CPU QuickLinks
Chipset QuickLinks
Video QuickLinks
Software QuickLinks
20XX
Q4 2008
Q3 2008
Q2 2008
Q1 2008
Q4 2007
Q3 2007
Q2 2007
Q1 2007
Q4 2006
Q3 2006
Q2 2006
Q1 2006
Q4 2005
Q3 2005
Q2 2005
Q1 2005
Q4 2004
Q3 2004
Q2 2004
Q1 2004
Q4 2003
Q3 2003
Q2 2003
Q1 2003
Q4 2002
Q3 2002
Q2 2002
Q1 2002
Q4 2001
Q3 2001
Q2 2001
Q1 2001
Q4 2000
Q3 2000
Q2 2000
Q1 2000
Q4 1999
Q3 1999
Q2 1999
Q1 1999
Q1 2002
01/02
Intel Celeron 1.3Ghz (Tualatin-256) was released on January 3rd.
Intel Celeron 1.0A & 1.1A were released on January 3rd with clock speeds of 1 and 1.1Ghz respectively. These processors are based on the Tualatin core (256Kb L2, 0.13micron) and replace the equivalently clocked Coppermine processors (128Kb L2 cache, 0.18micron).
Intel Pentium 4 (Northwood) 2.0 & 2.2Ghz, the 0.13 micron, copper interconnect successor to the Willamette based Pentium 4, was released on January 7th. Northwood features 512Kb of on-die L2 cache, rather than the 256Kb L2 of Willamette. Northwood runs with a core voltage of 1.5v and is shipped in the mPGA478 package. Preliminary benchmarks show Northwood to be between 4% and 10% faster than the equivalently clocked Willamette, depending upon the application.
AMD Athlon XP 2000+ (1.67Ghz) was released on January 7th.
Windows CE .NET, formally known as Talisker, was released on January 7th. Features include a 50% speed increase, support for Bluetooth and 802.11, IE5.5, Media Player 8, DirectX 8, SOAP, multiple displays, user profiles and a HTTP server. The featureset of the OS is fully customisable to allow developers to target a wide range of devices & applications.
nVidia nForce 415-D was announced on January 8th, with motherboards based around this chipset expected towards the end of January. The nForce 415-D is essentially a version of the nForce 420-D without the integrated NV11 (GeForce2 MX) graphics core. The 415-D features nVidia's MCP-D South Bridge, featuring 2 ATA100 IDE channels, 6 USB channels, a Modem, an Ethernet controller, a DirectX 8 compatible Audio Processing Unit (APU) and a Dolby Digital 5.1 encoder, alongside nVidia's new System Platform Processor (SPP) North Bridge. The System Platform Processor is identical to the Integrated Graphics Processor (IGP) found in nVidia's nForce 420 chipsets, but without the integrated NV11 graphics core. Like the nForce 420, one of the main features of the nForce 415 is a 128-bit data path to main memory provided by nVidia's TwinBank technology. For additional information about the nForce architecture see the nForce Roadmap entry.
Intel Pentium III-S (Tualatin) 1.4Ghz with 512Kb L2 cache, was released on January 8th.
Microsoft Visual Studio.NET was released through MSDN on January 16th. See the retail release date Roadmap entry for additional information.
VIA KM266 chipset for AMD processors was released on January 16th. The KM266 is based around the KT266A, but also includes an integrated S3 ProSavage8 graphics core. An AGP slot is supported, but is likely to be optional extra on motherboards. The KM266 chipset also makes use of VIA's VT8233A Southbridge, providing ATA133 support.
AMD Duron (Morgan) 1.3Ghz was released on January 21st.
Intel Price cuts occurred on January 27th. See the Intel CPU Prices page for additional information.
Mobile Athlon 4 1.3Ghz was released on January 28th.
AMD Price cuts occurred on January 30th. See the AMD CPU Prices page for additional information.
Mobile Duron 1.1 and 1.2Ghz were released on January 30th.
02/02
nVidia GeForce4 MX (NV17) was released on February 6th on a 0.15 micron process, with retail availability expected in March. The NV17 is the successor to nVidia's GeForce2 MX and is based around the core of the NV17M, incorporating features from both the GeForce2 and the GeForce4 architectures. The NV17 is essentially a DirectX 7 part, featuring 2 pixel pipelines with 2 texture units per pipeline (the same as the GeForce2 MX), no DirectX 8 Pixel Shaders and only limited Vertex shader support (the NV17 provides vertex shader 'assistance', with much of the processing left to the CPU). However, the NV17 core leverages the Accuview Anti-Aliasing engine, nView support and the improved Visibility subsystem from the GeForce4. It additionally features a slightly modified version of the GeForce4's LMA II (Lightspeed Memory Architecture II) memory controller - The NV17 core uses 2 x 64-bit memory controllers on its crossbar rather than the 4 x 32-bit controllers found on the NV25 core. The NV17 core also features dual integrated TMDS transmitters, for dual DVI output up to 1280x1024, and improved DVD hardware decoding with the introduction of hardware iDCT processing and per-pixel de-interlacing.
There are three versions of the GeForce4 MX - the GeForce4 MX 460 (300Mhz core, 275Mhz DDR, $179), the GeForce4 MX 440 (270Mhz core, 200Mhz DDR, $149) and the GeForce4 MX 420 (250Mhz core, 166Mhz SDR, $99).
Initial benchmarks show that the GeForce4 MX 460, generally speaking, offers performance around the level of the GeForce3 Ti 200. The GeForce4 MX 440 usually comes in slightly ahead of the GeForce2 Ti.
nVidia GeForce4 (NV25) was released on February 6th on a 0.15micron process, with retail availability expected in March. The NV25 core is essentially a highly optimised, higher clocked and slightly enhanced version of the NV20 (GeForce 3) core. The NV25 is an AGP4X part - AGP 8X was not finalised during the design time of the unit.
The Lightspeed Memory Architecture II is perhaps the main contributor to the NV25's enhanced performance over the NV20. LMA II offers a number of new performance enhancing technologies. The QuadCache architecture provides four caches for caching primitives, vertices, textures and pixels. Although much of this caching was done on the NV20 core, these and a number of other caches have been augmented and improved with the introduction of the NV25. LMA II additionally introduces DRAM precharge logic which makes intelligent guesses about which memory banks are to be accessed and precharges them ready for Read/Write operations, effectively reducing memory latencies. Z buffer compression has also been improved slightly to reduce memory load - around 85% of pixels can now be compressed losslessly. Perhaps the most significant performance increases, however, are due to improvements in the Visibility subsystem - the new Z occlusion culling algorithms are around 25% more efficient than those found in the NV20 core.
The nFiniteFX II engine improves upon the nFiniteFX unit found in the NV20, providing two vertex shaders with automatic load distribution rather than the single vertex shader found in the NV20 core. It should be noted that this feature was first introduced by nVidia with the NV2A graphics core of the X-Box, and is also present in the Radeon 8500's R200 core. The vertex shaders themselves have been improved over the vertex shader present in the NV20 and NV2A, effectively processing all vertex instructions in a single cycle by hiding their latency in the vertex pipeline. The introduction of the second vertex shader, its enhancements and the clock speed increase mean that the vertex performance of the NV25 is around 3 times that of the NV20. The NV25 also offers the Particle Engine first introduced with the NV2A. nFiniteFX II offers a single pixel shader, as found in the NV20 core. The NV25 core also features the same 4 pixel pipelines with 2 texture units per pipeline as found in the NV20.
The 'Accuview' Anti-Aliasing unit of NV25 offers a new 4XS FSAA algorithm as well as notable performance improvements - Quincunx AA is now as fast as 2X FSAA. The new 4XS AA mode provides higher quality FSAA than traditional 4X FSAA by offering more texture samples per pixel.
nView replaces TwinView as nVidia's multi-display solution, and is now present in nVidia's high-end GPUs. NV25 features dual integrated 350Mhz RAMDACs, providing support for two CRTs without additional circuitry, and supports up to 2 external TMDS transmitters for DVI. The NV25 also features an enhanced Video Playback Engine (VPE), which was first introduced with the NV17M core, to improve MPEG decoding. This provides hardware iDCT processing and per-pixel de-interlacing, both of which were not present in previous nVidia GPUs but were implemented on ATI products.
Although the GeForce4 offers little in the way of new features, nVidia have done a great job at optimising the GeForce3 core. The performance of the top of the range GeForce4 Ti 4600 is between 20 and 50% faster than the GeForce3 Ti 500, which is a much bigger performance gap than that offered by the GeForce3 over the GeForce2 Ultra when the GF3 was released (most games offered very similar performance on these two chips at that time!). Additionally it is worth noting that the greatest performance increases come from using newer titles, offering greater geometrical detail and using more texture passes, and by using Anti Aliasing and running at high resolutions. The lower clocked GeForce4 Ti 4400 also offers excellent performance, sitting approximately half way between the GF3 Ti 500 and the GF4 Ti 4600.
There were initially two versions of the GeForce4 - the GeForce4 Ti 4600 (300Mhz core, 325Mhz DDR), which initially retail for around $399, and the GeForce4 Ti 4400 (275Mhz core, 275Mhz DDR), which initially retailed for around $299. A lower clocked GeForce4 Ti 4200 (225Mhz core and a 250Mhz DDR (64Mb cards) or 222Mhz DDR (128Mb cards) memory clock) was released approximately 8 weeks later at around $199.
Microsoft Visual Studio.NET, the follow up to Visual Studio 6, was released on February 13th. In addition to updating Visual Basic, Visual FoxPro and Visual C++, Visual Studio.NET will include a number of new technologies including Microsoft's new language C# (C-Sharp), ADO+, SOAP, Windows Forms, Web Forms and Web Services. Additional ATL features and cross language debugging are also expected to available. The Web Service features will allow developers to link applications, services and devices over the internet using HTTP, XML and SOAP.
Visual Studio.NET will be released in three versions - Professional Edition, Enterprise Developer Edition and Enterprise Architecture Edition. The Enterprise products will feature pre-build eCommerce site components, testing tools, modelling software, team collaboration tools etc. The Enterprise Architecture Edition adds features for managers who need to work with developers to design and build distributed applications.
Microsoft X-Box was released in Japan in the week of February 17th. See the US release for additional details.
nVidia Quadro4 XGL series were released on February 19th. The Quadro4 XGL solutions are aimed at the 3D workstation market, and are built around either the NV25GL and NV17GL cores, depending on the model. Unlike previous generation Quadro GPUs, which were essentially just GeForce GPU's with certain features switched off though their chip id, both the NV25GL and the NV17GL are physically different from the NV25 and NV17 cores. Improved or new features over the NV25 and NV17 cores are improved hardware support for Anti-aliased lines and hardware support for overlay planes, XOR operations and window clipping.
The members of the Quadro4 XGL series are as follows:
Quadro4 900 XGL - NV25GL core, highest clock speed (300Mhz), 64Mb SDR
or DDR SDRAM
Quadro4 750 XGL - NV25GL core, mid range clock speed, 128Mb DDR SDRAM
Quadro4 700 XGL - NV25GL core, mid range clock speed, 64Mb DDR SDRAM
Quadro4 550 XGL - NV17GL core, low end clock speed, 64Mb DDR SDRAM
Quadro4 500 XGL - NV17GL core, low end clock speed, 64Mb SDR SDRAM
nVidia Quadro4 NVS series were released on February 19th. The Quadro4 NGS solutions are aimed at the 2D workstation market, and are built around the NV17GL core - see the Quadro4 XGL Roadmap entry above for details of this core. nVidia hope to aim at companies such as Matrox by providing high quality 2D output and multi-monitor cards with up to 4 outputs per card (any combination of VGA and DVI outputs) - up to 32 displays are supported with multiple PCI cards. nVidia do not recommend these cards for 3D output, as their 3D performance is expected to be fairly poor (due to low clock speeds and SDR memory interfaces). nVIdia's nView software has also seen improvements for use with these cards.
The members of the Quadro4 NVS series are as follows:
Quadro4 400 NVS - PCI, 64-bit DDR SDRAM, 4 monitor output (2 x NV17GL
chips)
Quadro4 200 NVS - AGP, 128-bit SDR SDRAM, 2 monitor outputs (1 x NV17GL
chip)
VIA KT333 chipset for the AMD Athlon processor is expected to be released on February 20th. The KT333 is the successor to the KT266A, adding support for PC2700 DDR SDRAM. The KT333 chipset will make use of VIA's VT8233A Southbridge, providing ATA133 support.
Intel Xeon (Prestonia) 1.8, 2.0 & 2.2Ghz, the 0.13u Northwood-based successor to Foster, were released on February 25th. Prestonia features the same 512Kb on-die L2 cache as Foster, but additionally features Intel's HyperThreading Technology, formally known as Jackson SMT technology. HyperThreading is Intel's implementation of Simultaneous Multi Threading (SMT), which allows a single CPU to interleave the processing of two threads together to make more efficient use of the processor pipeline. HyperThreading is expected to provide a significant performance increase in multi-threaded applications, with a 30% increase in rendering performance demonstrated at August 2001's IDF. A HyperThreading capable CPU is expected to appear as two CPU's to the operating system.
Intel E7500 chipset (Plumas / Plumas LE) for the Xeon processor was released on February 25th. Plumas features support for up to 16Gb of dual channel DDR200/266 DDR SDRAM (although the clock speed will be fixed at 200Mhz with a 400Mhz FSB) and PCI-X. The Memory Controller Hub (MCH/Northbridge) sports three Hub Link 2.0 channels (16bit, 266Mhz) to Intel P64H2 PCI/PCI-X bridges, and a Hub Link 1.5 channel (8bit, 266Mhz) to an ICH3-S chip (South Bridge). Each of the P64H2 chips can support two 64bit, 133Mhz PCI-X busses, giving a maximum of 6 PCI/PCI-X busses (this equates to 6 slots, as each PCI-X controller can support only one slot). ICH3-S is identical to ICH3 (6 USB ports, but otherwise the same as ICH2), but offers a faster link (Hub Link 1.5 rather than Hub Link 1) to the MCH.
ServerWorks Grand Champion GC-LE Foster chipset was released on February 26th. GC-LE is the cut down version of Serverwork's GC-HE chipset, designed for use in 2-way Xeon SMP environments. The memory controller supports up to 8 banks (16Gb max) of 2-way interleaved Registered PC1600 DDR SDRAM, giving a memory bandwidth of 3.2GB/s. The chipset also features dual PCI-X controllers and an On Line Spare Memory feature, which allows a spare DIMM to be used if another DIMM fails.
03/02
Mobile Pentium 4-M 1.6 & 1.7Ghz was released on March 4th. The Mobile P4 will be based on the Northwood core and feature 512Kb on L2 cache.
Intel i845MP chipset (Brookdale-MP), for use with the Pentium 4-M, was released on March 4th. The i845MP is expected to feature support for DDR266 SDRAM and will contain advanced power management logic, introducing "deep sleep" and "deeper sleep" low power states.
Intel i845MZ chipset (Brookdale-MZ), for use with the Pentium 4-M, was announced on March 4th with availability later in the year. The i845MZ is a cut down version of the i845MP.
Intel Price cuts for the Celeron series occurred on March 4th. See the Intel CPU Prices page for additional information.
SiS 745 chipset, for the AMD Athlon processor series, was released on March 5th. SiS 745 is the successor to the SiS 735, adding support for IEEE1395 and DDR333 SDRAM.
SiS 645DX chipset, for the Pentium 4, was released on March 10th. The SiS 645DX is based around the SiS 645 core, but includes support for ATA133 and a 533Mhz FSB speed.
Intel Xeon (Multi Processor) 1.4, 1.5 & 1.6Ghz, formally known as Foster MP, were released on March 12th. In addition to the 256Kb on-die L2 cache found in the Pentium 4 Willamette and Dual Processor Foster, the Multi Processor version of Foster will feature an integrated tertiary (L3) cache of 512Kb (1.4 & 1.5Ghz) or 1Mb (1.6Ghz). Foster MP will allow 4 and 8-way SMP.
ServerWorks Grand Champion GC-HE Foster MP chipset was released on March 12th. Grand Champion HE is the first chipset to become available for the Multi Processor version of Foster (Foster MP) and supports up to 4 Xeon processors, 32Gb of DDR SDRAM memory and PCI-X. The Grand Champion's memory controller features support for 4-way interleaved PC1600 DDR SDRAM (PC1600 is used to keep memory access synchronous with the quad-pumped 100Mhz FSB). This gives the Grand Champion 6.4GB/s of memory bandwidth (100 x 2 x 8 (64bit) x 4) - enough to keep 4 bandwidth-hungry Xeon's happy.
AMD Athlon XP 2100+ (1.73Ghz) was released on March 13th. This CPU is the last Athlon XP to be built around the Palomino core.
AMD Athlon MP 2000+ (1.67Ghz) was released on March 13th.
AMD Athlon 4 1600+ (1.4Ghz) was released on March 13th.
STM Kyro II SE (STG4800) was released on March 13th. The STG4800 is a higher clocked 200Mhz version of the original 175Mhz Kyro II (STG4500). The featureset is identical to its predecessor, although the drivers have been improved to include AGP4X Side Band Addressing and enT&L. enT&L is a software emulation of a DirectX 7 compatible, fixed function hardware T&L engine.
ATI Radeon IGP 320 & 320M, formally known as the ATI A3 North Bridge, for the AMD Platform were announced on March 13th, with availability expected in May. The Radeon IGP 320 and 320M are essentially the same chipset, with the M model designed for Notebooks - both chipsets support ATI's PowerPlay power saving technology. The IGP 320 series supports PC2100 (DDR266) and PC1600 (DDR200) DDR SDRAM over a 64-bit memory bus and contains integrated UMA graphics as well as an external AGP 4X slot. The integrated graphics core is based around ATI's RV100 (Radeon VE) core, containing 1 pixel pipeline with 3 texture units per pipeline, Hydravision multi-display support, hardware accelerated DVD decoding, HyperZ and TV-Out. The integrated graphics do not, however, offer support for hardware T&L or TMDS. The IGP (North Bridge) connects to the South Bridge via ATI's PCI compatible A-Link bus. This allows motherboard manufacturers to use either third party PCI connecting South Bridges (e.g. VIA's 686B) or ATI's own IXP South Bridges. As with nVidia's nForce, the drivers come in a single installation package, which ATI calls FlexFITT. The cost of the Radeon IGP series is expected to be noticeably less than nVidia's nForce chipset, although performance will be worse than the nForce 420, mostly due to the 64-bit memory bus.
Initial benchcmarks show that the Radeon IGP offers reasonable 3D performance compared with other integrated chipsets, but only around 50-60% of the performance offered by nVidia's nForce 420. System performance is also reasonable, but lags behind chipsets such as VIA's KT266A/KT333 and nVidia's nForce.
ATI Radeon IGP 330, formally known as the A4 North Bridge, for the Intel Pentium 4 Platform was announced on March 13th with availability expected in May. The Radeon AGP 330 is based around the same core as the IGP 320, but interfaces to the 400Mhz FSB of the Pentium 4. This chipset is targetted at the current 400Mhz FSB Pentium 4 Northwood core and the forthcoming P4 Celeron.
ATI Radeon IGP 340 & 340M for the Intel Pentium 4 Platform was announced on March 13th with availability expected in July. The Radeon AGP 330 is based around the IGP 330, but interfaces to the 533Mhz FSB of Northwood B. The graphics clock of the 340 series is expected to be higher than that found in the 330 & 320 series.
ATI Radeon IXP 200 & 250 Integrated Communications Processor (South Bridge) were announced on March 13th with availability expected in July. Both IXPs connect to the IGP through ATI's A-Link bus and include ATA100 support, an integrated 3Com 10/100 Ethernet controller, 6 port USB2 and 6 channel audio. The IXP 250 is aimed towards businesses and therefore contains Wake On Lan support, a Desktop Management interface, ASF support and a managed boot agent.
SiS 648 chipset, for the Pentium 4, was announced on March 13th and officially released on July 22nd. The SiS 648 is the successor to the SiS 645DX and includes support for a 533Mhz FSB speed, AGP 8X and DDR400 SDRAM. The SiS 963 Southbridge supports 3xIEE1394 ports, 6xUSB2.0 ports and ATA133.
nVidia nForce 620-D & 615-D chipsets for the AMD platform were announced on March 14th, although no nVidia subsequently decided not to release any of these chips. The nForce 620-D and 615-D are the successors to the nForce 420-D and nForce 415-D chipsets respectively. These chipsets include support for PC2700 (333Mhz) DDR memory, which will provide a memory bandwidth of 5.4Gb/s with nVidia's TwinBank architecture.
VIA P4X333 chipset for the Pentium 4 processor was released on March 14th. The P4X333 is based around the P4X266A, but additionally includes support for DDR333 SDRAM, a 533Mhz FSB speed and AGP8X. The memory controller is well optimised, being an updated version of the controller found in the KT333 Athlon chipset. The P4X333 North Bridge is paired with VIA's VT8235 South Bridge, which offers support for ATA133, 6 USB 2.0 ports, an integrated 10/100Mbps MAC, 6 channel AC97 audio and an AC97 modem. The North and South bridge are connected by VIA's improved 533MB/s V-Link bus.
Benchmarks suggest that the P4X333 is the highest performing Pentium 4 DDR chipset on the market, beating the i845E/G series by a small margin.
VIA Mobile C3 (Ezra T core) CPU was released on March 14th, at clock speeds up to 933Mhz. The C5M used in the Mobile C3 is the mobile equivalent of the C5C core which is built on a 0.13micron process and contains 128Kb of L1 cache and 64Kb of L2 cache.
Microsoft X-Box was released in Europe on March 14th. See the US release for additional details.
