home *** CD-ROM | disk | FTP | other *** search
- Path: lou.teclink.net!usenet
- From: rad@teclink.net (rad)
- Newsgroups: comp.sys.amiga.programmer
- Subject: Re: Processors
- Date: 28 Mar 1996 04:40:44 GMT
- Organization: TECLink Internet Services: info@TECLink.Net
- Message-ID: <2862.6660T1334T2954@teclink.net>
- References: <4iri6d$lim@columba.udac.uu.se> <1665.6656T1237T1226@teclink.net> <4j5rm3$cq9@brachio.zrz.TU-Berlin.DE>
- NNTP-Posting-Host: tc1_25.teclink.net
- X-Newsreader: THOR 2.2 (Amiga;TCP/IP) *UNREGISTERED*
-
- On 25-Mar-96 10:14:27, Philipp Boerker <rawneiha@w350zrz.zrz.TU-Berlin.DE>
- wrote:
- >rad@teclink.net (rad) writes:
-
- >>On 21-Mar-96 12:31:09, Kristofer Maad <m93kma@sabik.tdb.uu.se> wrote:
- >>>No, it doesn't. This has been said a thousand times on this and other
- >>>newsgroups: The '040 uses a double internal clock _only_ for pipeline
- >>>synchronization purposes. No instructions are performed in an odd
- >>>number of 80MHz-cycles. The fastest instruction takes one
- >>>40MHz-cycle. On the 486, though, the processor is _really_ clock
- >>>doubled, so some instructions take only one 66MHz-cycle to complete.
-
- >>Ummm, you're only half right. The integer unit runs at the rated speed;
- >>however, the FPU is run at the higher clock rate. Check out the 68040
- >>user's manual and you should see that several FPU instructions take
- >>fractional numbers of cycles in the Execution stage... (FDIV 37.5, FMOVE,
- >>1.5 or 4.5 FABS, FNEG 4.5) It has been confirmed by Motorola Engineers on
- >>comp.sys.m68k that the FPU is based on the "double" clock.
-
- >No, there is no double clock, there is only a clock, that is delayed half a
- >cycle. The fractional cycle is only a timing problem that needs the second
- >clock for handling! Believe me!
-
- Ok, you're going to make this difficult. First from the 68040 User's manual
- MC68040UM/AD Rev. 1 Page 5-2 to 5-3 Table 5-1 "Signal Index":
-
- signal name Mnemonic Function
- Bus Clock BCLK Clock input used to derive all bus signal timing
- Processor Cl PCLK Clock input used for internal logic timing. The PCLK
- frequency is exactly 2x the BCLK frequency.
-
- I refered to the PCLK as the "double" clock since the term would be more
- meaningful to those not familiar with the data-book. The CPU speed rating is
- the rated BCLK frequency. Therefor there IS a clock at twice this frequency.
- There however is no mention in the user's manual of a clock delayed half a
- cycle. Perhaps you could give a reference for this?
-
- Until I see some better explanations for what I've seen, I'll believe the
- poster on comp.sys.m68k who was posting from a Motorola site with the uP
- division listed in his .sig. He quite plainly stated that the IU was based
- off the BCLK and the FPU was based off the PCLK. This makes the fractional
- timings for some FPU instructions more logical to me. I would like to know
- how exactly integer and fraction cycle instructions can occur simultaneously
- without effectively operating at the double clock frequency?
-
- ---------------------------------------------------------------------------
- - Richard Deken EMail: (personal) rad@teclink.net -
- - VLSI Design Engineer (AuE) rad@aue.com -
- - Advanced Microelectronics PGP public key available -
- ---------------------------------------------------------------------------
-
-