krowitz@RICHTER.MIT.EDU (David Krowitz) (07/22/88)
More on the new workstations ... I was under the impression that Motorola had claimed that the 68030/68882 chips would be quite a bit faster than the 68020/68881 chips operating at the same clock rate. They said something about the on-chip memory management of the 68030 cutting down on the number of clock cycles needed to complete an instruction. In light of this, wouldn't you expect the 25-MHz 68030 in the DN3500 to have a higher MIPS rating than the 25-MHz 68020 in the DN4000? For that matter, why would the 33-MHz chip in the DN4500 be rated at 7 MIPS when the 25-MHz chip in the DN3500 is rated at 4 MIPS? That's a 75% increase in MIPS with only a 50% increase in clock rate. Do both machines have equivalent caches? -- David Krowitz krowitz@richter.mit.edu (18.83.0.109) krowitz%richter@eddie.mit.edu krowitz%richter@athena.mit.edu krowitz%richter.mit.edu@mitvma.bitnet (in order of decreasing preference)
joelm@apollo.COM (Joel Margolese) (07/28/88)
In article <8807212113.AA05459@richter.mit.edu> krowitz@RICHTER.MIT.EDU (David Krowitz) writes: >More on the new workstations ... I was under the impression that >Motorola had claimed that the 68030/68882 chips would be quite a >bit faster than the 68020/68881 chips operating at the same clock >rate. They said something about the on-chip memory management of >the 68030 cutting down on the number of clock cycles needed to >complete an instruction. In light of this, wouldn't you expect >the 25-MHz 68030 in the DN3500 to have a higher MIPS rating than >the 25-MHz 68020 in the DN4000? The DN4000 has a virtual cache which boosts performance by not having to go through the PMMU (at least when you hit the cache). The 3500 has no external cache, (just the relativly small caches within the 68030) with the result that it has about the same performance as the dn4000 did, with fewer chips. (Isn't that how progress always goes?) The DN4500 has a physical cache, as well as the higher clock speed, and therin lies the performance boost. (You can't have a virtual cache when the translation is all done on-chip.) For upgrade purposes, I think only the DN3550, (a dn3500 with a larger power supply) can be upgraded to the DN4500. We've been told that the upgrade deal is a good one, but I of course don't know any financial details. The DN3500 currently supports sr9.7 and both will support sr10.1 later this year. >For that matter, why would the 33-MHz chip in the DN4500 be rated >at 7 MIPS when the 25-MHz chip in the DN3500 is rated at 4 MIPS? >That's a 75% increase in MIPS with only a 50% increase in clock >rate. Do both machines have equivalent caches? > -- David Krowitz (of course this is my understanding, not Apollo's official position.) joel -- Joel Margolese UUCP: ...{attunix,mit-eddie,umix}!apollo!joelm Apollo Computer Inc. ARPA: joelm@apollo.com Chelmsford, MA