It seems IBM Corp’s Power Personal Systems Division has demonstration systems of Power Series machines running at 133MHz with 1Mb of level 2 cache and 60nS interleaved RAM that hit 190 Specmarks. Much better than the Specmarks of the machines we reported on just recently, which were well below the estimated SPECmarks of the chips. The big cache and fast memory in the demo versions account for much of the difference, but there is also a suggestion that IBM’s ‘Dakota’ PCI memory controllers still have some rough edges to be knocked off. Meanwhile, the company’s benchmarking team has been doing its own measurements. The nutshell claim from the division is that at the same clock speed, the PowerPC systems run Bytemarks benchmarks more than twice as fast (2.15 times to 2.17 times), as a top-of-the-line Compaq Computer Corp and National Software Testing Laboratory’s InterMark nearly twice as fast (1.94 times). The Spec figures look less impressive: 1.19 times for integer and 1.38 times for floating point, but these are compared against Intel Corp’s own figures presented to the SPEC committee. The benchtests were carried out under Windows NT, and IBM made a point of noting that the Motorola Inc compilers under NT are still in their infancy and as it matures significant performance improvements are expected. Apart from the synthetic benchmarks, the company has been testing a mixture of 32-bit NT applications, 16-bit to 32-bit conversions and 16-bit Windows applications running under emulation.
Dire performance
The most notable aspect of the results here is the dire performance shown by some aspects of Microsoft Corp’s 32-bit version of Excel. IBM used the National Software Testing Laboratory’s test script and is designed to produce a representative workload. With most of the functions, insert/delete; scaled cut and paste; file input-output; charting and so on, the PowerPC appears to outstrip the Pentium machine comfortably. But when it comes to measuring financials, statistical maths, addition and exponentials, speed drops way below the results for the Pentium. A number of functions in 32-bit Word suffer too – search and replace, block copy and page down, are all slower on the PowerPC. Why? The suggestion from within IBM is that Microsoft either originally implemented some of these functions as chunks of iAPX-86 assembler which it then converted, or that the functions are optimised for the iAPX-86 architecture. Fair enough, considering that the vast majority of NT installations are on the iAPX-86-compatible system, but less good as an example of Microsoft’s cross-system pretensions. It would be interesting to see the same tests run on Alpha and R-series systems. The same problems are even more pronounced when 16-bit Excel is run under emulation on Windows NT, although 16-bit Word runs consistently faster on the PowerPC than Pentium. As always, with benchmarks carried out by one vendor, it’s worth being cautious about the results. IBM is now pondering whether to go to the laboratories for a truly independent test.