Slashdot Mirror
LinModems?
Polo was the first of several to send us an article over at LinuxWorld about PC-Tel announcing LinModems, eg, software modems that can run under Linux as well. Some interesting comments in there about hardware modems being a "Luxury" item. Kinda amusing.
Although a hardware modem can cost up to five times more than a software modem,t hey are relatively cheap, with a current price tag at $100... ... 'I think that [in the future], a hardware modem will be a high-end luxury item'"
Okay. So even though hardware modems are *already* fairly inexpensive and will continue to go down in price, they are going to be high-end luxury items? $100 is not a high-end luxury price. The only way anyone could consider a modem high end is if everything else available was complete garbage. Of course, this *is* the prediction of a winmodem software manager
Personally, I fail to see how, with DSL, cable, and ISDN becoming cheaper and more available, any modem could be considered "high end". If you want high end connectivity, why modulate and demodulate a bunch of analog signal?
Traditionally, software modems have a bad reputation in the linux community. In fact, they've earned the nickname "WinModems" because many are optimized to work with the Microsoft Windows operating system, and refuse to cooperate with any other OS.
Funny, I always thought they "earned the nickname WinModems" because they say "WINMODEM" on the frigging box. It's the fricking model name, not a nickname.
They do not just have a bad reputation in the linux community, they have a bad reputation among just about anyone with any idea how modems work. And it's not because they aren't very cross-format, it's because they are lousy modems. I work at an ISP, and I can say with some certainty that the loathing and contempt our tech support has for WinModems is *not* because they only work on windows, but because they barely work at all
Phew. Felt good to get that out. So anyway. What planet did these people say they were from?
Another damned comic
+++ NO CARRIER
Really, this is a good thing. What is so great about Linux is the plethora of alternatives we can afford. Software modems are taking a heavy beating on Slashdot mostly beacuse people prefer not to use them. However, why is it that those same people are not complaining about Linux running other inferior hardware? Example: Why are attempts to run Linux on 386s not flamed? Why are we not all running Linux on Alphas when that is an architecture regarded by many as superior to x86?
My impression is that a company that wants to give another option to us is a good thing. Look at the fight we gave Nvidia to release Linux drivers. We fought and we won. We didn't say "I hate hardware acceleration." Now a company comes along to give us a product and we meet it with criticism.
Personally, I would rather chew sand than buy a software modem but who am I to decide what is best for all of us? As long as I have a choice, anyone that wants to give us more hardware is good.
-Clump
Well there goes my dream of bying a computer with a good modem
Accually he was probably a pompous ass cause he was frustrated. I know a person who works tech support who frequently gets calls about lexmarks (he doesn't work at lexmark tech support btw) and states that they are pure evil. Great for a while, but once they start acting weird there is almost nothing that can be done from the typcial user level over the phone.
My core disagreement with you is that you fail to take the past into account. Look at Unix. When more people adopted it, it became more supported and more powerful. The problem set in when it fragmented and became proprietary.
Linux is open-source and will remain as such. Myself and others certainly will not stand by and let companies get out of control. As long as the source code is there, nobody can monopolize Linux.
Further, paranoia over more people accepting Linux will get us nowhere. We can't say "I am afraid of your modem. Please keep it working only for Win32. I am afraid the Government will own Linux." Seriously, we need to realize that if we have the code, we have control. Anything too restrictive will be rejected or changed by the community.
-Clump
You mentioned the following:
For a perfect software modem, yes, this is true. But for a simple software modem, there are many things you can do to be "good enough", including the following:
After all, Win9x has the same problems as Linux does in this regards (eg. no guaranteed hard-real-time scheduling), yet software modems seem to function passibly in such an environment. If anything, Linux would probably do better than Win9x at this same task. In any case, missing a real-time deadline with a soft-modem would look like lag to the end user and little else. Annoying, but not fatal.
Now, I'm not running out and signing up to buy a soft modem (particularly since I'm about to get DSL service, obviating the need for a POTS modem almost entirely), but there are some interesting ideas that could make LinModems popular, depending on how open PC-Tel is.
- Software modem lights that actually work. (The ones I've seen in the past were worthless for actually diagnosing modem troubles. Eye candy at best.)
- Real-time modem connection quality statistics, such as number of retrains, current symbol rate, etc... perhaps via an entry in
/proc. The more detail, the better. - General on-the-fly control of modem parameters, such as explicitly requesting fallback/fallforward. For instance, lower symbol rates tend to work alot better for interactive sessions since latency is lower, but higher symbol rates tend to be better for downloading. You could actually request the modem to fall forward or fall backwards by signalling the driver directly. You can't really do that with an external modem.
--Joe--
Program Intellivision!
I don't see the problem. All drivers run in kernel mode, so by your logic we shouldn't support 3d video cards, since that's just one more driver running as root we don't need.
If you wish to do so, you can spend an extra $50-$75 on a hardware modem. For those who want to spend the $50-$75 on something else, a WinModem/LinModem is a viable option.
10 PRINT CHR$(205.5+RND(1)); : GOTO 10
PC-Tel seem to be the modem of choice for things such as e-machines, with cyrix processors that can't seem to provide the power the modem needs to operate. The *fastest* I have personally seen one of these connect(with a live connetion) is 28.8. Any noise on the phone line kills the connection. I feel like breaking out in tears as soon as I hear PC-Tel.... it's going to be a long call that I can't end quickly without being a jerk.)
As soon as someone calls up with one of these pieces of crap I tell them to buy a new modem.
As a ISP tech, I take all the crap from people saying they have a 56k modem and wondering why they are only conecting at 24000. I would just once love to tell them to go out, stick a crowbar in their wallet, and spring for a 150 dollar USR external.
It is a pain in the ass to explain to someone that since they just spent 2500 dollars on a new PIII from $major_manufacturer (that'll never be used for anything more then talking to friends online), perhaps they ought to spend another hundred on their connection to the outside world
The exact same thing goes for Rockwell HCF's (hint for any tech support people out there, "+MS=V34" is your friend) and Diamond Supras...even the externals.
If there is going to be a Lin-modem, it should at least be on a modem with a DSP, not these HSP pieces of crap that are currently being put out.
My rant is done, feel free to moderate, as an ISP bob, I'm just passionate about modems.
Mikesch -- admitting my main system is currently a Cyrix with an LT Win-Modem (not all that bad... for a WinModem)
When the CPU has been idle for some time, Linux (and other well-engineered OSs) issue the hlt instruction. It puts the CPU in a 'sleep' state from which it gets out in any interrupt. After returning from the interrupt, it already 'forgot' it was in an sleep state and continues working normally.
.com with only 18 hlts. When I run it, it waited for a second (timer is 18.2/sec under DOS, 100/sec under Linux/i386) and exited (ret to int 20 on psp).
When the CPU is idle, this means no proccess has anything to do, and the only things left to do are answering to keyboard, timers, modems, sound DMA exaustion, IDE read/write completed, mouse movement, and other things like that. And they all use interrupts, so it gets out of the halt state without losing any cycles (besides, the hlt is only used after the cpu has been idle for some time, so if you are fully using the cpu no hlt will appear).
When the proccesses are idle, they are all waiting for some event (blocking reads/writes and sleeping for some specified time being the most common). All those events can only happen after a ISR (so it'll wake up after a time interrupt, or after a disk interrupt, or after a keyboard interrupt, or after a NIC interrupt, etc.). So the CPU has actually nothing useful to do until an ISR happens.
Once I wrote (under DOS, before I got Linux) a
I hope someone understood.
Without a free software driver, we're stuck with a dumb piece of hardware and a closed driver. No thanks. The embedded intelligence in a modem doesn't cost more than $10, so I doubt that would fly.
I'd really like to encourage these guys...
Thanks
Bruce
Bruce Perens.
Nothing new. Embedded systems have long been implementing software features instead of hardware peripherals to save cost(and weight/size). The bottomline is that hardware will _always_ cost more than software. If you can stand the 1% performance hit, and it's worth the price cut, then there's no reason not to. As processors get faster and hardware gets smaller, software will dominate in our cost-cutting world.
-Adam
"The rich live hand-to-mouth too-just on a higher level."
-John Guare
That's not true though. I spend plenty for those 400Mhz+ processors. If a software modem will take 15% of the CPU time out of my system, I'll spend the extra money and get a real modem. The hit is a lot more than 1%. I think they even mention that in the article.
You won't see improvements, you didn't get first post, and Suse 6.2 is just about out (if it isn't already).
Happy birthday.
I have to disagree with the "another driver running as root" concern you brought up. By that logic linux shouldn't support ANY new hardware, since all drivers run in kernel mode (which is actually more privlaged than root) and can do anything they gosh darn please to do. Then again hardware is pretty much in the same situation.
Software modems themselves are bad. But they do allow for some really cool hacks. You could build an ultra cheap PBX out of some of these, or answering machine, or something else. Think of it more as a computer interface to a phone. Even TCP/IP telephony would be more attractive with a real phone.
If this company sells them cheap enough, and makes the docs completely open, I might pick up some and hack away. I'd never use one of these things as a modem though. Maybe it's just me but I don't think the $5 I'm gonna save is worth the CPU time it's gonna cost me.
-matt
A software modem is essentially a downloadable DSP chip with a fast interface to the computer on one side and a phone line on the other. To sell one to Linux folk you have to open-source the code - and probably the board schematic, too.
Think about the hack potential (both white and black) of such a device!
And don't forget tapping into the DSP lines behind the telephone interface.
I'm already drooling over the prospect of making a side-looking sonar for my wife's boat, to explore the local bay's floor, or a landmine-finder synthetic aperture radar hacked up out of a couple radar motion detectors and a laptop with one of these "modems". DSPs have specialized instructions for FFTs, which should make it MUCH easier than using a sound card and pure software.
Bantam Dominique roosters crow a four-note song. Once you've heard it as "Happy BIRTHday" you can't NOT hear it that way
fireproof wrote:
A wise man once told me "There are two ways to get something done -- the right way and the cheap way."
I've heard it a slightly different way:
Good, Fast, Cheap: Pick any two (you can't have all three)
-- From RFC 1925, The Twelve Networking Truths, by R. Callon, I.O.O.F.
----
----
Open mind, insert foot.
Linux needs this support. Not everybody can just buy a 'real' modem. Yes, winmodems suck ass, but almost every new laptop comes with a winmodem. Winmodems have less chips and require a lot less power than a normal modem which makes them ideal in the OEM's view for laptops. You can get real pcmcia modems, but even most of the new pcmcia modems are winmodems.
--- A Jesus Fish eating a Darwin Fish only proves Darwin's point.
I believe Rendus was referring to $50 being the cost of a hardware-driven DSP modem, thus being a very high percentage of the total $200 of these cheap computers they want to build.
:-) I've generally had very nice "good enough" success with cheap hardware, but modems are definitely not something I'll cheap out on again.
Those $20 modems you saw are the software-driven "Host Signal Processing" modems, which is what is being debated here (DSP vs. HSP).
I know this because I bought one the other day, and promptly returned it. I had not bought a modem for a couple of years, and had no idea that HSP == WinModem. I just saw that it was cheap, and ISA (it was intended for a 486/33).
The box described hardware requirements as being at least a 200MHz machine running Windows 95/98 or NT. I dismissed this because I frequently see "requirements" on products that are not necessarily true, but rather just there to comfort folks that don't "know better". "Hmm, I have Windows 95, I wonder if it will work with my computer. Yes! It says requires Windows 95, so it must work!"
Needless to say, I'm the one that "didn't know better" this time, regarding WinModems.
An interesting thought with the LinModem though -- if the driver and specifications are sufficiently open-sourced and stable, the $200 Linux box used for net surfing could have lots more potential as well, as software for voice mail could be written as well and integrated with the user interface.
Even if the user had a DSL or Cable Modem connection, a cheap "LinModem", given a stable enough driver, could still be used to provide said answering machine capabilities.
And what about a Linux "telephone server" in a wired home, where your conversation is transmitted digitally to the server before sent out over the phone line? Whoa, I'm starting to get some scary privacy-related issues here, I better stop writing...
"I think that [in the future], a hardware modem will be a high-end, luxury item."
Sounds like something from SegFault.
"Soon every household will have at least 512k SVGA Cards!"
_______
2B1ASK1
I've always thought that the reason WinModems aren't supported in Linux had nothing to do with the release of specs for the drivers. Rather, in order to correctly interprete the wave signals from the modem in software required that the driver execute at specific times. The only way to assure this would be to allow quality of service in the kernel. This warps the design of the kernel and has not been allowed by the kernel maintainers.
So, along comes someone trying to port their WinModem driver to Linux. How are they going to get the quality of service they need? Are they going to hack the kernel? Do you want your hardware vendor requiring you to hack your kernel in a way that the kernel maintainers themselves do not accept?
Or, is this a vaporware announcement intended to make their stockholders think that they are tapping into a market just waiting for a product.
Don't trust this announcement. I'll believe it when I see a Linux machine with a LinTel modem running.
I know I am probably going to be flamed for this but I don't think winmodems are really such a bad thing for home use in the same way IDE drives are not so bad for home use. I have an 28k external modem for my linux mach. and a 56k internal which i basically got for free from tigerdirect when I bought my machine.
:) Moreover, just to test out what the degradation really was like, I had the machine play mpegs (video) (I have no mpeg card so it was softwarebased decompression) at the same time and got less than 10% loss, still well above a typical 28k modem.
I think if you are logged in all day and run jobs in the background on a regular basis then by all means you should get a real modem, but to be honest everything i've ever done on the windoze side of the box (which basically involves downloading the random piece of software) the winmodem has pretty much worked exactly as specified and i get 56k transfers basically for free (which isn't so bad really).. running netscape isn't exactly CPU intensive
(Incidentally the box is an AMD 350-K2 w/ 64
megs of sdram, to give you an idea of the power)
I think if you are trying to build a $200 box for your parents so they can send email, read cnn news and other crap, winmodems are the way to go, since one of the nice things about linux is legally building costeffective machines where every penny counts, i think it would be nice to take advantage of the cheap hardware. (In the same way we now all take advantage of cheap IDE drives which are also more cpu intensive than your typical SCSI drive at home)
just so you know, more than likely if i actually buy a modem it'll be external etc but thats only because most of us are more than typical online users. (Incidentally i think i'll switch to SCSI as well next time around, mostly because 4 IDE device limit is out of control) but again the performance loss is really negligable as far as I can tell for doing typical home use stuff (not games obviously) but real_audio,surfing,email etc..
>CPU heat and CPU load have no relationship to each other.
Well, someone thinks it does! (enough to write a program for windows to do this)
CpuIdle lowers the CPU temperature by disabling it when not needed. This prolongs the CPU life (a decrease by 10C doubles the life span) and cuts power consumption.
cpu idle home page