The problem with Apple just porting OSX to x86 is that one runs straight into the lack of hardware support I described. But if there was an OSX build on top of FreeBSD, then Apple would be able to make immediate use of the device driver support for FreeBSD, which if it is behind Linux, is still pretty respectble (and FreeBSD had a decent ATAPI CDROM-burner driver earlier than linux...)
These kernel patches (Mandrake and SuSE) are not fragmentation: they are applied by groups of developers whose patches `track' the official kernel release, to provide features that in general Linus has agreed will go into the kernel at some future point. If any incompatibility is found between these patched kernels and the official kernel, the patches will be fixed (which is not what happens if there is fragmentation).
From a user point of view, there isn't a huge amount of difference between varieties of UNIX: most importantly, it's easy to get them to talk to each other and share services. But in terms of the fit between a particular UNIX and its realisation on a particular hardware setup, there is the world of difference: existence and quality of device drivers, support for SMP, clustering services, file systems, etc.
Linux/XFree86 has established a near invincible lead in this area for the x86 architecture. I can't help thinking that UNIX on the PPC is always bound to be a niche phenonmenon. Now if OS/X were to be ported to FreeBSD/x86...
Is this a foundry issue? This is one of the several weaknesses of Transmeta: that they had such a hard time getting a foundry to manufacture their chips, earlier this year, so perhaps this is indicitive of less than top-notch quality.
Maybe an idea is to have a `slow-but-sure' setting that routers don't have to deliver quickly, but try to ensure that they aren't dropped. This would be a bad setting for urgent content delivery, but it might be a good setting for spreading information about traffic spikes, etc.
I think that the dynamic compilation technology that Transmeta have bet their shirt on is a winner, and will in the long run render CISC procesors obsolete. But I doubt that Transmeta in it's current form will reap the benefits. Look at what has gone wrong for them:
Their unveiling really gave less than stellar performance: their 700MHz machine gave performance that is probably in the range 450MHz-500Mhz (but independent, comprehencsive benchmarks are not to be had). By all accounts this has been very disappointing to Tramsmeta's engineers.
This was at a start of a year that has seen strong increases in Intel's and AMD's flagship offerings. Transmeta have not shown off anything new in this time, so their already poor performance is falling behind.
They ran into horrible foundry problems at the beginning of the year, which, while settled looks to be settled in a quite expensive way for them.
OEMs have not exactly been flocking to their product.
Now this: this could really hurt consumer confidence in Transmeta notebooks.
Most likely outcome: they get bought out by a competitor (Intel? IBM? perhaps even AMD or Sony?), at a favourable or not so favourable price. To survive on their own they will need to do something surprising.
You make some interesting points: especially the point about memory support for concurrency is interesting. It looks to me that better exploitation of concurrency is really the only way to deal with the CPU / motherboard bus bottleneck.
I have only an incomplete understanding of how Rambus works. Is it fair to say that the Rambus system carries a big latency penalty compared to prevalent memory technologies, but it is a model in which increases in bandwidth can be designed in without incurring further costs in latency (ie. we have bandwidth scaleability)? Is this the strength that many designers have seen in Rambus?
I've thought about this some more, and I've changed my mind. Jerf's example is a good one, and I agree it is probably the kind of example the EFF were after.
Yes, the means to do something legal is made illegal. I think, though, this aspect of the DMCA is well worked over, and wasn't what the EFF was after: they were after instances of *fair use* of copyrighted material that DMCA forbids.
What was extraordinary about the DeCSS trial was that the RIAA didn't pretend that either DeCSS had been used to infringe copyright, or that it provided a viable technology for profitable piracy, but argued speculatively that changes in price could make it profitable, and this possible future use for piracy was the only relevant use of DeCSS. Well, what was extraordinary was that they got away with it...
Companies can attempt to assert copyright over material that they do create (ie. not created by employees): look at the recent story of AOL trying to assert copyright over the creative works of volunteer editors at dmoz (or Open Directory Project; this matter is still not resolved).
The DMCA specifies certain infringements that copyright holders may sue over. In this case, as the copyright holder, you cannot infringe your own copyright (and you could also grant waivers to other parties). So there is no illegality in your example. The examples that the EFF are after concern fair use of products whose copyright are held by other parties.
How likely do folks think that this will go to the Supreme Court? The case will only be heard if the Supreme Court decides to hear it, and they decided not to hear it earlier. My guess is that if the appeal court largely agrees with Jackson's legal judgement, then they will not examine the case, in which case we can expect the case to be finished by summer next year. Opinions?
I'd agree that Win98, Win98SE and WinME are not exactly stellar advances on Win95, but you're being unfair about Win2k: it offers a lot of important improvements over NT4 (eg. long uptimes, improved scripting and remote administration, much improved security features in the registry, rebootless device changes). Win2k is a plausible server OS, a claim I don't think was true of NT4.
The GPL requires that you be willing to give the source under a GPL lincense to anyone who receives the binary. Tom would therefore be entitled to the source. Unless you receive the binary, you would not.
Well, some of Intel's innovations were not so hot (eg. MMX). The P4 looks to be a big step towards a different model of processor (deep pipelining, sophisticated branch prediction) which whilst I agree in the long run is probably right, in the short run it might be a long time before it becomes an improvement on current technology. One could say the same about Rambus...
And if we are in the business of backing predictions about which will be the best architecture in the long term despite less than stellar short term performance, why should we believe the P4 architecture is better than that of rival VLIW architectures (eg. the Crusoe)? Following Intel's lead has been the right thing to do whilst Moore's law held, but now it rather looks broken...
Re:Damned if you do, Damned if you don't
on
Golden Rice
·
· Score: 2
I'm living in the States but I follow both the British and the German press...
There is concern about agri-business, but the high emotions seem to be mostly about GM foods, and equal standards are not being applied to both. Look at the furore about the Monarch butterfly, and compare it to the list of species that have become endangered by pesticides
Sure, there are dangers to GM foods, but they also promise to end one of the great environmental crimes of the modern day: drowning huge areas of land with dangerous bioactive chemicals, which cripple biodiversity and poison water supplies.
Slashdot Mirror
equilibrium.
A given planetary equilibrium is not necessarily very comfortable.
Think of Venus.
The problem with Apple just porting OSX to x86 is that one runs straight into the lack of hardware support I described. But if there was an OSX build on top of FreeBSD, then Apple would be able to make immediate use of the device driver support for FreeBSD, which if it is behind Linux, is still pretty respectble (and FreeBSD had a decent ATAPI CDROM-burner driver earlier than linux...)
D'oh. I'll take the foot out of my mouth. I thought you were describing the version of gcc that KDE uses, not RH.
The `peculiar flavour' of gcc (egcs) is the official compiler for the linux kernel.
These kernel patches (Mandrake and SuSE) are not fragmentation: they
are applied by groups of developers whose patches `track' the official
kernel release, to provide features that in general Linus has agreed
will go into the kernel at some future point. If any incompatibility
is found between these patched kernels and the official kernel, the
patches will be fixed (which is not what happens if there is
fragmentation).
From a user point of view, there isn't a huge amount of difference between varieties of UNIX: most importantly, it's easy to get them to talk to each other and share services. But in terms of the fit between a particular UNIX and its realisation on a particular hardware setup, there is the world of difference: existence and quality of device drivers, support for SMP, clustering services, file systems, etc.
Linux/XFree86 has established a near invincible lead in this area for the x86 architecture. I can't help thinking that UNIX on the PPC is always bound to be a niche phenonmenon. Now if OS/X were to be ported to FreeBSD/x86...
Is this a foundry issue? This is one of the several weaknesses of Transmeta: that they had such a hard time getting a foundry to manufacture their chips, earlier this year, so perhaps this is indicitive of less than top-notch quality.
Maybe an idea is to have a `slow-but-sure' setting that routers don't
have to deliver quickly, but try to ensure that they aren't dropped.
This would be a bad setting for urgent content delivery, but it might
be a good setting for spreading information about traffic spikes, etc.
their shirt on is a winner, and will in the long run render CISC
procesors obsolete. But I doubt that Transmeta in it's current form
will reap the benefits. Look at what has gone wrong for them:
700MHz machine gave performance that is probably in the range
450MHz-500Mhz (but independent, comprehencsive benchmarks are not to
be had). By all accounts this has been very disappointing to
Tramsmeta's engineers.
in Intel's and AMD's flagship offerings. Transmeta have not shown off
anything new in this time, so their already poor performance is
falling behind.
year, which, while settled looks to be settled in a quite expensive
way for them.
notebooks.
Most likely outcome: they get bought out by a competitor (Intel?
IBM? perhaps even AMD or Sony?), at a favourable or not so favourable
price. To survive on their own they will need to do something
surprising.
Big, established chip makers with a diverse product range can weather
this. Small IPOs with all their eggs in one basket can find
this...unsettling.
The opportune time would have been just *before* the IPO. It could
have been a lot worse for Transmeta a month ago...
Umm, you write bits in words in a big-endian notation, so the PDP output looks weird because you are mixing big endian and little endian.
Thanks, this answer is very helpful.
support for concurrency is interesting. It looks to me that better
exploitation of concurrency is really the only way to deal with the
CPU / motherboard bus bottleneck.
I have only an incomplete understanding of how Rambus works. Is it
fair to say that the Rambus system carries a big latency penalty
compared to prevalent memory technologies, but it is a model in which
increases in bandwidth can be designed in without incurring further
costs in latency (ie. we have bandwidth scaleability)? Is this the
strength that many designers have seen in Rambus?
How lame.
I've thought about this some more, and I've changed my mind. Jerf's example is a good one, and I agree it is probably the kind of example the EFF were after.
though, this aspect of the DMCA is well worked over, and wasn't what
the EFF was after: they were after instances of *fair use* of
copyrighted material that DMCA forbids.
What was extraordinary about the DeCSS trial was that the RIAA
didn't pretend that either DeCSS had been used to infringe copyright,
or that it provided a viable technology for profitable piracy, but
argued speculatively that changes in price could make it profitable,
and this possible future use for piracy was the only relevant use of
DeCSS. Well, what was extraordinary was that they got away with it...
Companies can attempt to assert copyright over material that they do
create (ie. not created by employees): look at the recent story of AOL
trying to assert copyright over the creative works of volunteer
editors at dmoz (or Open Directory Project; this matter is still not
resolved).
The DMCA specifies certain infringements that copyright holders may
sue over. In this case, as the copyright holder, you cannot infringe
your own copyright (and you could also grant waivers to other
parties). So there is no illegality in your example. The examples
that the EFF are after concern fair use of products whose copyright
are held by other parties.
How likely do folks think that this will go to the Supreme Court? The
case will only be heard if the Supreme Court decides to hear it, and
they decided not to hear it earlier. My guess is that if the appeal
court largely agrees with Jackson's legal judgement, then they will
not examine the case, in which case we can expect the case to be
finished by summer next year. Opinions?
I'd agree that Win98, Win98SE and WinME are not exactly stellar
advances on Win95, but you're being unfair about Win2k: it offers a
lot of important improvements over NT4 (eg. long uptimes, improved
scripting and remote administration, much improved security features
in the registry, rebootless device changes). Win2k is a plausible
server OS, a claim I don't think was true of NT4.
Where do you get this figure from? It sounds implausible to me.
The GPL requires that you be willing to give the source under a GPL
lincense to anyone who receives the binary. Tom would therefore be
entitled to the source. Unless you receive the binary, you would not.
looks to be a big step towards a different model of processor (deep
pipelining, sophisticated branch prediction) which whilst I agree in
the long run is probably right, in the short run it might be a long
time before it becomes an improvement on current technology. One
could say the same about Rambus...
And if we are in the business of backing predictions about which
will be the best architecture in the long term despite less than
stellar short term performance, why should we believe the P4
architecture is better than that of rival VLIW architectures (eg. the
Crusoe)? Following Intel's lead has been the right thing to do whilst
Moore's law held, but now it rather looks broken...
press...
There is concern about agri-business, but the high emotions seem to
be mostly about GM foods, and equal standards are not being applied to
both. Look at the furore about the Monarch butterfly, and compare it
to the list of species that have become endangered by pesticides
Sure, there are dangers to GM foods, but they also promise to end
one of the great environmental crimes of the modern day: drowning huge
areas of land with dangerous bioactive chemicals, which cripple
biodiversity and poison water supplies.