Path: utzoo!utdoe!generic!pnet91!ericmcg From: ericmcg@pnet91.cts.com (Eric Mcgillicuddy) Newsgroups: comp.sys.apple2 Subject: Re: ACCELERATOR - ZIP OR TRANSWARP? Message-ID: <731@generic.UUCP> Date: 15 May 91 23:30:03 GMT Sender: root@generic.UUCP Organization: People-Net [pnet91], Etobicoke, ON Lines: 40 >The Zip is usually faster than the transwarp at the same speed. Adding >cache and upping the speed generally should go hand in hand for best >price/performance (7mhz/8K, 8mhz/16K, 9mhz/32K, 10mhz/64K) -- roughly. >The upgrade prices change depending on availability so don't hold me to >closely to that. > >Todd Whitesel >toddpw @ tybalt.caltech.edu I agree with the first, but not the last. The size of the cache will affect the hit ratio (the probability of any given data being in "fast" memory). Suppose 8k gives a 50% hit ratio, 16k gives 75%, 32k gives 95% and 64k gives 97%. speed cache 7 8 9 10 8k 3.5 4 4.5 5 16k 5.25 6 6.8 7.5 32k 6.7 7.6 8.6 9.5 64k 6.8 7.8 8.7 9.7 The figures give the effective speed of the processor. The figures are not absolutely accurate, just estimates, but for most applications these are close enough. You can see that doubling the 8k cache to 16k gives better performance than increasing the speed. similarly a 64k cache is much better than an 8k cache at any available speed, although the improvement over 32k is negilible. The critical point is cost for the improvement. The Zip chip speed upgrade to 10MHz is around $200 CAN, the cost for 64k of SRAM is $60 CAN. This gives my 8MHz Zip more performance than a 16k 10MHz Zip, but only just, the cost however is much less. I think you will find incremental increases work out cheaper if you buy the memory first and then the speed. If you are planning ahead spend a little extra for faster than required memory to handle the higher speed later on. UUCP: bkj386!pnet91!ericmcg INET: ericmcg@pnet91.cts.com