home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!haven.umd.edu!decuac!pa.dec.com!granite.pa.dec.com!ajc
- From: ajc@pa.dec.com (AJ Casamento)
- Newsgroups: comp.arch
- Subject: Re: COMPAQ PROPOSED SCALABLE I/O ARCHITECTURE
- Date: 16 Dec 92 11:22:32
- Organization: Digital Equipment Corporation
- Lines: 128
- Message-ID: <AJC.92Dec16112232@thendara.pa.dec.com>
- References: <1992Dec15.171554.2781@twisto.eng.hou.compaq.com>
- <1992Dec15.194637.10009@eng.umd.edu>
- NNTP-Posting-Host: thendara.pa.dec.com
- In-reply-to: sysmgr@king.eng.umd.edu's message of 15 Dec 92 19:46:37 GMT
-
-
-
- In article <1992Dec15.194637.10009@eng.umd.edu> sysmgr@king.eng.umd.edu (Doug
- Mohney) writes:
-
-
- >> In article <1992Dec15.171554.2781@twisto.eng.hou.compaq.com>, leigh@croatia.
- >>eng.hou.compaq.com (Kevin Leigh) writes:
-
- >> [wondering what Yet Another Standard would do for anyone, since EISA has
- >> limited penetration and Local Bus/PCMICA <sp> have just emerged and are
- >> hitting the streets...]
-
- > You guys ever thought of licensing TurboChannel from DEC and building a
- > Pentium box/server around it?
-
-
- > I have talked to Ehud, and lived.
- > -- > SYSMGR@CADLAB.ENG.UMD.EDU < --
-
- Doug,
-
- Just to clear up one point in your response, there is no license for the use
- of TURBOchannel. It is completely free to both System and Option vendors. In
- point of fact, Digital no longer controls the specification. For almost the
- past two years the specification has been controlled by the TURBOchannel
- Industry Group. Specification changes are voted upon by the membership and are
- approved by the Steering Committee (9 members of which DEC has 1 vote). In any
- case, there is no hinderance for anybody using the TURBOchannel technology.
-
- Actually, I personally find a great deal of merit in the COMPAQ proposal. I
- quite agree that a modular I/O design (whether one uses the terminology of a
- "star topology" or, as we do here in DEC, "radial") has a great deal to
- recommend it. Consider a TURBOchannel implementation like this:
-
- ------------ -------
- | | | |
- | Memory |========| CPU |
- | | | |
- ------------ -------
- | |
- | <--- DMA |
- | |
- ---------- | <--- PIO
- | | |
- | Tc/CTL |--------------
- -------| |_________________________
- / ---------- |
- / | | |
- / | |___________ |
- / | | |
- / | | |
- ------- ------- ------- -------
- | | | | | | | |
- |sysIO| | Tc0 | | Tc1 | | Tc2 |
- | | | | | | | |
- ------- ------- ------- -------
-
-
- Where each of the links to the TURBOchannel Controller (Tc/CTL) is a standard
- 25MHz 32bit wide TURBOchannel. The DMA path could be a 50MHz 64bit wide version
- of TURBOchannel (in order to balance the I/O requirements. This would generate
- a system with an aggregate bandwidth of 400MB/s of I/O. The key point here is
- obviously the Tc/CTL ASIC as it will require careful design to support this
- kind of bandwidth (additionally, one could imagine a true dual ported memory
- being used such that the CPU is not starved when DMA access is underway).
-
- This is a potential high end workstation style design where:
-
- sysIO - Is a collection of all of the base module I/O. SCSI, Ethernet, ISDN,
- Audio, ACCESS.bus, Synch/Asynch comm, Centronics and so on. At 100MB/s.
-
- Tc0 - Is a completely radial TURBOchannel at 100MB/s.
-
- Tc1 - Is a completely radial TURBOchannel at 100MB/s.
-
- Tc2 - Is a completely radial TURBOchannel at 100MB/s.
-
- The benefits of this in terms of really high speed graphics or networks is an
- obvious one. Given a sufficiently powerful design of the Tc/CTL ASIC, an option
- should be able to depend upon a given bandwidth and latency (response to a DMA
- ~rReq or ~wReq ought to be immediate). Video teleconferencing and other such
- high bandwidth consumers ought to do very well with such a system.
-
-
- Given that one needed to increase the bandwidth even further, one could make
- the radial TURBOchannels (Tc0, Tc1, Tc2) into speed selectable versions (this
- is briefly described in a TURBOchannel presentation I put up on gatekeeper over
- a year ago in gatekeeper:pub/DEC/TriAdd/TURBO.PS) that can run at either
- 25MHz or at 50MHz. This would insure all the old options would be compatible
- and that any new option could have a full 200MB/s throughput if it is needed.
- We know that TURBOchannel works at this kind of throughput because we use
- variants of TURBOchannel (different versions of datapath width and freuency)
- internally on several of our designs.
-
- A low end version of this (with an enclosure something like a DEC 3000/500)
- might have a single TURBOchannel that covered the base system I/O and three of
- the TURBOchannel slots (for all of the slow speed stuff that won't break 40MB/s
- or so) and could still have three complete radial (speed selectable or not)
- TURBOchannels for the other three option slots (the DEC 3000/500 has 6 slots
- that are customer configurable).
-
- This kind of an implementation puts a one-time cost into the Tc/CTL versus a
- shared-wide fully bussed interconnect which puts such a cost into each of the
- add in cards as well. Plus, with a full bussed interconnect you will be using
- the same resource for your SCSI system disk that you are trying to run high
- speed video images across.
-
- In any case, none of the above discussion guarantees that Digital is (or isn't)
- developing anything like this. It's just my own personal views on the topic.
-
- Sorry for the long soapbox discussion. I had originally intended to just clear
- up the use of the term license and I got carried away.
-
- I hope that helps.
-
- Thanx,
- AJ
-
-
- **********************************************************************
- * AJ Casamento "The question is not whether or *
- * Digital's TRI/ADD Program not the opinions are mine; but *
- * 529 Bryant Ave. PAG-2 rather, which of my personalities *
- * Palo Alto, CA 94301-1616 do they belong to?" *
- * 415.617.3460 *
- * ajc@pa.dec.com *
- **********************************************************************
-