melling@cs.psu.edu (Michael D Mellinger) (05/15/91)
Here are two articles that were posted by Michael Slater in comp.arch a couple months ago about the 68040 that I thought the everyone might be interested in. -Mike ============================================================ From: mslater@cup.portal.com (Michael Z Slater) Newsgroups: comp.arch Subject: 68040 Status (long) Date: 3 Mar 91 03:42:03 GMT Organization: The Portal System (TM) A month or so ago, I posted a query on the status of the 68040. Most of the responses I received were from NeXT users, all of whom said they were satisfied with their machines. I also got several defensive responses from Motorola employees, several of whom were sure I could get a bug list from Motorola just by asking. I've asked, and they won't give it to me. I'd still be glad to find one in my mailbox. In any case, the chip does seem to be shipping. The following is the article I wrote for Microprocessor Report on the 68040 status. After the article is a table of SPEC data for the 68040 and other major 32-bit processors. --------------------------------------------------------------------- Despite continued rumors of bugs and production problems, systems based on Motorola's 68040 are finally being shipped to customers. The only system that has been shipped in any quantity (as of early February) is the NeXT machine; several users we contacted all were happy with its performance and stability. HP issued a statement on January 29th that it had begun volume shipments of its 25-MHz 68040-based workstations. Force Computers has announced immediate availability of a single-board VME-bus computer based on the 68040, and several vendors showed 68040 accelerators for the Macintosh at MacWorld Expo. While Motorola and HP claim to be satisfied with the chip's performance, the benchmark results fall short of Motorola's initial claims. The table below shows the SPEC benchmark results. For this table, we have chosen 25-MHz systems for consistency, except for the i860 whose minimum clock rate is 33 MHz. The table shows that Motorola's claims of superiority over the 486 (for integer code) and SPARC don't hold up. With a 128-Kbyte external cache on all three systems, the 68040, 486, and SPARC are all within a few percent for integer SPECmarks. For floating-point, the 68040 is roughly the same as SPARC and the 486 falls far behind. The 68040 is available only at 25 MHz, while 486 systems are shipping in volume at 33 MHz and SPARC systems are shipping at 40 MHz. By the time 33-MHz 68040 chips are available, we expect to see 50-MHz 486 processors. Soon thereafter, superscalar SPARC implementations with separate, on-chip instruction and data caches will appear, and these will be dramatically faster than today's implementations. Motorola characterizes the 68040 as 3.8 times the speed of a 68020 and 3.2 times the speed of a 68030 at the same clock rate on integer programs, and three to seven times faster than a 68020 with a 68882 on floating-point programs. This claim appears to be upheld by the SPEC benchmarks. The performance ratios between HP's 68040 system without cache and its 33-MHz 68030 system, when adjusted for clock frequency, are 3.1 for integer programs and 4.1 for floating-point. Motorola won't release the bug list to us, but sources indicate that it is still lengthy with most bugs relating to the FPU and MMU. Most customers seem satisfied with the work-arounds to the remaining bugs, but there is still some dissatisfaction with the state of the chip. One area of particular concern is how soon Motorola will begin production of the 33-MHz version, since the 25-MHz part lacks the performance to compete with high-end RISC-based workstations. NeXT is gluing heat sinks on the chips, and power dissipation apparently remains a sore point. HP says it considers the current chips to be production-level silicon, and has no plans to upgrade systems in the field when new chips become available. Force Computers, on the other hand, is providing a bug list to its customers and promising to provide upgrades to future versions of the chip. At the recent Montgomery Securities conference, Motorola VP Murray Goldman claimed that the company would ship 400,000 units of the 68040 in 1991. He said that it will be March or April before Motorola catches up with demand for the 68040, and that the 33-MHz version wouldn't be shipped until the second half of the year. (Oddly, HP is promising first shipments of 33-MHz systems in the second quarter.) When the chip was announced just over a year ago, Motorola said that 33-MHz chips would follow the 25-MHz parts by approximately three months, but it now appears that the gap will be six months or more, despite the fact that the 25-MHz parts are over six months behind the schedule given a year ago. Motorola also projected at the introduction that 40-MHz parts would ship about 3 months after the 33-MHz chips. -------------------------------------------------------------------- Vendor Moto Moto Intel Sun MIPS Moto IBM Intel Achitecture 68040 68040 486 SPARC R3000 88000 RS/6000 i860 Cache none 128K 128K 128K 128K 128K 64K/8K none Clock (MHz) 25 25 25 25 25 25 25 33 -------------------------------------------------------------------- Integer gcc 12.4 13.8 13.8 13.8 15.5 17.5 17.8 11.5 espresso 12.8 13.4 12.2 11.6 17.7 19.4 20.7 20.0 li 14.9 15.5 16.8 11.2 20.4 20.7 19.8 17.7 ------------------------------------------------------------------- FP spice2g6 11.8 13.1 8.8 11.4 12.1 12.5 27.6 14.8 doduc 7.7 8.1 5.8 9.5 15.9 10.1 27.7 15.6 nasa7 11.5 12.1 5.8 14.0 18.1 15.2 35.5 45.0 matrix300 10.8 11.5 9.5 14.7 13.8 18.4 21.8 21.5 fpppp 12.1 13.4 7.0 13.1 17.8 14.7 54.7 21.8 tomcatv 8.4 9.1 4.3 8.2 13.9 11.6 75.7 34.0 -------------------------------------------------------------------- Means SPECmark 11.0 11.8 8.8 11.8 16.1 15.2 28.9 20.3 Integer 12.3 12.9 13.3 12.3 17.6 18.3 20.2 16.4 FP 10.2 11.0 6.6 11.6 15.1 13.5 36.7 23.4 ------------------------------------------------------------------ SPEC benchmark results for CISC and RISC systems at 25 MHz (except i860 at 33 MHz). All results as published by SPEC, except for 486 and 860 results published by Intel and 68040 results provided by HP. The systems are the HP 425t, HP 425s, an unspecified 486 system, Sun SPARCstation 330, MIPS RC3240, Motorola 8864SP, IBM RS/6000 Model 530, and an 860 PC add-in board with static column DRAM. Cache sizes shown are external (second-level) caches only, except for RS/6000. MIPS and 88K caches are 64K instruction and 64K data. ------------------------------------------------------------------- Michael Slater, Editor and Publisher, Microprocessor Report mslater@cup.portal.com 707/823-4004 fax: 707/823-0504 874 Gravenstein Hwy. So., Suite 14, Sebastopol 95472 ============================================================ From: mslater@cup.portal.com (Michael Z Slater) Newsgroups: comp.arch Subject: Re: 68040 Status - SPEC table with missing line added Date: 4 Mar 91 16:35:04 GMT Organization: The Portal System (TM) As several observant net readers have pointed out, my table of SPEC results inadvertently deleted the "eqntott" line. The means shown did include all 10 benchmarks. Here is the corrected table: -------------------------------------------------------------------- Vendor Moto Moto Intel Sun MIPS Moto IBM Intel Achitecture 68040 68040 486 SPARC R3000 88000 RS/6000 i860 Cache none 128K 128K 128K 128K 128K 64K/8K none Clock (MHz) 25 25 25 25 25 25 25 33 -------------------------------------------------------------------- Integer gcc 12.4 13.8 13.8 13.8 15.5 17.5 17.8 11.5 espresso 12.8 13.4 12.2 11.6 17.7 19.4 20.7 20.0 li 14.9 15.5 16.8 11.2 20.4 20.7 19.8 17.7 eqntott 9.7 9.8 11.0 12.6 17.1 16.0 23.0 17.8 ------------------------------------------------------------------- FP spice2g6 11.8 13.1 8.8 11.4 12.1 12.5 27.6 14.8 doduc 7.7 8.1 5.8 9.5 15.9 10.1 27.7 15.6 nasa7 11.5 12.1 5.8 14.0 18.1 15.2 35.5 45.0 matrix300 10.8 11.5 9.5 14.7 13.8 18.4 21.8 21.5 fpppp 12.1 13.4 7.0 13.1 17.8 14.7 54.7 21.8 tomcatv 8.4 9.1 4.3 8.2 13.9 11.6 75.7 34.0 -------------------------------------------------------------------- Means SPECmark 11.0 11.8 8.8 11.8 16.1 15.2 28.9 20.3 Integer 12.3 12.9 13.3 12.3 17.6 18.3 20.2 16.4 FP 10.2 11.0 6.6 11.6 15.1 13.5 36.7 23.4 ------------------------------------------------------------------ SPEC benchmark results for CISC and RISC systems at 25 MHz (except i860 at 33 MHz). All results as published by SPEC, except for 486 and 860 results published by Intel and 68040 results provided by HP. The systems are the HP 425t, HP 425s, an unspecified 486 system, Sun SPARCstation 330, MIPS RC3240, Motorola 8864SP, IBM RS/6000 Model 530, and an 860 PC add-in board with static column DRAM. Cache sizes shown are external (second-level) caches only, except for RS/6000. MIPS and 88K caches are 64K instruction and 64K data. ------------------------------------------------------------------- Michael Slater, Editor and Publisher, Microprocessor Report mslater@cup.portal.com 707/823-4004 fax: 707/823-0504 874 Gravenstein Hwy. So., Suite 14, Sebastopol 95472