HD/FD autonegociation

I am not sure what you mean by "at the time where the both modes existed" ?

Half duplex came first. When full duplex was first implemented, standards in autonegotation did not exist, and it was not uncommon to see early devices that could not autonegotiate at all (but which could be configured for either half duplex or full duplex for any given port.) I would need to check historical references to find out if autonegotation was a goal in the development of full duplex or whether autonegotation was later than full duplex.

Bonjour Walter,

Sorry for the late, and thanks for the comments.

I meant the time where the collision domains were common, with hubs separated by switches and routers.

OK, I can understand that.

Full duplex come in march 1997 with 802.3x, and autonegociation with

802.3u in june 1996. So it seems that the automatic selection of the rate was also important, and duplex selection has probably raised that fact.

I read this morning, in an old book, that autonegociation was essential, necessary, with 1000Base-T. Was it correct in 2000, and is it always correct today?

Thanks for the new comments, best regards, Michelot

That's when the standards were approved.

The Synoptics 28115 and 28104 switches, and the Intel EtherExpress Pro/100 ethernet card were jointly announced May 9, 1994, with June 1994 delivery for the 28115 and August 1994 delivery for the EtherExpress. The announcement claimed that they were the world's first 100M ethernet products.

("Intel, SynOptics announce new fast Ethernet devices NetworkWorld" If you google "synoptics 28115 released" then it is the second link, but you will have to ask google for the cached version as the primary is no longer available through that link.)

Yes, as far as I know; that's what the experts here keep saying.

They can *claim* that, but the world's first 100M Ethernet products were announced and shipped in 1992 by Grand Junction Networks (later acquired by cisco).

Yes. Auto-Negotiation is (and always has been) mandatory for 1000BASE-T.

-- Rich Seifert Networks and Communications Consulting 21885 Bear Creek Way (408) 395-5700 Los Gatos, CA 95033 (408) 228-0803 FAX

Send replies to: usenet at richseifert dot com

Bonjour Rich, and Walter

Could you pleased details a little bit some reasons, or some elements to understand that? Would it also valid, the same thing for 10Gbase-T?

Thanks, best regards, Michelot

In 1000BASE-T, it is not just the data rate and duplex settings that are being configured. Auto-Negotiation is also used to select the Clock Master (there is one master and one slave on each link), and a litany of other parameters related to the link signaling method.

-- Rich Seifert Networks and Communications Consulting 21885 Bear Creek Way (408) 395-5700 Los Gatos, CA 95033 (408) 228-0803 FAX

Send replies to: usenet at richseifert dot com

(snip)

Do you mean 100baseVG?

I believe I have an HP NIC that is 10baseT/100baseVG, which of course won't work at 100baseTX like all the others that I might try to connect it to.

-- glen

No, I mean 100BASE-X (what would later become 100BASE-TX). (HP was the only firm assert the disingenuous claim that VG was somehow "Ethernet".) Grand Junction shipped a 24+2 port switch (24 ports of 10BASE-T, 2 ports of pre-standard "100BASE-TX"), along with a NIC for connecting a device (typically a server) to the 100 Mb/s ports. We demo'ed it at InterOp at the Moscone Center in San Francisco in October 1992.

[Full disclosure: While I was never an employee of Grand Junction, I did provide consulting services to them at that time.]

-- Rich Seifert Networks and Communications Consulting 21885 Bear Creek Way (408) 395-5700 Los Gatos, CA 95033 (408) 228-0803 FAX

Send replies to: usenet at richseifert dot com

Bonjour Rich,

Thanks. After my post, I read that in a old reply from you.

Could I ask you more details about the clock negociation :

(1) It seems that only 100Base-T2 et 1000Base-T are concerned by the clock negociation. What is the reason?

(2) 100Base-TX has no clock negociation. Is it not possible to have a

100Base-TX device with a local clock?

(3) It seems that the scrambling operation is only in relation with the clock negociation (certainly to have a sufficient density of transitions for the clock recovering)

(4) Today, is it common to have master 1000Base-T devices, or the peer- to-peer exchange is only used?

Sorry for these basic questions, best regards, Michelot

100BASE-T2 and 1000BASE-T transmit and receive on the same pair(s) at the same time. Thus, the receivers must accurately cancel out any "echoes" (reflections caused by impedance discontinuities and crosstalk) that appear on each pair from their own transmitted signals on that, and the other pair(s). It is *much* easier to do this if the transmitted and received signals are synchronous, i.e., using the same clock, because then then only phase differences (as opposed to frequency differences) need be accounted for. 10BASE-T and 100BASE-T do not transmit and receive on the same pairs at the same time, and so the echo cancellation is not needed, and the clocks at the two ends do not need to be synchronized.

Quite the opposite. In 100BASE-TX, each device transmits to the other using its local clock, which is not frequency-locked to the other station. The clocks in each direction are independent.

I'm not sure what you are asking here. Scrambling is performed to ensure both sufficient transition density, and to spread the transmit energy more uniformly across the spectrum (to reduce EMI). This function is the same whether the clocks are synchronized or not.

As for the PHY clock, there is always a Master and a Slave on each twisted pair link. Which device becomes the Master and which the Slave is resolved during Auto-Negotiation. The "master" need not be a more powerful computer than the slave; e.g., a laptop may be the master in a link connecting it to a mainframe computer.

-- Rich Seifert Networks and Communications Consulting 21885 Bear Creek Way (408) 395-5700 Los Gatos, CA 95033 (408) 228-0803 FAX

Send replies to: usenet at richseifert dot com

Bonjour Rich,

How do you do to be so clear that we can understand? I was looking for some information in Google, and very sad to be not satisfied by the descriptions.

...in base band. For 10Pass-TS e.g. transmit and receive are also simultaneous on the same pair, but in different frequency bands.

Now, I'm understanding

OK

Very interresting.

Thanks for your time, Michelot