz FSB would be the maximum speed where AGP and PCI speeds would remain standard.
Where this really comes into play is when using a FSB that results in non-standard AGP and PCI speeds. Overclocking most 66 MHz Celerons to the 75 or 83 MHz FSB is usually pretty easy, but it can result in some other system problems. When running on an 83 MHz FSB, the AGP speed is also 83 MHz and the PCI speed is over 41 MHz. These speeds can result in problems with some AGP and PCI cards, and many IDE hard drives will display errors when using a high PCI speed. As a basic rule, speeds of 66-99, 100-132 and 133+ MHz FSB will use the AGP and PCI divider of the lowest default FSB speed in the range. There are exceptions to this rule, and certain motherboards offer higher AGP/PCI dividers at FSB speeds such as 90-99 MHz or 124-132 MHz, which will result in a lower than standard AGP and PCI speed. Running on a 60 MHz AGP and 30 MHz PCI speed may slow your overall system performance just a bit, but it is preferable to running seriously over spec.
The overclocked FSB also has an effect on your system memory, and depending on its quality, may contribute to system instability. Your system FSB is also by default, the speed of your system memory as well. This means is that when using PC100 memory on an overclocked 120 MHz FSB, the memory may not be able to keep up to the higher 120 MHz memory speed. Some motherboard chipsets, such as the VIA products, may allow memory speed to be increased or decreased relative to the system FSB. This can come in handy when overclocking, but those with older chipsets and memory should keep an eye out for memory speed issues, and possibly lower the RAS and CAS BIOS settings accordingly. 


The CPU Multiplier and its Impact
In order to determine an overclocking path, it is advised that you become very familiar with your CPU and its capabilities. The first step is figuring out the CPU multiplier and how it will be affected by increases to the FSB. For example, a Pentium III 500 has a 5X multiplier, which means that when used on the default 100 MHz FSB, it performs at a 500 MHz core speed (5X100). Conversely, a Celeron 700 has a default FSB of 66 MHz, and therefore has a higher 10.5X multiplier. When overclocking the Celeron 700, each FSB increase would be multiplied by 10.5X, while the Pentium III 550 would only experience an increase of 5X. To put this in a proper perspective, overclocking the FSB by only 10 MHz would translate into a 105 MHz jump for the Celeron 700, but only 50 MHz for the Pentium III 500. The higher the CPU multiplier, the more difficult it is to move to a much higher FSB. Knowing your own CPU multiplier can allow you to calculate what the new core speed will be at each new FSB, as well as identifying FSB speeds that may well be out of reach.
This CPU multiplier has been locked on Intel CPUs for a long time now, but the AMD Athlon does allow multiplier changes through the use of a Goldfinger overclocking device, and a few Thunderbird/Duron motherboards also include options to override the default CPU multiplier.
CPU Analysis
This is the research phase of any overclocking exercise, and involves reading up on your current CPU, or one that you may potentially purchase. Recommended reading includes CPU overclocking articles, product reviews and even online opinions. USENET newsgroups are also a good place to gather information about your particular CPU. This research is intended to outline the potential overclocking speed of your processor model and to determine a realistic baseline for the core speed. While it is certainly true that each CPU may not all overclock to the exact same speed, look for overall trends or high success rates. If you read several reviews of the Celeron 566 and find that all of the tested units were able to easily reach the 850 MHz overclock, then this is a trend you can possibly use.
Let's Overclock!
At this point you should have a good working knowledge of your motherboard and its features, along with a pretty good idea on the overclocked core speed that your processor may potentially handle. With this information in hand, it is now time to move on to the basic steps and techniques of CPU overclocking. 


Increase the Core Voltage
The first area that needs to be addressed is the CPU core voltage (Vcore). Increasing the CPU voltage can allow higher CPU clock speeds and at greater stability compared to using the default setting. The core voltage needs to be adjusted before any FSB tweaking is performed and should only be raised by approximately 5-10% at the start. The level of voltage required varies between processors, with some able to make huge overclocking jumps with minimal voltage tweaks, while others may need a larger voltage increase.
Depending on the motherboard design, adjusting the voltage could be as easy as changing a setting in the system BIOS or as difficult as having to open up the PC to adjust an on-board jumper. If your motherboard does not allow core voltage increases, then prospects of overclocking success may be severely limited. Many Athlon Goldfinger devices and Intel Slocket adapters will have options to increase the CPU core voltage as well.
Set AGP and PCI Dividers
The next step is to determine if the AGP/PCI divider needs to be changed to support the desired core speed. Although the KX133A chipset is right on the horizon, current AMD systems use only the 100 MHz speed and this step is not required. Intel owners seeking only a small core speed increase, say from 66 to 75 MHz or from 100 to 110 MHz, can also skip onto the next section. Some Intel motherboards do not allow direct manipulation of the AGP/PCI settings, and use a CPU Detect option for all CPUs. These CPU Detect motherboards can be troublesome to overclock, and while Slot 1 motherboards may be overridden using a Slocket adapter, Socketed PPGA/FCPGA motherboards will simply not allow processors to jump from the 66 to 100, or 100 to 133 MHz FSB.
When looking to overclock your Intel CPU into the next default FSB setting, such as from 66 to 100 MHz, or 100 to 133 MHz, adjusting the AGP/PCI divider is a requirement. Consult the motherboard manual and make the necessary changes, either through the system BIOS or via a jumper setting. If you are using a Slocket adapter, you can usually leave the motherboard set as "CPU Detect" and just change the default FSB on the adapter card itself. For safety reasons, many motherboards also limit the available FSB speeds to only those matching the current AGP and PCI divider settings.
If the divider settings need to be changed, then the PC will need to rebooted with the new settings. On some older BX motherboards, moving from 66 to 100 MHz was a seamless process, but it is a bit more difficult on boards using the newer Intel and VIA chipsets. These boards usually feature a set of hardware jumpers, and if the CPU cannot reach that setting's minimum default speed (66, 100 or 133), then the system will refuse to boot.
If this happens, then resetting the jumpers to CPU Detect will usually allow the PC to be restarted. Then use small increases to the CPU core voltage until the PC finally responds to the new setting. If the PC refuses to boot with the core voltage at a 10-15% increase, then it may be time to lower your overclocking expectations. Some CPUs will simply refuse to make that large a FSB jump and further voltages increases may finally jolt the CPU into action, there are some risks involved. Be very careful when raising the core voltage, as the CPU temperature can also rise precipitously, and heat buildup may damage the CPU or degrade it over the longer term. 


Increasing the FSB
Since Intel processors have a locked clock multiplier, and those from AMD require additional hardware or motherboard features to override the multiplier, increasing the FSB is by far the most popular form of overclocking. It involves changing the system bus speed from the CPU default, and thereby increasing the overall core speed of the processor. A Celeron at 566 MHz runs on the 66 MHz FSB by default (8.5x66), but by simply increasing the FSB speed to 100 MHz, the Celeron would then be overclocked to an 850 MHz core speed (8.5x100). The Celeron 566 remains at the 8.5X multiplier, and the FSB increase would be the sole driver of the higher 850 MHz clock speed.
The FSB options will be accessed either in the system BIOS, jumpers on the motherboard itself, or through the use of a software overclocking utility. There are many different FSB overclocking formats so this is where your intimate knowledge of the motherboard really helps out. Use the information presented earlier in the article and select the appropriate FSB speed and reboot/restart. If the required FSB speeds do not show up in the BIOS, confirm that the correct AGP/PCI divider has been enabled and confirm in the user manual that the board does in fact support this speed.










 

Also, keep an eye on the memory, AGP and PCI speeds that may result from using a non-standard FSB. While the CPU may easily handle a 83 MHz or 124 MHz overclock, some of the other components may not be as forgiving. Front-side bus overclocking is one area where Intel CPUs have the advantage over similar AMD models. Due to the DDR design of the AMD processors, any increase to the FSB actually doubles its effect through the CPU and system bus, so do not expect more than a 110-115 MHz FSB overclock. 

AMD Overclocking With their seemingly-locked multipliers and minimal FSB returns, the AMD Athlon, Thunderbird and Duron processors may look like overclocking chumps next to their Intel competition. Nothing could be further from the truth, and you simply need to explore other options to get the most out of your AMD processor. The most important is the potential to raise the CPU multiplier, which can be a very effective method of overclocking. By increasing a 600 MHz CPU (6X multiplier) up to a 7X multiplier, you will have effectively raised its core speed to 700 MHz, but with none of the stability issues of using a non-standard, >100 MHz FSB. For owners of older Slot 1 Athlons, the solution is to use a Goldfinger overclocking device that fits right onto the processor itself. The outer shell of the CPU will need to be removed, but once attached it will offer total control over the CPU multiplier and voltage settings. This is a powerful tool, and even inexperienced users will find it extremely easy to move the clock up one or two levels and achieve noticeable performance increases.     With the Thunderbird Athlon and Duron processors, overclocking success is almost entirely a factor of the motherboard design. The Socketed Duron and Thunderbird processors do not support external Goldfinger devices, but certain motherboards such as the MSI K7T Pro2, ABIT KT7, ASUS A7V and Soyo K7VTA do have on-board voltage and multiplier controls. Like the Goldfinger devices, these on-board features are easy to use and can yield some significant overclocking gains, especially with some of the lower speed Duron models.

Troubleshooting
When the PC experiences boot or stability problems, troubleshooting the issue is a very important and time-consuming part of any CPU overclock. Here are a few of the common problems you may encounter after finally getting the overclocked PC to boot, as well as some tips as to what could be causing it, and some possible solutions:
1. PC Will Boot but Halts before OS Load
This is a very common overclocking scenario, where the PC boots up perfectly at the higher core speed, but then halts before or during the operating system load. This is usually a problem associated with CPU stability and it may be alleviated with further core voltage increases. If the voltage is already set very high, or the PC refuses to respond after further voltage tweaking, then choosing a lower FSB speed may be required.
2. Error Loading OS
This is very similar to the above example, except it displays a specific error when loading the operating system. Increasing the CPU core voltage, or lowering the FSB can usually alleviate this problem. In some cases, the system memory is at fault, especially when using non-standard FSB speeds that increase the memory speed far beyond its rated spec. Check to make sure the FSB is not above the memory's rated speed, and if so, try lowering the RAS and CAS memory settings to see if this solves the problem.
3. Hard Drive Errors
This type of error occurs because the PCI speeds are out of spec and unsupported by your current hard drive. Older Maxtor drives are notorious for not working on >33 MHz PCI bus speeds and few models will run perfectly when using a FSB such as 83 or 124 MHz. Unless you plan on replacing your current hard drive, lowering the FSB is really your only option.
4. Registry Errors
This is a common occurrence when trying to push the overclocking envelope and exhibits itself by the system halting with an “Error in the Registry” message. The display will then ask you to reboot, which usually creates an endless loop. By far the most common reason for this error is when running your system memory beyond its rated speed. As with other memory issues, try lowering the RAS and CAS BIOS settings and reboot. There can also be other causes, such as CPU or chipset instability, and adding a bit of juice to the Vcore (CPU) or VIO (chipset) voltages may alleviate the problem.
5. OS Crash While Performing Tasks
This type of crash occurs when using a seemingly stable system, and the OS does a hard lock. In select instances the application or game may terminate with an error, while still allowing you access to the OS. This is also one of the toughest problems to troubleshoot, since it may be caused by a multitude of factors. Processor heat buildup is the most common culprit, so keep an eye on the CPU temperature using a diagnostic program such as SiSoft Sandra or by consulting the system BIOS. If the CPU temperature rises dramatically and only then exhibits problems, then adding in a better CPU cooler or enhancing existing system cooling are both options worth exploring.
If the CPU will allow a significant level of activity before crashing, then it is likely not due to the core voltage supplied and it is likely heat related. The shorter the time before a crash, the higher the possibility that overall CPU stability is the issue. Do not discount the impact a non-standard FSB can have on system memory, and AGP and PCI peripherals. Some games can really put your whole system to the test and illuminate problems with a out of spec memory, AGP or PCI speed that may have previously gone unnoticed. 


Finding the Perfect Speed
Finding the optimum overclocked speed of a given CPU is a process of trial and error, spiced with a touch of compromise between speed and stability. There are basically two schools of thought in achieving the optimal overclocked speed.
The first is to decide on a specific overclocked speed and then put all your energy into hitting that target. Perhaps you read all about Celeron or Duron overclocking and purchased a model for the express purpose of getting an XXX MHz overclocked processor. In this case, reaching that speed would be deemed a success and going higher is not an important consideration. The system should be configured expressly for the desired overclock speed, with only small adjustments to the core voltage to determine the safest and most stable operating environment. In the event of the processor not reaching the intended speed, then the process starts again from a slightly lower FSB target.
The other method is to not have any overclocking preconceptions and to simply throw all the FSB speeds and voltages at the processor and see what sticks. This involves a slow and steady progression up the FSB ladder, with stops to increase core voltage or change AGP/PCI dividers as required. At some point you will reach a happy medium of overclocked performance, stability and core voltage. What the final speed ends up being is dependant on the hardware and cooling, and exactly where your risk tolerance lays.
Please remember there are inherent risks to any overclocking exercise, such as long term CPU degradation and serious heat issues. One area you should be extremely careful is when increasing the CPU core voltage above the 5-10% level, as this is the easiest way to damage your processor when overclocking. Using non-standard FSB speeds can also yield AGP and PCI speeds that are far out of spec, and there is a chance that peripheral cards can be damaged.
Of course, some overclockers upgrade their processors on a regular basis and feel that the risk is minimal considering the performance returns and the short amount of time they intend to own the processor. Others are more careful and only overclock to a speed that can be maintained using minimal voltage tweaks and stays within reasonable AGP and PCI bus speeds. Tailor your own overclocking to the amount of risk you are willing to take, as compared to the potential performance benefits.
Test the Overclock
After the optimal overclocking speed has been determined and any system errors properly diagnosed, it is time to put the CPU to the test. Running programs such as Prime 95 for extended periods can illuminate most stability problems, and will give your CPU a good burn-in period. If you do not have that sort of time, then we suggest simply using the system as you normally would, and playing extended sessions of Unreal Tournament and Quake 3 if possible. Those two games, especially UT, will put some serious strain on the CPU and will outline any chinks in the overclocking armor. Another good test is to use a compression program such as PKZip and then select several extremely large files and add them to an archive. Use the highest compression setting available and then run through a few rounds of compress/decompress and watch for any errors.
While testing the overclocked CPU, it is important to take breaks to check the current system, motherboard and CPU temperatures. Different games and applications can tax the system in different ways and watching how the processor reacts to each test can illustrate possible cooling issues. The amount of heat an individual processor can handle can be very different, but as a basic rule, aim for a temperature in the mid 30's ( C) or lower. You may want to explore additional cooling options if the CPU temperature increases rapidly and moves well above 40 degrees C. Again, these are not strict rules, and overall system stability can be a better gauge of the specific CPU's heat tolerance. 

Additional Cooling The prospect of adding additional cooling to your CPU is an inviting one, but first make sure that it will actually do some good. Many times a CPU will not even post at the overclocked speed, and the canned advice may be to “add more cooling”, when in fact, this will probably not help in the least. If the CPU will not even initialize at a given speed, then you will gain nothing by slapping on an expensive CPU cooler. There are exceptions to this rule, and they involve specialized devices such as Peltiers or liquid cooling. The technical expertise required to install these goes far beyond the scope of this article, which deals solely with standard air-cooled devices. Where adding better CPU cooling really comes into play is when the processor will post and load the OS, but then becomes unstable after a short period of time. This is usually indicative of an overheated processor and increasing the cooling effect can in turn increase the stability of the overclocked CPU. The heatsink compound (the material used to link the heatsink to the CPU) is also important, and different types can yield greater heat transfer. The standard silicon-based compound will work best for most overclocking, but those wanting an extra boost should investigate compounds using aluminum, silver or gold elements. Just keep in mind that these metallic substances can conduct electricity as well as heat, so special care should be taken in their use. No CPU cooler will work at peak efficiency if the computer case is filled with very hot air. Keep an eye on the system temperature as well, and if the system starts to overheat, it will have a detrimental effect on the CPU temperature. Adding a couple of case fans may do more for overall cooling and stability than installing a monstrous CPU cooler. In Closing   The process of overclocking a CPU should first be one of gathering information and making sure that you have a good working knowledge of your motherboard features and capabilities. Only then will you have a good chance of reaching the optimal overclock for your particular system. We hope that this guide has helped outline the very basic aspects of CPU overclocking, and will aid novice users get a better idea on how the entire process works. For additional information on specific CPUs and their overclocking potential, be sure to check out the following Sharky Extreme Overclocking Guides: Intel Celeron FC-PGA Overclocking Guide Overclocking the AMD Thunderbird Athlon & Duron Intel Pentium III Overclocking Guide Duron vs. Celeron Overclocking Shootout Athlon GFD Overclocking Guide For more detailed information on specific CPU speeds and pricing, check out Sharky's Weekly CPU Prices Vince Freeman Editor