Re: [OT] Stupid consumers and inferior hardware

[Enrico Weigelt <weigelt@metux.de>]

* briand@aracnet.com <briand@aracnet.com> schrieb:

> yes i am capable of modifying drivers, and have done so, although
> not for video. 

Same for me. It's one of the things I earn my money with.

> having source means not (necessarily) having to wait on nvidia to apply
> a patch.  sometimes patches to fix a problem are very simple and even
> somebody not that familiar with the subsystem in question can fix it.
> when the graphic card, or any other hardware, gets old enough which is
> common for MANY linux systems, nvidia will stop fixing bugs completely
> for old cards. planned obscelence and all that.

ACK.

BTW: Nvidia's binary blobs never really worked for me.
(okay, >10yrs ago, the riva128 X-server somehow worked, but badly)
Last time I had the unpleasant experience was on my previous notebook,
the blob kernel module crashed the whole system as soon as X came up.
A little analysis showed up funny things like self-modifying code.

No, these guys simply cannot be trusted. I don't let any proprietary
code into privileged areas (not even speaking of the kernel)
whatsoever. Period.

In recent years I've warned by customers not to buy nvidia stuff
anymore. My little contribution that about 1k of nvidia boards
did not get sold. Not enough for them to care, but if more and more
people do so, they will have to, or get optimized away by the market.

> why do companies keep source "secret" for their old tech 10/100 cards.

That happens when excelsheet jerks rule over things they dont 
understand. They simply can't imagine that there're really valid
reasons for some people bying mainframes worth >100k$ which run
seamlessly for decades.

BTW: one of the funniest arguments for not releasing the sources I
got back from NV jerks was the fear that leaks could be found.
Well, seems they *know* how bad their code is ;-o

> They have bugs because they rev the silicon to save 0.005 on the
> price of the chip and then don't tell anybody.

Funny: these excelsheet jerks are really good in calculating the
total of their milicent-per-unit savings, but they dont have the
slightest idea to what damage do themselves with their bad quality.

> lack of robustness is due to cost cutting and general ass-hattery
> on the part of the mfrs.  it will always be thus unless source is
> available, or even better, open hardware.

Well, better hardware would probably be the better approach in
the long run. Why does every GPU require its own kernel driver ?
Why can't they simply have one generic hardware interface/protocol
for attaching shared memory and command channels and let the 
rest be done via some standardized highlevel language ?

Todays GPUs are so powerful that they can easily render whole 
hires 3D scenaries completely on their own, but making them
complete display servers speaking some standard protocol would
be too expensive ? Absurd.


cu
-- 
----------------------------------------------------------------------
 Enrico Weigelt, metux IT service -- http://www.metux.de/

 phone:  +49 36207 519931  email: weigelt@metux.de
 mobile: +49 151 27565287  icq:   210169427         skype: nekrad666
----------------------------------------------------------------------
 Embedded-Linux / Portierung / Opensource-QM / Verteilte Systeme
----------------------------------------------------------------------