Good Cheap Hardware Modem?

[bluezebra1098]

Zorba: You may find the OEM of a US Robotics PCI hardware modem (i think the model is 2796 or something like that) but it won't be $30...maybe more like $50.

As for the rest of you...I can't believe you're gonna butt heads with Modus.....that's about as bad as a Sony vs KDS thread...LOL

Modus: FYI...I don't read the whole thing either cause it looks the same...maybe a big UPDATE to catch everyone's attention is in order.

 

[Hyper99]

Hardware modem is alway better, if you willing to pay more then 30 bucks for it.
cheapest I can find is 38 bucks generic hardware and at least 68 dollar
for USR hardware ISA or PCI
Well?
There is no use for them unless you run a BBS in dos.

 

[Zorba]

Modus: Do you know a modem that uses that Lucent chipset?

 

[Zorba]

Nevermind I found one that looks good. Actiontec 56K ISA Master V.90, $53 at buy.com. Hopefully it works .

 

[imhotepmp]

Modus this may see like a weird question, and i maybe be slow at this but..
youre name seems frightingly similar to modem
but for all those latin buffs out there modusis singular for the plural modem, since the suffix determines the number. AM i right or am i way too observant

 

[Modus]

Well you are on the right track, in that Modus is a Latin word, namely the first part of the phrase Modus Operandi: method of operation.

If you've ever watched the old detective movies (before they switched to the M.O. abreviation) you'd hear that phrase quite a bit. It's a police term for the distinguishing actions of a criminal. For instance, part of a bank robber's modus operandi might be that he always wears a Richard Simmons mask. Some criminals tend to commit the same crime the same way many times, making them easier to catch once their M.O is established.

I chose Modus because Modus Operandi is too much of a mouthful. My participation in almost every modem-related thread is simply an unfortunate coincidence for AnandTech's members.

Modus

 

[NFS4]

<< Boy I wish I did work for Lucent, instead of selling computers for a lousy 10% profit and pushing through university. >>


My dad works for Lucent Real Estate

 

[Workin']

Jeez, all this fuss and only 2 specific modems are mentioned. That's not really being helpful, is it? Obviously there are pros and cons to the whole hardware vs. winmodem debate, but for cryin' out loud it is clear that Zorba knew what he was asking for, and instead he gets a bunch of bull$h!t &quot;I'm smarter than you are&quot; crap.

 

[Modus]

Workin,

<<for cryin' out loud it is clear that Zorba knew what he was asking for, and instead he gets a bunch of bull$h!t &quot;I'm smarter than you are&quot; crap.>>

Of course Zorba knew what he was asking for. He just didn't realize that some winmodems (the HCF variety especially) can work in Linux. Similarly, if a fellow came in and asked for the best deal on an i820/RDRAM combo, you would be downright evil not to tell him that a VIA/PC133 setup could do the job for half the price. This doesn't mean you are smarter, it simply means you happened to know something he wasn't aware of.

<<Jeez, all this fuss and only 2 specific modems are mentioned. That's not really being helpful, is it?>>

About as helpful as complaining that no one is being helpful. . .

Actually, if you'd take the time to read the posts you rail about, you'd see that, like video boards, the brand of the card matters much less than the brand of the controller chip. So it is enough to say (like and several others did) that a person should look for a modem with a particular chipset.

Modus

 

[random]

All right Modus, here it is:

(But my time is worth more than cutting/pasting all the time:

Well I finally decided to try it out for myself....

Hypothesis:
Are hardware modems an expensive waste?

Test:

Athlon 750, lucent pci winmodem (not completely software, has signal modulator on board) with latest driver and rom version:
connected at 45333

C:\WINDOWS>ping -n 10 206.53.x.y

Pinging 206.53.x.y with 32 bytes of data:

Reply from 206.53.x.y: bytes=32 time=126ms TTL=29
Reply from 206.53.x.y: bytes=32 time=134ms TTL=29
Reply from 206.53.x.y: bytes=32 time=196ms TTL=29
Reply from 206.53.x.y: bytes=32 time=240ms TTL=29
Reply from 206.53.x.y: bytes=32 time=258ms TTL=29
Reply from 206.53.x.y: bytes=32 time=240ms TTL=29
Reply from 206.53.x.y: bytes=32 time=350ms TTL=29
Reply from 206.53.x.y: bytes=32 time=278ms TTL=29
Reply from 206.53.x.y: bytes=32 time=128ms TTL=29
Reply from 206.53.x.y: bytes=32 time=131ms TTL=29

Ping statistics for 206.53.x.y:
Packets: Sent = 10, Received = 10, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 126ms, Maximum = 350ms, Average = 208ms

486 dx2/66, aopen f56 (hardware), now unsupported so no recent driver... (not even v.90) connected at 28800:

C:\WINDOWS>ping -n 10 206.53.x.y

Pinging 206.53.x.y with 32 bytes of data:

Reply from 206.53.x.y: bytes=32 time=171ms TTL=29
Reply from 206.53.x.y: bytes=32 time=186ms TTL=29
Reply from 206.53.x.y: bytes=32 time=154ms TTL=29
Reply from 206.53.x.y: bytes=32 time=136ms TTL=29
Reply from 206.53.x.y: bytes=32 time=144ms TTL=29
Reply from 206.53.x.y: bytes=32 time=136ms TTL=29
Reply from 206.53.x.y: bytes=32 time=136ms TTL=29
Reply from 206.53.x.y: bytes=32 time=138ms TTL=29
Reply from 206.53.x.y: bytes=32 time=188ms TTL=29
Reply from 206.53.x.y: bytes=32 time=251ms TTL=29

Ping statistics for 206.53.x.y:
Packets: Sent = 10, Received = 10, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 136ms, Maximum = 251ms, Average = 164ms

Same OS (95 osr2.5), same version of winsock (dun 1.3, winsock 2 with y2k patches), same phone line, same cable, done less than two minutes apart.

THIS WAS NOT A FLUKE:

Testing Procedure:

1. send 10 pings to the router on the other side of the link. (Not determined by the windows tracert as that skips some, but by linux ifconfig showing local and remote address)

2. Repeat 2 times.

3. Throw any that are way off from each other.

4. Find the best time of the three for each.

Data:
Lucent: 136 ms with 2 packets dropped [unacceptable and tossed]
215 ms
208 ms (listed above)
Aopen: 274 ms [thrown. one ping of 500 ms and one of 740]
195 ms
164 ms (listed)

Analysis:
Even with the slower CPU and lower connection rate, the hardware modem won. In
addition, it didn't drop any packets. Average higher ping would be preferable over complete packet loss. However, the Lucent is good for long downloads because pauses of up to 10 seconds (which I have experienced) are offset by the extra
~1.5k/s.

Conclusion:

I should continue to use the 486/66 as a linux gateway/firewall. Long downloads will continue to be connected directly using the winmodem. I will also continue to recommend hardware modems. =)

Final note: hardware modems don't cause the whole system to freeze (winamp and mouse stop) for a second and a half every time it goes on or off-hook

 

[Modus]

random,

I'm devasted! You have finally shown me the err of my ways. I will never again preach the devil's gospel of winmodem value superiority.

Actually, there are some obvious problems with your methodology (aside from the fact that your results go against the personal experience of most people on this forum and the other slight problem that your data actually proves my point for me, but we'll get to that later):

1) You say you used the latest driver and rom version for the Lucent LT. First of all, there is no &quot;rom version&quot; for LT winmodems. I'm not sure what you're talking about there. The driver, however is very important. Are you sure you were using the latest one? Which version was it?

2) The machines are not identical. No objective comparison of computer hardware can be made when the test beds are so wildly dissimilar. You might argue that the A/750 favors the winmodem, but CPU speed is not an issue past the minimum requirements. Who knows what software, drivers, configuration settings, etc. have further widened the gap between the two systems? Your data is rendered nearly worthless (though I'll use it later to prove my argument in spades.)

3) Your ping times are quite high for any modem. The results in my own test above and the experiences of most people on this forum indicate that a 56k HCF winmodem gets average pings in the 95-130 ms range. Possibly your phone line suffers from moderate noise or your ISP is overloaded. Regardless, your tests are analogous to a magazine declaring that a Toyota Tercel gets better mileage than a Dodge Neon while at the same time commenting that neither scored more than 10 miles/gallon -- maybe so, but what's wrong with your gas?

4) Your practice of taking the best results from each trial is clearly biased toward the hardware modem, as it increases the weight of its one good trial (164ms) while eliminating the median result (195ms). A fair approach would be to run many tests and throw out the high and low value, or better yet, use the standard deviation formula to iron everything out. Still, working with only the data you provide, we come to see that the average score on the winmodem's two good trials was 212ms while the average score on the hardmodem's two good trials was 180ms. This is only an 18% difference -- as opposed to the 27% spread in your original biased comparison (208ms vs. 164ms). The difference between ping times (latency) for each modem starts to look insignificant, doesn't it?

Still, all these flaws in your testing are moot since your own data actually proves my point. Why?

Whether you take the hardware modem's lead as 18% or 27%, it still comes nowhere near close enough to make up for its added price. Hardware modems are not 27% more expensive, they're 100% more expensive, and sometimes much more than that depending on which company rips you off. So you end up paying for far more than you actually get. And remember, even in your strange testing conditions, both modems would be completely adequate for most online activities.

<<But my time is worth more than cutting/pasting all the time>>

Cutting and pasting saves time. If you didn't cut and paste that post, you would have had to spend quite a while typing and composing something that would only be torn down later. So it's actually a good thing you only pasted it in

Modus

 

[Vrangel]

Simple, his line isnt very good.
On a bad line HW modem will kill winmodem any day.

 

[Workin']

Modus -

You make some good valid points. But.... I looked through the handful of modems I have and couldn't tell you what chipset is on any of them. They are all from different mfg's and every one has a really sticky sticker on the chip.

And I don't think the Best Buy &quot;help&quot; would be able to tell you, either. So if someone goes there and says they need a modem with chipset &quot;x&quot;, it isn't much good.

And I also know this post isn't really adding to the value of this thread, so I'm going now.

 

[Modus]

Vrangel,

The line noise in random's test may have revealed the slightly higher latency of the winmodem -- 18% higher, to be exact -- but it also showed that the winmodem had almost double the throughput. So which is better now?



Modus

 

[JellyBaby]

Winmodems take up a pci slot. I don't like that. Winmodems require a special driver(s). I don't like that. No gaming magazine or site has ever recommended a soft modem. I don't like that.

External hardware modems (USR) have volume controls for the speaker. I like that. Hardware modems have a bank of LED lights for troubleshooting and status. I like that. Hardware modems can easily be moved from system to system. I like that!

In the final analysis hardware modems are definitely worth it for me. I wouldn't think to get a soft modem for my needs but it may work great for someone who doesn't need the above benefits.

 

[random]

1) You say you used the latest driver and rom version for the Lucent LT. First of all, there is no &quot;rom version&quot; for LT winmodems. I'm not sure what you're talking about there. The driver, however is very important. Are you sure you were using the latest one? Which version was it?
My mistake: &quot;FIRMWARE version.&quot; 5.75

2) The machines are not identical. No objective comparison of computer hardware can be made when the test beds are so wildly dissimilar. You might argue that the A/750 favors the winmodem, but
CPU speed is not an issue past the minimum requirements. Who knows
what software, drivers, configuration settings, etc. have further
widened the gap between the two systems? Your data is rendered nearly
worthless (though I'll use it later to prove my argument in spades.

You're telling me that in the probably 8 years between the release of the 486 and the Athlon, they managed to make drivers WORSE? There was no other software running when this test was running. It was a clean installation of Win95 on the 486 (usually only has a proxy running) and a new installation of Win95 (new system). These processes were running:

kernel32.dll
msgsrv32.exe
mprexe.exe
mmtask.tsk
explorer.exe
systray.exe

No customizations were made at all (If you tweak it, it's akin to telling someone to buy a slower processor and OC it making it a better deal than a faster one). And how do you know that it's not everyone else's &quot;tweak's&quot; that hurts the hardware modem?

3) Your ping times are quite high for any modem. The results in my
own test above and the experiences of most people on this forum
indicate that a 56k HCF winmodem gets average pings in the 95-130 ms
range. Possibly your phone line suffers from moderate noise or your
ISP is overloaded. Regardless, your tests are analogous to a magazine
Not everyone will have a &quot;perfect&quot; line. However, I do connect at 44000 or higher. And it's not the ISP. I tried it on multiple ISPs and dialup numbers. And maybe the original poster wants a hardware modem because he has a bad phone line. Or maybe it's going in a 386 for a firewall, where a winmodem wouldn't meet the requirements, ever think of that?

declaring that a Toyota Tercel gets better mileage than a Dodge Neon while at the same time commenting that neither scored more than 10 miles/gallon -- maybe so, but what's wrong with your gas?

Well, seeing as the Tercel (hardware) didn't need gas in the first place... And if someone told me I couldn't drive the Neon (winmodem) on Interstate 80 (beos, freebsd, etc), I would still buy the Tercel, even if it got less mileage. In addition, gas costs money in case you hadn't noticed. Some of us actually have to deal with that.

4) Your practice of taking the best results from each trial is
clearly biased toward the hardware modem, as it increases the weight
of its one good trial (164ms) while eliminating the median result
(195ms). A fair approach would be to run many tests and throw out the
Hey, I have a life. I don't have all that time.


high and low value, or better yet, use the standard deviation formula
Ummm, standard deviation tells you how much it deviates from a normal curve. How does that help here? How would you even apply it? Working with the statistical mode would work however, but if you do too many tests, there is no guarantee of identical conditions on the other side of your line.

to iron everything out. Still, working with only the data you
provide, we come to see that the average score on the winmodem's two
good trials was 212ms while the average score on the hardmodem's two
good trials was 180ms. This is only an 18% difference -- as opposed
to the 27% spread in your original biased comparison (208ms vs.
164ms). The difference between ping times (latency) for each modem
starts to look insignificant, doesn't it?

Well you can't ONLY throw out the worst one. Here, YOU are manipulating the data. Remember, Disraeli (corrollary): &quot;there are Lies, Damn lies, Statistics, and Benchmarks&quot;. This was not intended as a end-all discussion experiment. The only way to do that would to run it with expensive Telco equipment locally and at least two identical computers. Even then, you could say that one of them may have been defective. Better start getting out the cash. Anyone can manipulate data to their advantage. However, even this proposed test would not be effective. It does not simulate real world conditions. Better run around the country (and the world) connecting to different ISPs multiple times and running this test to make sure it's not just the ISP. Oh, and better use multiple winmodems (including the nice and cheap $80+ 3com winmodems =P )

Whether you take the hardware modem's lead as 18% or 27%, it still
comes nowhere near close enough to make up for its added price.
Hardware modems are not 27% more expensive, they're 100% more
expensive, and sometimes much more than that depending on which
company rips you off. So you end up paying for far more than you
actually get. And remember, even in your strange testing conditions,
both modems would be completely adequate for most online activities.

Well, I guess that depends. You could say that hardware modems are infinitely more expensive if you work after rebate, but I've seen no name brand software modems for $30 as well as cheap hardware modems for about $40. I suppose that whenever you buy a new processor, you calculate &quot;hmm.... starting at 500Mhz, i get 10% more performance to go to a 550 but only pay 5% more so I'll go up ... whoops, at 700 Mhz, I pay 10% more but only get a 6% increase from 650.&quot; And seeing as he's planning to run linux, maybe we should use linux to run the ping test. That makes your data invalid too. The latest released Linux module is 5.68, which is what he would have to use. Then you can't compare windoze and Linux data because of the deficiencies of the windows tcp/ip stacks.

The line noise in random's test may have revealed the slightly higher latency of the winmodem -- 18% higher, to be exact -- but it also showed that the winmodem had almost double the throughput. So which is better now?
Which part of the test showed that?!? I don't remember that data at ALL. Can you show me where that came from?

 

[random]

And you forget a VERY important part about the linux driver.

Having a closed-source, unoptimized driver in the kernel space makes the whole system slower. You also fail to note that the linux driver WILL NOT LOAD into any kernel version OTHER than the 2.2.x series. Anyone running an older 1.x (hey, it's still more functional than running an old version of win3.1 or as I remember, 2.0) or the latest and greatest 2.4.x kernels will be unable to use their modem at all. If the driver were open source and work was done to optimize this driver, I would have quite a bit less opposition to winmodems.

(maybe zorba is really Michael Callahan, Al Longyear, or Paul Mackerras and needs a hardware modem to continue development)

cheers!

 

[random]

anyways, as to the original query, the actiontec call waiting modem (hardware and pci) can be found at buy.com for 69.95. Should be able to get at least $10 off of that or their other PCI hardware modem for less by searching pricewatch.

I see 3com's usr 2977 (oem) for $50 with $5 shipping.

Another option would be to hit the second hand market and get it from some lucky sap with cable/dsl/etc. Searching e.b.a.y. for &quot;pci hardware modem&quot; landed one at auction currently for $28CDN which comes to $20.

HTH

 

[Vrangel]

Modus, talking about throughput.

He used his grandpa's 28.8 HW modem.

And its still showed that having a dedicated controller has its benefits.

BTW winmodems have a tendency to connect at aggressive speeds which cannot be maintained on noisy lines.

Which results in frequent disconnects.I'd say spending extra $30 for HW modem to get reliable connection is quite cost effective.

A while ago I did internet setup for a friend with noisy lines.

Conexant winmodem that came in a system kept disconnecting even after reformatting HD, latest driver didnt help either.

I got her Actiontec Callwaiting and problem disappeared.

 

[Modus]

random,

<<Or maybe it's going in a 386 for a firewall, where a winmodem wouldn't meet the requirements, ever think of that?>>

Boy, I never though of that! I guess hardware modems do have a purpose after all. I'm sure the world is just teeming with people who think they need a firewall and decide to use an old 386 for it -- all those dozens and dozens of people.

<<Well, seeing as the Tercel (hardware) didn't need gas in the first place... And if someone told me I couldn't drive the Neon (winmodem) on Interstate 80 (beos, freebsd, etc), I would still buy the Tercel, even if it got less mileage. In addition, gas costs money in case you hadn't noticed. Some of us actually have to deal with that.>>

You know, you managed to miss the point of my analogy altogether and still plow through it with authority. That's quite impressive. Anyways, what I was trying to say was that when two things are compared and the results (as is the case with your high ping times for both modems) are abnormal, then the winner of the comparison is moot. In other words, neither car can be declared to have better mileage because something was wrong with the gas used in the test. Comprende?

<<Hey, I have a life. I don't have all that time.>>

If you acknowledge that your testing was not thorough and that by using only the best of the three results for each modem it inflates the score for the hardware modem and that more trials would be needed for a proper comparsion and that you don't have the time or inclination to do such a comparison, then why did you post that skewed information in the first place?

<<Well you can't ONLY throw out the worst one.>>

Why not? Sure you can. And if not, then how can you ONLY use the best one, as you did? See what I mean? You chose the best method of representing your data so as to make the hardware modem appear better. I simply chose a more objective method. In fact, you even admitted in your paste that the worst trials for each modem should be tossed.

Another good way would be to just average all three results and be done with it. I wonder why you didn't do that in the first place? It would have been winmodem (186) vs. hardmodem (211). Oh, now I see why.

<<I suppose that whenever you buy a new processor, you calculate &quot;hmm.... starting at 500Mhz, i get 10% more performance to go to a 550 but only pay 5% more so I'll go up ... whoops, at 700 Mhz, I pay 10% more but only get a 6% increase from 650.&quot;>>

Yes, I do, as would any sane person. In this case the 650 would be a better deal than the 700, unless there is some other factor.

<<Which part of the test showed that?!?>>

You said yourself that since the hardware modem only connected at 28800 and the winmodem was normal at 45333, throughput was better on the winmodem. And this brings us to another issue: if your lines are so bad that a supposedly fantastic hardware modem only conects at 28800, are they the most appropriate test bed?

<<Having a closed-source, unoptimized driver in the kernel space makes the whole system slower.>>

That's a pretty ignorant statement. Closed-source just means compiled without source code available. Linux couldn't care less whether it was closed source, open source, or the Lone Ranger's source. Once something's compiled, that's it. The fact that it was a closed-source program doesn't affect its performance in the slightest. And &quot;unoptimized&quot;? Why exactly is it unoptimized? What does that even mean? It just means &quot;not yet made better.&quot; Well, so what?

<<You also fail to note that the linux driver WILL NOT LOAD into any kernel version OTHER than the 2.2.x series.>>

The vast majority of commercial Linux distro's still use the 2.2.x kernel, as it the most stable and bug-free for the time being. When 2.4.x gains acceptance, I'm sure Lucent will release a new driver.

Vrangle,

<<Modus, talking about throughput. He used his grandpa's 28.8 HW modem.>>

Wrong. It was a perfectly good AOpen 56k. Read.

JellyBaby,

<<Winmodems take up a pci slot. I don't like that.>>

Um. . . why? Where else is it going to go? Most new motherboards don't even have ISA slots. And besides, with 5 or 6 PCI slots on a standard motherboard, there is plenty of room for one measly card. Plus, keep in mind that very few ISA hardware modems are manufactured these days. The vast majority are PCI. Get used to it.

<<Winmodems require a special driver(s). I don't like that.>>

Then what are you doing in a forum like this? Almost every piece of useful computer hardware requires a driver of one form or another. What difference does it make if it's a 700k compressed EXE or a 12k INF file? Installing a driver is easy and fast. You do it once, and forget about it. It is NOT something to base your buying decision on.

<<No gaming magazine or site has ever recommended a soft modem. I don't like that.>>

It's a well known fact that mass media computer magazines are clueless when it comes to computer hardware. Just flip through PC magazine, or even Maximum PC for that matter. Many looked at winmodems three years ago, wrote them off as cheap junk, and ignored all the recent improvements.

<<External hardware modems (USR) have volume controls for the speaker. I like that.>>

Well, there you are. Right there, that's it. That's why they're so expensive -- it's the bloody volume control! Now I know why everybody around here loves them. Boy, was I clueless. That's definitely worth twice the cost. Ooh, and do they have a motion sensor that triggers &quot;Don't Worry, Be Happy&quot; music when you walk by? Now that would be something!

<<Hardware modems have a bank of LED lights for troubleshooting and status.>>

And have these ever honestly helped you diagnose a problem? If so, what was the solution? To hang up and connect again? You don't need bunch of pretty lights to tell you that.

<<[External] Hardware modems can easily be moved from system to system. I like that!>>

You could even take them to parties. What an asset! Imagine all the women you could meet. It'd be just like having a cute dog with you all the time. Hmmm. . .

&quot;Hey good lookin'. Want to see something special?&quot;

&quot;Um, OK. But I have a root cannal later tonight so hurry up.&quot;

&quot;See this? This is an external hardware modem. It cost me over a hundred dollars.&quot;

&quot;That's great. Did it come with a mint?&quot;

&quot;No, no, this is really neat. See this thing on the back here? It's a serial port connector. And here, let me get something out of my pocket --&quot;

&quot;I really have to be going!&quot;

&quot;No, look, it's just my serial cable. It lets me connect my external hardware modem to ANY COMPUTER I WANT!&quot;

&quot;That's fantastic, but I --&quot;

&quot;Oh and look at this. I can keep the AC power adaptor in my fanny waist pack here and pull it out whenever I need to. It plugs into the wall and the modem and then the modem LIGHTS UP. Look! Look at all those pretty blinking lights!&quot;

&quot;Great, I just --&quot;

&quot;And because it's a hardware modem and costs more money, it must be better, right? And it is. On some really bad phone lines, its ping times are up to 18% faster than a winmodem. Do you know what a winmodem is?&quot;

*sigh*

&quot;No, what is it? I'm DYING to hear.&quot;

&quot;It's this really lousy kind of modem that doesn't even have lights on the front, and it needs drivers to work, and the drivers take so much time to install, I mean really, who has the attention span to wait for a computer to reboot? And they're so sneaky because now the winmodems are practically as good as hardmodems and so much cheaper and run in every popular operating system. Pretty soon everyone will be using winmodems. It's awful.&quot;

&quot;But aren't they almost as good as hardware modems in everyday tasks?&quot;

&quot;Uh, I guess but --&quot;

&quot;And don't they cost many times less than hardware modems, often selling for insanely low prices after mail-in rebates?&quot;

&quot;Well yeah, but --&quot;

&quot;And isn't there a kind of winmodem called HCF that exhibits fairly low latency and fairly high throughput, performing similar to hardware modems of over twice their cost?&quot;

*dumbfounded*

&quot;OK. Well, that's a really nice external hardware modem you've got there, but I've got a root cannal to get done so, as much as I'd love to continue this conversation, I'll have to see you later. OK?&quot;

&quot;Oh sure, sure. I'll see you real soon OK? You have fun at the dentist!&quot;

*waves at her with the modem until she recedes into the crowd*

&quot;I'm such a stud.&quot;

Modus

 

[random]

Boy, I never though of that! I guess hardware modems do have a purpose after all. I'm sure the world is just teeming with people who think they need a firewall and decide to use an old 386 for it -- all those dozens and dozens of people.

You totally missed the point. Just because it's good enough for you doesn't mean that your &quot;solution&quot; will work for everyone. Rather than pasting in your unbelievably long post every time someone asks about a modem, if they SPECIFICALLY say IT HAS TO BE HARDWARE, give them the benefit of the doubt that THAT'S WHAT THEY REALLY REALLY WANT.

why did you post that skewed information in the first place?
To counter your one single example of how good winmodems are.

In fact, you even admitted in your paste that the worst trials for each modem should be tossed.

Wrong. I said that anomolous data can not be counted. Standard procedure, even in any high school science class.

Another good way would be to just average all three results and be done with it

As explained, dropped packets are unacceptable in any case. You going to figure that in? The delay caused by dropped packets is more than sufficient to add to the average time. Given that ping waits 1000 m/s, rather than the reported 136 ms, using the maximum 1000 for those two packets raises that time to 308 ms. Still better?

You said yourself that since the hardware modem only connected at 28800 and the winmodem was normal at 45333
That's because the hardware modem is not v.90 but 56k flex. The ISP I did that demo on was an X2/v.90

they the most appropriate test bed?
Didn't I say that you need to use multiple modems? Nothing makes your tested modem more appropriate than any other one.

Closed-source just means compiled without source code available.
No, but it is commonly known that open sourced code that has been audited is often more efficient than before. Mixing one small bit of bad code into an otherwise efficient system kills the productivity of the system.

&quot;not yet made better.&quot;
Well I guess you're running Windows.... &quot;not yet made better.&quot;

The vast majority of commercial Linux distro's still use the 2.2.x kernel, as it the most stable and bug-free for the time being. When 2.4.x gains acceptance, I'm sure Lucent will release a new driver.

That's because commercial companies are always behind the times. Many times, people have older systems running because there are mission critical software running that is running. There is no upgrade performed on it because downtime is not an option. (hence the reason it's not running Windows) I would also not bet on Lucent releasing those drivers. Given that the binary release is always lagging behind the Windows version, I would say it is safe to assume that Linux is not a high priority for them. Rewriting the driver to interface with the new symbols is not going to be something that happens soon.

perfectly good AOpen 56k.
Perfectly good? It doesn't even have a v.90 upgrade

Plus, keep in mind that very few ISA hardware modems are manufactured these days. The vast majority are PCI.
Bull. Total bull. Read the ads and shop the stores. Almost all hardware modems are ISA.

What difference does it make if it's a 700k compressed EXE or a 12k INF file?
The difference is that the more drivers you have loaded into memory, the worse Windows runs. Why do you think that when you need to &quot;optimize&quot; Windows, you are supposed to GET RID OF drivers and unneccesary crap.

ou could even take them to parties. What an asset! Imagine all the women you could meet. It'd...
Someone has WAY too much time on his hands.

run in every popular operating system
Wrong. Mac, *BSD, Solaris, PalmOS (PCMCIA of course, but still), SCO ....

 

[Zorba]

Damn, I didn't mean for this to turn into another Soft vs Hard modem debate. I thought that winmodems wouldn't work in Linux, period. But I also know I would feel better buying a hw modem that I know will work. If what random said is true, then I don't want it, since I plan on updating my kernel often.

Now since most modem boxes don't say what chipset the modem has, anyone know of a good modem that will work with linux (the actual modem not just the chipset). I don't need a great Brand name or anything, I plan on getting DSL before the year is out (as soon as the phone company gets it ready).

Okay I gotta add my part into this debate. Hardware modems are the best speed wise, Software are the best (by far) value wise. So in most cases Software is the best, but in other case hardware is. This on going debate in everything theard that has the word modem in it is stupid. The people debating will never change their minds, and it just goes on and on.

Anyway, thanks for the help.

 

[Modus]

random,

<<You totally missed the point. Just because it's good enough for you doesn't mean that your &quot;solution&quot; will work for everyone.>>

Obviously the number of people in the universe who are unsuited to winmodems is greater than zero. There are also some people in this universe who refuse to eat ice cream. But to say that I should not advocate winmodems because a puny minority of people cannot use them is akin to saying that I shouldn't suggest ice cream on a hot day because a minority of people are lactose intolerant.

Besides, Zorba's not running a 386 router.

<<Rather than pasting in your unbelievably long post every time someone asks about a modem, if they SPECIFICALLY say IT HAS TO BE HARDWARE, give them the benefit of the doubt that THAT'S WHAT THEY REALLY REALLY WANT.>>

Deep breath. Calm down. Ready to think straight? Alright then:

First of all, I will suggest what I want to whom I want regardless of what you say. So don't waste your breath.

Second, if some one wonders in and posts a message asking for advice on the purchase of a certain type of product, I will not necessarily answer his exact request. Instead, I will discern the reason behind his request and offer the best solution for his needs. That is what makes me a person and not an automaton -- I think rather than compute.

In this case, the original poster, Zorba, requested a hardware modem. I was right in my assumption that he wanted one because he desired Linux compatibility, stable operation and good performance. However, if a winmodem, softmodem, or richardsimmonsmodem can meet those needs for less money, then he should give them serious consideration. That's what I pointed out.

<<To counter your one single example of how good winmodems are.>>

Again, you've already admitted that a truly fair, thorough test would require more time and effort than you're willing to put in. Yet you still expect us to accept your skewed results? You can't have it both ways, I'm afraid.

<<Wrong. I said that anomolous data can not be counted.>>

Alright then, have it your way:

Drop the anomolous results for both modems, and we end up with (as I've already tired to explain to you) an average of 212ms for the winmodem and 180ms for the hardmodem -- a paltry 18% spread. That is nowhere near close enough to make up for the enormous 300% pricing spread between a Lucent LT winmodem and an ISA hardware modem ($9 vs. $36 on Pricewatch). Obviously, the winmodem's value is far superior.

<<That's because the hardware modem is not v.90 but 56k flex. The ISP I did that demo on was an X2/v.90>>

Yet another reason to disregard your entire test based on the odd conditions.

<<they the most appropriate test bed? Didn't I say that you need to use multiple modems? Nothing makes your tested modem more appropriate than any other one.>>

I am going to wear tube socks to my next job interview.

<<Mixing one small bit of bad code into an otherwise efficient system kills the productivity of the system.>>

Again, you're assuming that close-sourced code is inherently bad code. Please, you're grapsing at straws here. Drop it. The Lucent LT Linux driver does work. Its only drawback is that it does not yet support the 2.4.x kernel that is in limitted use. Big deal.

<<I would also not bet on Lucent releasing those drivers.>>

Ever? I'll take that bet.

<<Rewriting the driver to interface with the new symbols is not going to be something that happens soon.>>

It's not as big a job as you're making it out to be. The 2.4.x kernel was not designed to force a complete overhaul of binary drivers. Modifications will be minor.

<<Bull. Total bull. Read the ads and shop the stores. Almost all hardware modems are ISA.>>

Bull. Total bull. Read the ads and shop the stores. Almost all new modems are PCI and that includes hardware modems.

Come on, you're really reaching here. ISA is dead. The slots aren't even present on most modern motherboards. Let it go. Live in the Now.

<<The difference is that the more drivers you have loaded into memory, the worse Windows runs.>>

Windows does not run measurably slower when a properly written hardware driver is loaded. That's a fact. And winmodem drivers have had a long time to improve. They don't slow down your system to any significant degree.

<<Someone has WAY too much time on his hands.>>

Some one needs a sense of humor.

<<Wrong. Mac, *BSD, Solaris, PalmOS (PCMCIA of course, but still), SCO>>

Those are popular operating systems? I guess under your definition of popular, Kathy Lee Guiford is a cultural icon.

Zorba,

<<Okay I gotta add my part into this debate. Hardware modems are the best speed wise, Software are the best (by far) value wise.>>

Speed is download throughput and in that department nearly everyone agrees that winmodems are as fast as or faster than hardware modems. As far as ping times (latency), my results indicate that HCF winmodems will match or beat conventional hardware modems. Even tests done by certain people in this thread bent on promoting hardware modems show that their sometimes better ping times are never good enough to justify their added cost.

Modus

 

[Vrangel]

Regardless of amount of spam posts, HW modems are better for most games
and of course are the solution for noisy lines.

And check Id3dwiz post here.http://forums.anandtech.com/messageview.cfm?catid=27&amp;threadid=188504
Awhile ago on a long modem thread he asked about modems and got usual replies.
Now after two winmodems he is finally looking for a hardware one...