NetNews Offline 2/NetNews Offline Volume 2.iso/news/comp/dcom/modems-part1/3493

Text File | 1996-08-05 | 6.0 KB | 127 lines

Newsgroups: comp.dcom.modems
Path: FreeNet.Carleton.CA!an171
From: [e-mail] (Anthony Hill)
Subject: Re: Hayes vs USR
Message-ID: <[e-mail]>
Sender: [e-mail] (Anthony Hill)
Reply-To: [e-mail] (Anthony Hill)
Organization: The National Capital FreeNet
References: <4ebjhe$[e-mail]> <4ebmq4$[e-mail]> <4eiq0c$[e-mail]> <[e-mail]> <4eohtn$[e-mail]>
Date: Thu, 1 Feb 1996 04:31:20 GMT


Tom Hanson ([e-mail]) writes:
> In article <[e-mail]>, [e-mail]
> says...
>>Ricky Lacy ([e-mail]) writes:
>>> In article <4ebmq4$[e-mail]>, [e-mail] says...
>>>
>>>>Two independent testing organizations tested a handful of similairly
>>>>priced modems - including the Sportster and the Hays Accura -
>>>>you can read the results on USR's webpage....
>>>>Accura was one of the worse.....
>>>>
>>>
>>> "Independent testing organizations" can be two guys in a truck.  Ask
> yourself
>>> who paid their bill since it's being used in advertisements.
>>
>>        FWIW, the two tests were conducted by real testing firms.  They
>>weren't exactly the most complete tests I've ever seen, but what results
>>they did have seemed quite accurate (and other non-USR related tests
>>agreed with those results), and they were certainly a LOT better then
> many
>>of the other tests I've seen recently (like that recent C-Net test in
>>which the Hayes Business modem placed quite highly).
>>
>>> More well known labs such as NSTL have consistently ranked Hayes very
> high, in
>>> fact the OPTIMA has won a couple of awards from them over the past
> year or so.
>>
>>        I have yet to see a decent test in which Hayes Optima v.34s have
>>been even good performers, let alone top performers.  But then again,
> I've
>>only seen two decent tests of v.34 modems (PC Mag's March test and
>>Computer Shopper's May test).  All other tests varied from poor to
> worse.
>>
>>Anthony
>>--
>>Anthony Hill | [e-mail]
>
> Those 'tests' run by USR weren't exactly the most real world tests
> everdone on modems.  They involved sending only 2560 characters.  Way to
> small for a real test.  This might have been due to the so called

    Unless of course all you're testing is the ability for the modem
to actually connect.  If you look at the results, I think you'll find that
many modems simply failed to send those 2560 characters.  If they can't
even send 2560 characters, they sure aren't going to perform too well when
you try to download a few megs.

> 'Spiralling Death bug' that USR had in its firmware at the time.  Also
> the supervisor date of the firmware USR tested was Jan 2 1995.  Don't
> believe that USR ever shipped that rev (Hmmm... preparing a special rev
> for at test house...) could be wrong on that but as I recall from
> following CDM USR initially shipped a Nov 94 rom then issued an April
> release to fix the spiralling death bug.

    They used a beta firmware, this should come as no suprise, all
other companies ship their modems to be tested with the newest beta
firmwares as well.  Of course, hardly any companies ship beta firmwares to
the public because they don't want to get a reputation of constantly
sending out new chips which always have bugs and glitches in them.  If you
don't believe that, just look at how many firmwares Zyxel USA distributes
on their BBS as compared to the number of firmwares that are released in
other countries (who's distributers have a somewhat less extensive beta
testing program).

> The tests only checked two impairments noise and 3002 gain/delay curve
> ignoring the other impairments that affect modem transmission.  Also the
> test sent data in both directions an area that USR does do well but in
> the real world most of us don't send data both ways simultaneously.  So
> good bidirectional data throughput is hardly a good reason to spend the
> extra $50 that USR charges for their modem than most of the other modems
> in the test.
>
> Biggest complaint about them is that these test were non standard tests.
>  There is a quite comprehensive test suite - TSB 37A and TSB 38, that has

    All the above comments where only applicable for the first of the
two tests.  The test run by Henderson Labs used two of the tests from TSB
38 (the PSTN simulation as per TSB 37A and the connect reliability test).
It was also a much more recent test.  The Sportster also won both of these
test.  FWIW, PC Mag did a similar test shortly afterwards, the Sportster
won the PC Mag performance tests as well, and in both tests the Motorola
LifeStyle/Power was a close second (actually I believe the USR and
Motorola tied at 100% in both of the connect tests, while the Sportser had
a slight edge on the PSTN simulation.. It would be interesting too see if
the newest Motorola firmware will change that).

> been developed to avoid people making up their own tests so they win and
> then claim that they are the best.  As far as USR's claim that 'the test
> pushed the modems to their limits' its kind of hard to do this when you
> ignore the majority of impairments that are found on phone lines.

    As I said, it wasn't too bad a test, just somewhat limited.  The
Henderson test which did use TSB38 and TSB37A only tested 95% of the
network.  That's actualy more than MANY other tests do (it's rare to see a
test that uses more then 90% of the network sim.), but it's not a complete
test.  It also did not do the throughput test, in which USRs typically
don't do as well as PPI or Hayes modems (these tests are run with NO line
impairments).

> The only thing these tests prove is that if you pay a consultant 25
> grand...

    Or it could probe that if you look hard enough you'll find a test
in which your modem won.  Just about every major modem company has either
complete tests or at least quotes from tests availible in which their
modems did exceptionally well.

Anthony

--
Anthony Hill | [e-mail]