GartnerGroup research note - December 1999
As the gap between hardware performance and mainstream application requirements widens, enterprises must shift their focus from megahertz and megabytes to platform stability and product continuity
---------------------------------------------------
Intel Desktop Configurations: Winter 1999-2000 Update
K. Knox
We update our recommended desktop configurations to reflect ongoing improvements in hardware technology and the growing need for enterprise stability.
--------------------------------------------------------------------------------
Core Topics
Desktop Platforms, Markets and Strategies ~ Hardware & Operating Systems
PC/Laptop -- Procurement, Maintenance and Obsolescence Strategies ~ IT Management
Key Issue
How will companies match the growing variety of hardware technology to organizational and user requirements?
--------------------------------------------------------------------------------
As the gap between hardware performance and mainstream application requirements widens, enterprises must shift their focus from megahertz and megabytes to platform stability and product continuity; however, doing so is becoming increasingly difficult. Despite the initial efforts of Intel and others to ease product transitions and improve platform stability, selecting and standardizing desktop hardware is more complex and challenging than ever. Product introductions, platform transitions and uncertain external market factors are driving significant changes in desktop hardware. We examine key technology and market trends and offer recommended desktop configurations.
Microprocessor: Processor performance continues to outpace most mainstream application requirements significantly, with no major discontinuities on the three- to four-year horizon to force a major hardware transition. Enterprises should continue to ride down price curves for mainstream requirements and focus on stability and continuity in lieu of megahertz performance. Intel, however, will continue to drive processor performance to new levels and is currently in the midst of a transition to its next-generation, 0.18-micron architecture, code named "Coppermine," which will allow even-faster processor performance levels to be achieved (see Note 1). Despite these improvements, enterprises should continue to focus on the entry level of the Pentium III product line (currently 500MHz) for entry-level and mainstream users (see Note 2). For high-end desktop and low-end workstation requirements, where processor performance is still an issue, acquisitions should be focused on Intel's highest-performing Pentium III processor (currently 733MHz).
--------------------------------------------------------------------------------
Note 1
Coppermine
Coppermine chips are made with a 0.18-micron process, which allows a smaller, less-expensive, faster chip. The new process also enables Intel to implement a new integrated transfer cache technology. By integrating the cache on the processor unit, Intel is able to reduced the amount of cache being used from 512Kb to 256Kb, while improving overall performance. The specification for "512Kb L2 cache" should be immediately eliminated from all desktop RFPs.
Note 2
Celeron
For Intel's low-cost Celeron processor, the issue continues to be packaging. Today's Celeron-based systems generally have shorter product life cycles, less consistency and continuity, limited manageability, and, as a result, higher TCO. However, we believe that highly managed, stable Celeron platforms will emerge in 1H00 and will be a viable alternative to Pentium III.
--------------------------------------------------------------------------------
Chipset: Selecting a configuration solely on the basis of processor speed is no longer viable; instead, enterprises must now look beyond megahertz and focus on the underlying chipset as a more critical factor in the selection process. Moving from a 500MHz to a 600MHz processor is a relatively easy and straightforward process. Chipset changes, on the other hand, almost always include new drivers and system software, which require enterprises to requalify applications and create new system images ? a very labor-intensive, disruptive process. For mainstream requirements, enterprises should continue to purchase 440BX-based systems through 1Q00 and make a transition only when a stable next-generation chipset platform (see Note 3) has been established by their OEMs, offering 12- to 18-month image stability and product availability. Despite the advanced feature set and performance capabilities of Intel's 820 chipset, external factors and technical delays will limit this chipset to only high-end, performance-driven configurations (see Note 4).
--------------------------------------------------------------------------------
Note 3
Next-Generation Chipsets
As a result of issues related to the 820 chipset and Rambus memory, PC OEMs have adopted several different mainstream chipset strategies. Although the underlying chipset technologies do vary slightly, enterprises should focus on stability and transition management in selecting an appropriate platform on which to standardize.
Intel 810e offers many of the same features as the 820, with the key differences being slower bus speeds, an integrated video implementation and the use of SDRAM memory.
Non-Intel chipsets (e.g., Via) have most of the same features as the 820, with the exception of SDRAM support rather than RDRAM.
Intel 440BX: Some OEMs are not making the transition and will ride 440BX into 2H00.
Note 4
The Intel 820
Intel's 820 chipset has been delayed several times since it was announced in 1998. Rambus memory availability and pricing have been major problems. RDRAM is still significantly more expensive than SDRAM and will continue to be through 2Q00. The 820 has also been affected by a number of technical issues. The 820 chipset does offer several major new enhancements and features:
133MHz front-side bus
Support for RDRAM
ATA-66 disk interface
AGP 4X graphics
Security technologies (e.g., a hardware random-number generator)
Manageability technologies (e.g., Alert on LAN)
--------------------------------------------------------------------------------
Memory and Disk: Despite the current spike in SDRAM pricing, memory continues to be a great investment for enterprises and should not be compromised. Enterprises should make 128MB a minimum requirement for all new system purchases and consider moving to a standard of 256MB. In 2000, we will begin to see a transition away from traditional SDRAM to much-higher-performing Rambus memory, but availability and pricing will limit the appeal of this technology to high-end, performance-driven configurations. Unlike memory, hard-disk capacities continue to outpace most application and data requirements. Disk capacities are increasing at a rate of two to three times every 12 months. Enterprises should continue to focus on entry-level disk capacity, which today is 6.4+ gigabytes for mainstream systems.
Display: Seventeen-inch CRT monitors continue to be the mainstream corporate desktop display standard and will remain so through 2002. Although FPDs do offer ergonomic and performance benefits over CRTs, the three- to four-times cost differential prohibits broad use. However, enterprises purchasing 19-, 20- and 21-inch monitors, or those constrained by space, should consider FPDs as a viable alternative.
Figure 1
Intel Desktop Recommended Configurations
Source: GartnerGroup
--------------------------------------------------------------------------------
Acronym Key
AGP Accelerated Graphics Port
ATA Advanced Technology Attachment
CRT Cathode ray tube
EIDE Enhanced Integrated Drive Electronics
FPD Flat-panel display
L2 Level 2
OEM Original equipment manufacturer
RDRAM Rambus dynamic random-access memory
RFP Request for proposal
SCSI Small Computer Systems Interface
SDRAM Synchronous dynamic random-access memory
TCO Total cost of ownership
--------------------------------------------------------------------------------
Bottom Line: Our recommended configurations (see Figure 1) should be used as a guideline for the rational deployment of PC desktop hardware on the basis of user and organizational needs.
--------------------------------------------------------------------------------
This document has been published by:
Service Date Document #
End-User Computing 13 December 1999 T-09-3470
PC/Mobile Buyer 13 December 1999 T-09-3470
Personal Hardware 13 December 1999 T-09-3470
PRISM for Distributed Computing 13 December 1999 T-09-3470
Mobile Business Strategies 13 December 1999 T-09-3470
Equipment Asset Management 13 December 1999 T-09-3470
Equipment Procurement - North America 13 December 1999 T-09-3470
Midsize Enterprise IT Management & Support 13 December 1999 T-09-3470
LAN and Desktop Management 13 December 1999 T-09-3470
PRISM for IT Asset Management 13 December 1999 T-09-3470
NT Strategies 15 December 1999 T-09-3470
--------------------------------------------------------------------------------
|
