bose@iuvax.UUCP (12/07/87)
The Norton SI index for the PS/2 Model 30 is 1.9. My PC XT has a SI index of 1.8 with a V20 chip. However the Model 30 seems to run faster. Can anyone tell me what it is that I am observing?
smvorkoetter@watmum.UUCP (12/08/87)
In article <36300005@iuvax> bose@iuvax.cs.indiana.edu writes: > >The Norton SI index for the PS/2 Model 30 is 1.9. My PC XT has a SI index >of 1.8 with a V20 chip. However the Model 30 seems to run faster. > >Can anyone tell me what it is that I am observing? You are observing the uselessness of the SI benchmark. SI does a lot of processor intensive work, such as multiplication. The V20 is very good at such things, and thus can do the 1.8 times as fast as an 8088. However, for normal processing, which is more bus intensive, the 8 bit bus on the 8088/V20 is the bottleneck. This is why you will only observe a 5-10% improvement with the V20 for general processing tasks. The Model 30 on the other hand has a 16 bit 8086, which means it can fetch twice as much data in one bus cycle, thus speeding up operations quite a bit for average processing. The '86 in the Model 30 also runs at 6Mhz instead of 4.77, which also increases the speed. So, the 1.9 reading for your model 30 is probably pretty close to a useful result, indicating that the 30 is about 1.9 times as fast. However, SI is useless at comparing V20's with 808x's. Stefan Vorkoetter Symbolic Computation Group University of Waterloo
johnl@ima.ISC.COM (John R. Levine) (12/08/87)
In article <36300005@iuvax> bose@iuvax.cs.indiana.edu writes: >The Norton SI index for the PS/2 Model 30 is 1.9. My PC XT has a SI index >of 1.8 with a V20 chip. However the Model 30 seems to run faster. >Can anyone tell me what it is that I am observing? You're observing the oft noted fact that Norton's SI numbers don't mean much. In particular, the value that SI reports is heavily affected by the speed at which a chip can do multiplication instructions, and the V20 multiplies much faster than an 8088 or 8086. But the PS/2 has an 8086 with a 16 bit memory, while the XT has only an 8-bit memory, so that the XT has to make roughly twice as many memory cycles as the PS/2 to do the same work. That's why the PS/2 is faster. -- John R. Levine, IECC, PO Box 349, Cambridge MA 02238-0349, +1 617 492 3869 { ihnp4 | decvax | cbosgd | harvard | yale }!ima!johnl, Levine@YALE.something The Iran-Contra affair: None of this would have happened if Ronald Reagan were still alive.
berger@clio.las.uiuc.edu (12/09/87)
You're observing the meaninglessness of the Norton SI figure.
It must rely heavily on data transfers, because my NEC APC,
running at 5 MHz with an 8086, gets a 1.6 rating. It reads
2.5 with a V30 installed. It certainly doesn't run any real
programs 2.5 times faster than a PC.
Mike Berger
Center for Advanced Study
University of Illinois
berger@clio.las.uiuc.edu
{ihnp4 | convex | pur-ee}!uiucuxc!clio!bergerz7m8@sphinx.uchicago.edu (ari shlomo zymelman) (12/14/87)
You are observing the uselessness of the SI benchmark. SI does a lot of processor intensive work, such as multiplication. The V20 is very good at such things, and thus can do the 1.8 times as fast as an 8088. However, for normal processing, which is more bus intensive, the 8 bit bus on the 8088/V20 is the bottleneck. This is why you will only observe a 5-10% improvement with the V20 for general processing tasks. The Model 30 on the other hand has a 16 bit 8086, which means it can fetch twice as much data in one bus cycle, thus speeding up operations quite a bit for average processing. The '86 in the Model 30 also runs at 6Mhz instead of 4.77, which also increases the speed. So, the 1.9 reading for your model 30 is probably pretty close to a useful result, indicating that the 30 is about 1.9 times as fast. However, SI is useless at comparing V20's with 808x's. Could someone please comment on the difference between the new PS/2 30 and the AT&T 6300 which also has an 8086. Is there any speed difference, or is IBM finally putting out a comparable machine? -- Ari Zymelman