garyb@gallium.UUCP (Gary Blumenstein) (11/21/89)
In his article, asv@gaboon writes:
.................................................
. I've recently installed an Equinox Megaport 24 port board in my 386/25
. box. (It sounds like alot of ports but their 12 port board hasn't hit
. the streets quite yet.) This board is GREAT! I've been able to
.................................................
I recently took the plunge and got one of these too. I've heard good feedback
about this board from other people. I was also impressed after visiting
Equinox at the recent UNIX Expo. They had a demo showing all 24 ports sending
continuous screenfuls of text to 24 terminals at 38,400 baud. There was no
visible delay between the terminal on the first and last ports and the whole
thing took up hardly any CPU time.
..........................................
. I used to have the Telebit and the mouse on the standard COM ports
. along with the mx80 printer. By using the Equinox board I was able to
. get all my devices on one board and eliminate the COM board and free
. up a slot.
..........................................
Me too. This was the deciding factor as I was able to remove 5 interrupt
driven devices from my box (2 serial, 2 parallel, plus the old multiport
board). Now I finally have available slots and I can go ahead and add another
hard disk controller, Ethernet board, and whatever other goodies I want (funds
permitting :-)) So, after all is said and done, I wound up with 12 more ports
than when I started. I'd have to agree that if you're into saving real
estate, the Equinox is the way to go.
I was also able to get my Trailblazer to do UUCP on one of these ports. I'm
using SCO HoneyDanBer and their supplied dialer program. I gave up a long
time ago trying to get a Telebit to work on other smart i/o boards. I
struggled for years with them and could never get them to work right under
"real world" conditions. At best, I could only get them to work half assed,
either uucp world work and getty would break, or it would be the other way
around. On an Equinox port I can have UUCP poll out and then use the same port
for interactive dial ins at slower speeds or UUCP logins. Modem control works
like it should.
In addition, I'm seeing the best transfer rates I've ever seen with these
boards. I get 1MB compressed news batches coming in consitently at >1500
cps. I'd like to know if anyone else is seeing the same thing? Also,
I don't fully understand how this board can install without using an IRQ.
It is advertised as using "polling" and that this makes a faster setup
when used with a well written driver. Can someone explain this?
One last question; Will the mouse work under VP/ix on one of these ports and
can you send me your cable configuration?
** Discalimer ** I'm not affiliated with Equinox and all this is just my
humble opinion.
--
Gary Blumenstein, UNIX Network Administrator // CIBA-GEIGY CORPORATION, USA
===========================================================================
Voice: (914) 347-4700 7 Skyline Drive, Hawthorne, NY 10532
FAX : (914) 347-5687 uucp: ...{philabs, gaboon}!crpmks!{sysadm, garyb}jrh@mustang.dell.com (jrh) (11/23/89)
In article <458@gallium.UUCP>, garyb@gallium.UUCP (Gary Blumenstein) writes: > One last question; Will the mouse work under VP/ix on one of these ports and > can you send me your cable configuration? > > ** Discalimer ** I'm not affiliated with Equinox and all this is just my > humble opinion. I recently received a Equinox 24 port also, but I have not set it up to test yet, to busy with other projects. However, I am using Computone 8 and 16 port boards, and running a mouse off of these. I would imagine the cabling should be the same. From the 25pin serial out on the multiport, attach a null modem adapter, then a 'gender mender' and a 25-9 converter to connect the mouse. Of course, this doesn't look very pretty. If you have the need for beauty, build a null modem adapter with 25-pin male on one end, and 9-pin male on the other, and attach the mouse. This works with X-windows and the mouse on /dev/ttyi04 (computone node name). The only drawback with this setup is the mouse sometimes leaves trash on the X display, and you need to refresh. If the Equinox is that much faster, without delay on the lines, this problem will probably be lessened. ------------------------------------------------------------------------- James R. Howard !s: cs.utexas.edu!dell!mustang!jrh
jmm@eci386.uucp (John Macdonald) (11/24/89)
In article <458@gallium.UUCP> garyb@gallium.UUCP (Gary Blumenstein) writes: > > [ a very favouable review of the Equinox 24 port serial board ] > > Also, >I don't fully understand how this board can install without using an IRQ. >It is advertised as using "polling" and that this makes a faster setup >when used with a well written driver. Can someone explain this? When an interrupt occurs, the system must stop executing the user process that was being executed, save all register and control information needed to be able to restart the process that might be affected by running the interrupt handling routine, invoke the interrupt routine, check whether a different process has become highest priority because of the interrupt, and restart the (old or newly available) user process. If the interrupt is occuring because of an input character from a com board, then the amount of work to process the input character (the total time spent in the interrupt handling routine) is generally much less than the amount of work involved in getting in and out of the interrupt handler routine (as described in the above paragraph). Most (if not all) systems have a clock interrupt that is used for controlling system internal timing functions (like time-of-day, driver timeouts, and user process alarms). This interrupt generally occurs a number of times per second (60 or 100 is typical) - the number is chosen by the system implementors to provide a good balance between getting the smallest time resolution needed and putting the smallest possible load on the system with the interrupt handling. A polling device driver arranges to have the timer interrupt handler invoke it occassionally (perhaps 10 times per second). Since the interrupt overhead has already been budgeted for the clock interrupt, there is no extra overhead for the system to invoke the polling driver (except a function call, but that is trivial compared to the process state saving and restoring). The polling driver then checks to see if there is any work for it to do, does what it can, and then returns to the clock interrupt handler. This makes the clock interrupt take slightly longer for no reason whenever there is nothing that must be done on for the polled device. However, it saves one or more (possibly a huge number) of process state save and restore operations whenever there is work that it can do. A com board's input is an especially good match for this type of handling. On output you usually can set up a large buffer of characters to send at once, and this can usually be done so as to only generate one interrupt, so there is not much difference (although a polling driver can keep filling the output buffer before it is fully emptied, so there is some gain here). On input, however, such buffering is not easily done. When a character arrives, the system has no way of telling whether it is necessary to process this character immediately, or whether it is save to wait a while, so interrupt-based drivers must usually get an interrupt for every incoming character. (If the driver falls behind, it may end up processing more than one character per interrupt, but in that case the system is being overloaded by the input processing and cannot continue with anything else until the input level slows down.) With a polling driver, there is no need to get an interrupt - the character will be picked up by the next poll. If input is coming at 9600 baud, then there might be 100 characters all processed during a single (tenth of a second) poll. At high baud rates, there might be more processed at once. If multiple ports are active, more might be processed at once also. A human user will not notice a tenth-second delay in echoing (if they do, then the polling could be done more often). A machine to machine transmission will normally be done without echoing, but even if there is echoing, the first character will be delayed and then both input and output will proceed at full speed. Of course, a polling device driver is not always the best choice. To work well it requires that the hardware be able to take care of itself for long periods of time. Thus a com board must have enough intelligence to collect lots of input characters and buffer them until the polling driver happens to come along and accept them. It is generally not appreciated for input characters to be lost because of insufficient buffering. It is also desired that there be enough output buffering to be able to maintain full speed, since sending one character at 9600 baud every tenth of a second looks much the same as full speed transmission at 100 baud. (A processor is not necessarily required to make a polling driver possible, a UART with 100 character input and output buffers could be handled this way as long as the polling frequency is high enough to sustain the transmission rate without loss.) -- 80386 - hardware demonstrating the fractal nature of warts. | John Macdonald EMS/LIM - software demonstrating the fractal nature of warts. | jmm@eci386
asv@gaboon.UUCP (Stan Voket) (11/25/89)
In article <458@gallium.UUCP>, garyb@gallium.UUCP (Gary Blumenstein) writes: > >One last question; Will the mouse work under VP/ix on one of these ports and >can you send me your cable configuration? > Gary, I have a Microsoft serial mouse working on the Equinox Megaport board. It runs fine under both X and vpix. Below is my cabling diagram. Have fun with all those newly freed slots and interrupts! I'm also getting very high throughput on the Telebit. - ---RJ11 to DB25 adapter diagram.--- Microsoft serial mouse to Equninox Megaport RJ11 6 wire. Equinox Megaport, 6 wire RJ11 adapter, reversing cable (not straight through) six wire RJ11 cable end DB25 pins (at mouse) spring contacts on bottom. 6---------------------------20 |_______4 5----------------------------3 4 3----------------------------2 2----------------------------7 1----------------------------8 See Megaport installation Manual pg. 3-7 for diagrams of modular cables abd adapters. -- +----------------------------------------------------------------------+ | - Stan Voket, asv@gaboon - OR - ...uunet!hsi!stpstn!gaboon!asv | | Land Line: (203) 746-4489 TELEX 4996516 - | +----------------------------------------------------------------------+
mel@fleet.UUCP (mel) (11/28/89)
In article <4576@gaboon.UUCP> asv@gaboon.UUCP (Stan Voket) writes: >I'm also getting very high throughput on the Telebit. I understand that the demos shown were for 24 terminals streaming output at 38,400 baud. But what about input? Can an Equinox board take input of binary data in as BBS program to the upload section. Seems like six ports is the max recommended number of serial lines useable with BBS/modems due to the fact that most multi-port boards are optimized for output not input and get overloaded when that many serial lines are receiving data at high speed at one time. Mel Shear !rex!mgse!fleet!mel
jim@tiamat.fsc.com (Jim O'Connor) (11/29/89)
In article <458@gallium.UUCP>, garyb@gallium.UUCP (Gary Blumenstein) writes: > > about this board from other people. I was also impressed after visiting > Equinox at the recent UNIX Expo. They had a demo showing all 24 ports sending > continuous screenfuls of text to 24 terminals at 38,400 baud. There was no > visible delay between the terminal on the first and last ports and the whole > thing took up hardly any CPU time. I'm impressed, too. Enough at least to ask a few more questions about this board. :-) > I was also able to get my Trailblazer to do UUCP on one of these ports. I'm > using SCO HoneyDanBer and their supplied dialer program. I gave up a long > time ago trying to get a Telebit to work on other smart i/o boards. I My Telebits work with the Bell Tech (ah, excuse me, Intel) ACE board in this system, but I didn't get the throughput improvement that I though I would. The number of interrupts caused during a UUCP transfer did decrease dramatically, but the actual speed of the transfers only incresed slightly. > One last question; Will the mouse work under VP/ix on one of these ports and > can you send me your cable configuration? Here's another VP/ix question: will this board let me run VP/ix from a serial terminal like the Wyse60, which runs in PC mode during the VP/ix session? Last question (for now anyway): what physical attachment does the board use? The picture in Unix World looks like it uses a cable that connects to the back of the card and then (I suppose) to a breakout box. If this is basically it, then 1) how long is the cable and 2) what kind of jacks are used in the breakout box? I knew I'd have another question: I heard some mention a 12-port version. Is this a product yet and how much is it, or when will it be a product and how much will it be? I'll quit asking now. Any, and all, answers will be deeply appreciated. ------------- James B. O'Connor jim@tiamat.fsc.com Ahlstrom Filtration, Inc. 615/821-4022 x. 651 *** Altos users unite! mail to "info-altos-request@tiamat.fsc.com" ***
jamesd@qiclab.UUCP (James Deibele) (11/30/89)
In article <134@tiamat.fsc.com> jim@tiamat.fsc.com (Jim O'Connor) writes: >I'm impressed, too. Enough at least to ask a few more questions about this >board. :-) I've ordered one of the darn things on Monday, so I'll be interested in seeing what the thing actually does once it arrives. I talked with two distributors and Equinox themselves before making the decision to buy the board. I've got marketing propaganda on the way, too. In the meantime, what I remember them saying is that the board will do 38.4K output on 24 ports, no problem. I don't about input, and I'm embarassed to say I never even thought of asking (although I expect the output to input ratio to be 20:1 or more). The board is made with two ASIC chips instead of UARTs, chips that were taken from some design that Equinox did somewhere else (they're supposed to be a big deal in mini telecomm) and decided would make a nice design for a board. The 24-port version is $1795 (list), the 12-port version is $1295 (list, and is supposed to be shipping this week). There are supposed to be many different ways of hooking things up. I want to hook up modems, and maybe a terminal or two, so I ordered the DB-25 connectors, which are supposed to be 18" long. If they are what I think I ordered, you should be able to mount the modems on a shelf or rack behind the machine and save some dollars on cabling. More when I get my hands on the thing. -- James Deibele jamesd@qiclab BBSs: (503) 760-1473 or (503) 761-7451 TECHBooks: The Computer Book Specialists --- Voice: (503) 646-8257 12600 SW 1st Beaverton, OR 97005 --- Book reviewers wanted for computer science & electronics - contact us for more information.
garyb@gallium.UUCP (Gary Blumenstein) (12/04/89)
In article <134@tiamat.fsc.com> jim@tiamat.fsc.com (Jim O'Connor) writes: >> using SCO HoneyDanBer and their supplied dialer program. I gave up a long >> time ago trying to get a Telebit to work on other smart i/o boards. I >>>>>>>>>>>>>>>>>>>>>> >My Telebits work with the Bell Tech (ah, excuse me, Intel) ACE board in this >system, but I didn't get the throughput improvement that I though I would. >The number of interrupts caused during a UUCP transfer did decrease >dramatically, but the actual speed of the transfers only incresed slightly. >>>>>>>>>>>>>>>>>>>>>>>>> Keep in mind that your actual throughput is also affected by the condition of the line, the modem, and your software. The board istself may not necessarily affect transfer speed as much as these other factors. However, with conditions being equal, for the exeption of the line quality (which I have no way of measuring). I have seen my highest throughput when using a Telebit on the Equinox. Overall, the worst I'm getting is around 1,250 cps and the highest is over 1,500 cps. I was never able to get 1,500 cps when using another smart i/o board, and I've tried several of these . The figures are taken from xferstats (obviously with the HZ problem fixed) and with files >500K both sending and receiving. As far as interrupts go, there was a good posting earlier which explained this better than I could but because the Equinox utilizes polls every x milliseconds when clock signals occurr, you are basically eliminating extra interrupt servicing and context switching that you get with interrupt driven i/o boards. The ASIC processor is also smart enough to handle simultaneous 38.4K baud to and from all 24 channels unlike most other smart boards which can barely handle more than 9,600 baud if that. >>>>>>>>>>>>>>>>>>>>>>>>> >Here's another VP/ix question: will this board let me run VP/ix from a >serial terminal like the Wyse60, which runs in PC mode during the VP/ix >session? >>>>>>>>>>>>>>>>>>>>>>>>> Yes, the Equinox board supports VP/ix sessions nicley. In addition, VP/ix running under Interactive UNIX can be used over a modem. I've tested this. Two other BIG pluses for the Equinox board that I forgot to mention in my original posting (I found out by accident!) is that their drivers support line disciplines for both SCO Multiview and their office automation application (I forget the name of this). When I enabled the line discipline, Multiview flys! While I havn't measured it yet, there is a HUGE perceived difference in speed when running Multiview on, say a Wyse 60 with this feature enabled. It makes the application completly usable, and its very kind to the cpu. In fact, when Equinox releases their new drivers that have multiple sessions, I probably won't use them because Multiview works so well. The other nice feature about the Equinox is that the drivers allow Foxbase+ to run at 38,400 baud on your terminals without blowing up the screen. >>>>>>>>>>>>>>>>>>>>>>>>> >Last question (for now anyway): what physical attachment does the board use? >>>>>>>>>>>>>>>>>>>>>>>>> They have everything under the sun. What I have, and I suspect will be the most common are two 4 foot PBX cables attatched to what they call "Modular Splitters". This is a little black box containing 12 RJ-11 jacks. Some of the other wiring alternatives you can get include male or female DB25 distribution panels, punch blocks, DB25 or RJ-11 fan out cables, etc. They also sell RJ-ll to DB25 adapters for modems and terminals and the coolest short haul modems I've ever seen. I didn't have to futz around with a breakout box at all. Just plug and go. My question is, why couldn't these boards have been around years ago when we originally configured our systems! >>>>>>>>>>>>>>>>>>>>>>>>> >I knew I'd have another question: I heard some mention a 12-port version. >Is this a product yet and how much is it, or when will it be a product and >how much will it be? >>>>>>>>>>>>>>>>>>>>>>>>> Yes, I'm pretty sure it's shipping now. I had to get the 24 port board because at the time the 12 wasn't available. I'm 99.9% sure they have one now. Also, they're about to release the drivers that do multiple sessions. -- Gary Blumenstein, UNIX Network Administrator // CIBA-GEIGY Corporation, USA =========================================================================== Voice: (914) 347-4700 7 Skyline Drive, Hawthorne, NY 10532 FAX : (914) 347-5687 uucp: ...uunet!philabs!crpmks!{sysadm, garyb}
karl@ddsw1.MCS.COM (Karl Denninger) (12/05/89)
In article <459@gallium.UUCP> garyb@gallium.UUCP (Gary Blumenstein) writes: >In article <134@tiamat.fsc.com> jim@tiamat.fsc.com (Jim O'Connor) writes: >>> using SCO HoneyDanBer and their supplied dialer program. I gave up a long >>> time ago trying to get a Telebit to work on other smart i/o boards. I >>>>>>>>>>>>>>>>>>>>>>> >>My Telebits work with the Bell Tech (ah, excuse me, Intel) ACE board in this >>system, but I didn't get the throughput improvement that I though I would. >>The number of interrupts caused during a UUCP transfer did decrease >>dramatically, but the actual speed of the transfers only incresed slightly. >Keep in mind that your actual throughput is also affected by the condition of >the line, the modem, and your software. The board istself may not necessarily >affect transfer speed as much as these other factors. However, with >conditions being equal, for the exeption of the line quality (which I have no >way of measuring). I have seen my highest throughput when using a Telebit on >the Equinox. Overall, the worst I'm getting is around 1,250 cps and the >highest is over 1,500 cps. I was never able to get 1,500 cps when using >another smart i/o board, and I've tried several of these . The figures are >taken from xferstats (obviously with the HZ problem fixed) and with files >>500K both sending and receiving. Did you try the Specialix board? I have one here on evaluation, and by Gods is this board nice! A wonderful subsystem! Transfer rates? We went from somewhere around 10k to somewhere around 14kbaud on our Telebit when we switched today. Nothing other than the I/O card was changed. >As far as interrupts go, there was a good posting earlier which explained >this better than I could but because the Equinox utilizes polls every x >milliseconds when clock signals occurr, you are basically eliminating >extra interrupt servicing and context switching that you get with interrupt >driven i/o boards. The ASIC processor is also smart enough to handle >simultaneous 38.4K baud to and from all 24 channels unlike most other >smart boards which can barely handle more than 9,600 baud if that. Oh, from the numbers in the Specialix manual I get: 640 KB/sec (bits/sec) [Total throughput per card] or 64 Kilobytes/sec [Total throughput per card] or 64 9600 baud channels unidirectional 32 bidirectional or 32 19200 baud channels 16 bidirectional or 16 38400 baud channels 8 bidirectional or 8 56800 baud channels 4 bidirectional at 100% loading without losing ANYTHING! Yes, it will run 56Kbaud! It uses the "50 baud" setting on stty to do this. We have a couple of terminals which can handle the speed, and it really, truly, does get it without a single hitch. Zmodem transfers are incredible at 56K! With two 56K channels up receiving, and a couple of Telebit connections running at 12-14KBaud there was no noticable degredation in performance. None. Absolutely zero. I am >very< impressed. Evaluation board? Hell no, this one you couldn't tear from my hands! >>>>>>>>>>>>>>>>>>>>>>>>>> >>Here's another VP/ix question: will this board let me run VP/ix from a >>serial terminal like the Wyse60, which runs in PC mode during the VP/ix >>session? >>>>>>>>>>>>>>>>>>>>>>>>>> > >Yes, the Equinox board supports VP/ix sessions nicley. In addition, VP/ix >running under Interactive UNIX can be used over a modem. I've tested >this. Same with the Specialix. >Two other BIG pluses for the Equinox board that I forgot to mention in >my original posting (I found out by accident!) is that their drivers >support line disciplines for both SCO Multiview and their office automation >application (I forget the name of this). When I enabled the line >discipline, Multiview flys! While I havn't measured it yet, there is a >HUGE perceived difference in speed when running Multiview on, say a Wyse >60 with this feature enabled. It makes the application completly usable, and >its very kind to the cpu. In fact, when Equinox releases their new drivers >that have multiple sessions, I probably won't use them because Multiview >works so well. That I haven't tried. On the other hand, the line discipline code is on the card when you're using line discipline 0 (standard). Unlike Anvil, the board doesn't suffer from this offloading at all! >The other nice feature about the Equinox is that the drivers allow >Foxbase+ to run at 38,400 baud on your terminals without blowing up >the screen. Try it at 56K. You think 38,400 is nice? 56K is like being on the console! You really have to see it to believe it. >>>>>>>>>>>>>>>>>>>>>>>>>> >>Last question (for now anyway): what physical attachment does the board use? >>>>>>>>>>>>>>>>>>>>>>>>>> > >They have everything under the sun. What I have, and I suspect will be >the most common are two 4 foot PBX cables attatched to what they call >"Modular Splitters". This is a little black box containing 12 RJ-11 jacks. >Some of the other wiring alternatives you can get include male or female >DB25 distribution panels, punch blocks, DB25 or RJ-11 fan out cables, >etc. They also sell RJ-ll to DB25 adapters for modems and terminals and >the coolest short haul modems I've ever seen. I didn't have to futz >around with a breakout box at all. Just plug and go. My question is, >why couldn't these boards have been around years ago when we originally >configured our systems! The Specialix connector scheme is a female DB25. The I/O modules (8 or 4 ports each) plug into >each other< allowing from 4-32 ports on one I/O interface. Each "interface" board in the computer (there can be 2 total) has a 20Mhz (!) Z280 onboard and up to 1/2Meg of buffer RAM. One of the nicest things about this card is that you can add additional ports (given that you aren't beating the snot out of them!) until you have 32 (given all 8-port modules) on a single I/O interface, but buy them 8 or 4 at a time. Thus this card is >truly< expandable as your needs grow, something that most other intelligent cards only dream about. This thing comes with drivers for 386/IX 2.x and SCO Xenix. Others are rumored on the way. Pricing runs from ~$1k - $2500 depending on the number of ports; the 16-port version is somewhere around $1700. Do I sound like a salesman? Not yet we're not. But if this thing continues to live up to it's initial promise, we'll be carrying them shortly (as in as fast as we can get them in the building). I can't say enough nice things about these ports boards; they are real wonders. Anyone want to buy an Anvil Stallion-8 card for somewhere around $700? I have one for sale :-) -- Karl Denninger (karl@ddsw1.MCS.COM, <well-connected>!ddsw1!karl) Public Access Data Line: [+1 708 566-8911], Voice: [+1 708 566-8910] Macro Computer Solutions, Inc. "Quality Solutions at a Fair Price"
garyb@crpmks.UUCP (Gary Blumenstein) (12/06/89)
In article <3323@qiclab.UUCP> jamesd@qiclab.UUCP (James Deibele) writes: >>>>>>>>>>>>>>>>>>>>>>>>> >[..] In the meantime, what I remember them >saying is that the board will do 38.4K output on 24 ports, no problem. I don't >about input, and I'm embarassed to say I never even thought of asking (although >I expect the output to input ratio to be 20:1 or more). >>>>>>>>>>>>>>>>>>>>>>>>> It's more like 1:1. The ASIC provides baudrate processing of 38.4K x 24 channels on output, 38.4K x 24 on input, and a total aggregate baudrate for both input and output processing at 38.4K x 48. In other words, the processor handles these baudrates on all 24 channels for both input and output simultaneously and with very little cpu overhead because no extra interrupt servicing is required. It's going to be tough to beat that with a stick, let alone with some of the other serial boards and drivers out there. >>>>>>>>>>>>>>>>>>>>>>>>> > The board is made with >two ASIC chips instead of UARTs, chips that were taken from some design that >Equinox did somewhere else (they're supposed to be a big deal in mini telecomm) >and decided would make a nice design for a board. >>>>>>>>>>>>>>>>>>>>>>>>> This is what I heard too. The ASIC was taken directly from thier telecom switches. From what I understand, (and I'm no expert!) the underlying philosophy with the data switch market dictates no such thing as claimed and delivered baud rates. In addition, distictions between input and output baudrates are simply not made. Data switches are made to be totally invisible in the data path when it comes to routing calls to various destinations. The application specific processor approach is just an existing design philosophy that was finally carried over to the AT bus, multi-serial i/o market. Another thing I recently learned about the board is that it makes very efficientuse of the bus. My vendor tested it at 12 mHz, 0 wait states and Equinox rates it up to 16 mHz. >>>>>>>>>>>>>>>>>>>>>>>>> >More when I get my hands on the thing. >>>>>>>>>>>>>>>>>>>>>>>>> I know there are several others on the net who have tried this board and like it. I'd be interested in hearing about your experiences with it. * * Disclaimer * * Neither myself or this company has any affiliation with Equinox whatsoever. -- Gary M. Blumenstein, UNIX Network Administrator // CIBA-GEIGY Corporation USA ----------------------------------------------------------------------------- Phone (914) 347-4700 7 Skyline Drive, Hawthorne, NY 10532 FAX (914) 347-5687 UUCP ...uunet!philabs!crpmks!garyb
richard@pegasus.uucp (Richard Foulk) (12/13/89)
>>[..] In the meantime, what I remember them >>saying is that the board will do 38.4K output on 24 ports, no problem. I don't >>about input, and I'm embarassed to say I never even thought of asking (although >>I expect the output to input ratio to be 20:1 or more). >>>>>>>>>>>>>>>>>>>>>>>>>> > >It's more like 1:1. The ASIC provides baudrate processing of 38.4K x 24 >channels on output, 38.4K x 24 on input, and a total aggregate baudrate for >both input and output processing at 38.4K x 48. In other words, the processor >handles these baudrates on all 24 channels for both input and output >simultaneously and with very little cpu overhead because no extra interrupt >servicing is required. It's going to be tough to beat that with a stick, let >alone with some of the other serial boards and drivers out there. > That seems mighty optimistic, at least in the real world. Sure the board may be able to pump data at some wonderful rate in some ideal situation, but real world applications aren't that cooperative. Output with Unix is usually no problem -- high performance is relatively easy. Input with Unix is just the opposite. So how do these boards perform in the real world, handling real tasks. How do they compare to their competition? Unix I/O is the sort of thing that you can easily throw all kinds of expensive hardware at and not gain a whole lot. Or employ some careful tuning and make all the difference in the world. -- Richard Foulk richard@pegasus.uucp richard%pegasus.uucp@nosc.mil richard@uhccux.uhcc.hawaii.edu