VLab's Development of a Turn-key System

VLab's  demonstration system features five SGI O2 client systems, and is configured with an SGI Indigo server system server system.

The SGI 02 is a system that has been optimized for working with video data.

This platform is currently being used to test the VPS system.  For a descriptin of VPS see the overview.

A review of the 02 system can be found in the  December 1997 (pg 58-63) issue of DV magazine.

Forthcoming

video data rate specs.
System data rates.

SGI Spec ratings and Comparisoms

The SPEC CPU2000 benchmark suite is the recognized industry standard for assessing processing performance. SPEC developed the CPU2000 benchmarks based on actual end-user applications. These tests depend on processor, memory subsystem and compiler performance when executing integer and floating-point computations.

The SPEC CPU2000 benchmark suite comprises four metrics ÷ SPECint_base2000, SPECfp_base2000, SPECint_rate_base2000, and SPECfp_rate_base2000 ÷ to represent different performance characteristics of the system.

SPECint_base2000 and SPECfp_base2000 measure the speed of a single task ÷ either an integer calculation or a floating-point calculation ÷ executing on a single processor. Each test measures how long the processor takes to complete the benchmark set of single tasks.
 
 

Origin200 360MHz R12k

Integer                     .pdf

Floating Point          HTML
 

IBM RS 6000
 

Floating Point           .pdf

IBM pSeries

Floating Point           .pdf

IBM Power 4 (PPC G5)

G4

The SPEC benchmarks disable a second CPU if present, and don't take advantage of the Altivec and Screaming Sindy dedicated FP instructions - that's down to the
compiler.  The problem is, one former Apple engineer told us, in serializing the twenty five year old BSD layer with the fifteen year old code of the extensions NeXT
began to add in the mid 1980s.  Serialization is one of the toughest problems for an engineer to solve.

'll jump the gun a bit by saying that the G4e's reliance on PC133 SDRAM is its Achilles heel. The dual-channel RDRAM subsystem with which
the P4 is usually coupled is miles ahead of anything that Apple has released.

The G4e's Altivec implementation is by far its most compelling feature. The processor boasts four independent Altivec units, three of which are fully pipelined and can operate on multiple instructions at once.

All of this SIMD execution hardware is tied to a generous register file that consists of thirty two, 128-bit architectural registers and 16 additional vector renames.