Expected network throughput on Mediator?

[kolla]

Well, I'm sorry to see you strugling with your network, as well as your grammar.

[Oldsmobile_Mike]

Well, I'm sorry to see you strugling with your network, as well as your grammar.

Unnecessary and stupid personal attack. Thanks for keeping this thread classy! Sorry I get my head bit off for trying to correct someone, I know how much this site likes actual facts! :roflmao:

[kolla]

Oh, facts. Too bad I'm far away from any hardware to do performance tests myself for you, but here is a paper on throughputs with various Windows releases, notice how most numbers are well above 94% efficiency.

http://www.zti-telecom.com/BrochuresN/LanTraffic%20V2%20%28V2.6%29%20Performances%20Measurements.pdf

As for "personal" attack, consider it a tongue in cheek attack on all your countrymen, who seem to strugle so hard with what you consider your own language (which btw, it is not)

[danbeaver]

Since some people have resorted to personal insults, the above mentioned document references results using a specific benchmark in a test lab.

The English language is a rough development of old German (both "Anglo" and "Saxon" are from German regions) with addition of Latin derivations and words that transcend country boundaries.  Spelling and grammar differ from the US to that used in Canada and the UK. Aside from some small typing errors that could be colloquialisms, his grammar is fine.  Some people's use of "effin" for fornication is not, or at least I would never have turned in a document in undergrad or grad school with that word in it; furthermore it is insulting.

[duga]

http://en.wikipedia.org/wiki/Ethernet_frame#Maximum_throughput

http://packetpushers.net/tcp-over-ip-bandwidth-overhead/

"If you add Ethernet (and VLAN tagging) into the mix (see the calculations from Wikipedia here) then the throughput of a 100Mb link is 100 X 0.9733 (TCP/IP efficiency) x 0.9728 (Ethernet (with tagging) efficiency) which equals 94.68Mbps, which I assume means the combined efficiency is 94.68%."

[Duce]

Thanks for sharing the PacketPushers article, Duga - very well written and I'd hadn't seen it before.

[danbeaver]

Well, at least the 94 Mbs is reachable in a closed laboratory setting between two PC's with no hubs and using ttcp (Test TCP) provided the "sender" only sends and the "receiver" only receives.

But, to the best of my knowledge we were talking about Amiga and their Ethernet cards and throughput over the Zorro 2/3 bus, multitasking, error checking and the like.

As stated, and kindly read the second article for "Assumptions" and "Other Overheads" since this is not a real-world calculation.

[olsen]

http://en.wikipedia.org/wiki/Ethernet_frame#Maximum_throughput

http://packetpushers.net/tcp-over-ip-bandwidth-overhead/

"If you add Ethernet (and VLAN tagging) into the mix (see the calculations from Wikipedia here) then the throughput of a 100Mb link is 100 X 0.9733 (TCP/IP efficiency) x 0.9728 (Ethernet (with tagging) efficiency) which equals 94.68Mbps, which I assume means the combined efficiency is 94.68%."

The figures quoted refer to efficiency in terms of bandwidth used by and available for the payload (as opposed to the protocol overhead), and they specifically exclude the work needed to get the data from one place to another, which would also have to account for the effects of memory management in the TCP/IP stack and whatever the operating system has to do at the time.

That is to say the 97% efficiency figure probably does not translate into practical throughput of 97 MBit/s on Fast Ethernet. It's likely a bit lower.

[psxphill]

Sheesh, FFS!!

He said FastEthernet *interfaces*, not network, stop twisting this around!

The question asked was about ftp transfers using a Mediator, the answer he gave is a theoretical maximum you can push through an Ethernet PHY when you don't have annoying things like a cpu/ram/bus/os to get in the way.

His answer was unhelpful at best, misleading at worst.

[olsen]

The question asked was what is the expected network throughput on Mediator, the answer he gave is a theoretical maximum you can push through an Ethernet PHY when you don't have annoying things like a cpu/ram/bus/os to get in the way.

You should include "physics" in that list, too  Signal propagation speed and latency need to be taken into account as well.

[kolla]

What this started with was Oldsmobile claiming that 94-97Mbit on a fastether interface is BS. It is not, and you don't need a "lab setup" to prove that, just hook a fairly modern machine with a fastether NIC to a modern switch and do your measuring. There is no difference in a "lab" and real world here, the equipment is the same.

[Oldsmobile_Mike]

What this started with was Oldsmobile claiming that 94-97Mbit on a fastether interface is BS. It is not, and you don't need a "lab setup" to prove that, just hook a fairly modern machine with a fastether NIC to a modern switch and do your measuring. There is no difference in a "lab" and real world here, the equipment is the same.

I was really just going to ignore you, but please remind us again how all your trolling about "theoretical maximum throughput" helps this guy with his Mediator question at all?

[paul1981]

Sheesh, FFS!!

He said FastEthernet *interfaces*, not network, stop twisting this around! Today equipment is typically geared toward Gbit speeds, and yeah, 97% efficiency on 100Mbit interfaces is quite doable - nowhere did he mention protocol overhead, because that is irrelevant when it comes to measing efficiency of *hardware*. You chose to use a slow protocol with lots of overhead, that your effin problem!

But you haven't explained what part the Fast File System plays in this?

[danbeaver]

But you haven't explained what part the Fast File System plays in this?

Paul1981 is quite correct (but more to the point about Amiga HDD transfers); those people spitting out "Theoretical" and "Lab Results," fail to remember that the "lab" part is from RAM to Ethernet to RAM.  The original question was on an Amiga doing FTP which involves HDD to Mediator to Ethernet to Server HDD.  These Real World transfers eat CPU cycles and are limited by bus speeds and finally the file system (to some extent), further the server is an unknown and we can't guess the efficiency of it.

I'm going to "guess" that if Jens can get a lab result under optimal conditions for his X-surf 100 of 1600kBytes/s [1.6 MB/s] , then a plain 8139 card sitting on a Mediator that transfers to a Z3 bus and then to a hard drive is going to achieve a value significantly less.