I hope the game hasn't fallen in the "technology trap" as its predecessor - in MOO2 it wasn't really worth playing anything but Psilons, as the quick research advancements very easily overweighed the intrinsic advantages of other races (e.g. growth, flying skills).
That said, I can hardly wait to get my hands on it!
And it's not something a compiler can just optimize automatically
All local variables whose references are not passed/copied can very easily placed on the stack by the compiler (bypassing the GC altogether). This has to be done conservatively, of course, so quite a few optimization oportunities would be missed this way. But I believe it's a very good compromise (efficiency is only one of the design goals of most programming languages, power and clarity are IMHO more important)
I'm sorry, I don't really get it. I find believable the fact that you can get better uptime in controlled environments with Win2k, but why would Linux perform worse in such an environment ? Can you elaborate on the "FAR higher" part ?
I mean, c'mon, my Dell box preinstalled with Redhat Linux is up for 84 days, and I haven't really done anything to it.
IDC says factors other than software acquisition cost--particularly staffing and downtime--are the most significant factors when determining TCO over a long-term period.
While it's true that Win2K does much better in terms of uptime than Win NT, it still doesn't even come close to Linux.
Think in terms of dataflow. Given a simple, acyclic circuit, you form a loop by taking an output and wiring it to one of its inputs (directly or via another acyclic circuit). For instance if you have 2 chained inverters and connect them in a loop you get a latch. A Meally/Moore automaton is a memory element + a simple circuit that closes another loop on top of the first one....
I'm not doing quantum computing, but AFAIK quantum circuits can't form loops.
A memory element (latch) needs a loop. A Meally/Moore automaton - 2 loops. A circuit that emulates a Turing Machine - 3 loops. Something that's also programmable - 4 loops.
Most people don't realize that a quantum computer can't function by itself, i.e. it needs a traditional "front-end". This is mostly due to the fact that quantum circuits can't form cycles, and in order to have a Turing-complete system you need at least 3 loops on top of each other.
Moreover, the peculiarities that make quantum computing interesting (e.g. the ability to factorize in polynomial time) also make it completely inappropriate for mundane tasks. So please stop the "google in a cube" shit.
Clockrate is a fixed constant of the processor. IPC (instructions per cycle) also depends on the instruction level parallelism of the program. Taking it to the extreme, if your program is a chain of fully-dependent instructions (i.e. instruction n+1 depends on the outcome of instruction n, for all n), the processor won't ever be able to execute 2 instructions in a single cycle
(there's a lot more to this, like cache misses & branch mispredictions)
K7 processor manage to beat Pentium processors running at the same frequency precisely because of IPC (they get more work done per cycle)
While I don't disagree with the fact that this is probably the chip with the highest clockrate ever built, performance has another ingredient - instructions per cycle (IPC). Now, clockrate remains the same, while IPC is strictly tied to a benchmark, and that's why people buy GHz, not performance.
Such claims have to be backed by benchmark runs. The PIV, when released, had a perf improvemnt of only 15->20% when running at 1.5GHz compared to a PIII running at 1 GHz
The difficulty is not to conglomerate processing power... you can do that relatively easily with
Benjamins... the real difficulty is in either parallelizing your computations, or making a single processor work faster.
So the Teraflops they're mentioning are just a theoretical upper bound, don't get too aroused when you see it.
Well, in the world of architecture, small is fast and large is slow. Memory is already an order of magnitude slower than the CPU.
Furthermore, RAM drives are really meant for servers. Such a server will most likely use a fast internal memory (like RAMBUS) and cheaper, slower & much larger SDRAM 100 for the RAM drive.
It's good that they setled the dispute, however I think it would have been much better for the comunity if GPL were tested and proven valid in court. The reality is, there's tons of very valuable software under GPL, and noboy knows whether GPL is truly enforceable.
The answer is pretty simple - bad management. Most points that you made (and which I consider completely valid, especially given your impressive experience), can be in fact reduced to bad management. Good management is the exception, not the rule.
The Raven.
P.S. The ability to manage people has nothing to do with one's school or even GPA.
... but the production of antimatter. If I remember correctly, the energy used to produce antimatter atoms was more than a million times the energy resulting out of it.
So antimatter is a very compact storage of energy, but we just can't produce it efficiently.
... and claim it's gotten them better performance ? With bean-managed persistence the developer writes the SQL code for accessing the beans; this gives a lot of flexibility, but prevents the container from doing a lot of optimizations.
Anyway, such a comparison is flawed from the start. Bench suites should be developed by independent 3rd parties, or consortiums like SPEC and NOT by vendors.
I actually don't find the results surprising. Microsoft's pet store is heavily optimized for an app server/SQL server; the standard EJB pet store should work with minimal tweaking on any EJB-compliant app server / SQL server pair.
Slashdot Mirror
That said, I can hardly wait to get my hands on it!
That's what you, people with small CRTs, always say.
Imagine a Beowulf cluster of these!
All local variables whose references are not passed/copied can very easily placed on the stack by the compiler (bypassing the GC altogether). This has to be done conservatively, of course, so quite a few optimization oportunities would be missed this way. But I believe it's a very good compromise (efficiency is only one of the design goals of most programming languages, power and clarity are IMHO more important)
4MB L2 cache => *huge* die => low yield => huge cost. Yeap, it's that simple.
I'm sorry, but I think you haven't looked enough ... How about Eclipse ?
KDE 3.1 ?
Very simple ... Sharp Zaurus with tkcPlayer
I mean, c'mon, my Dell box preinstalled with Redhat Linux is up for 84 days, and I haven't really done anything to it.
While it's true that Win2K does much better in terms of uptime than Win NT, it still doesn't even come close to Linux.
I'm not doing quantum computing, but AFAIK quantum circuits can't form loops.
This is precisely what a country with a GNIPC (gross national income per capita) of ~750$ (see WorldBank) needs these days.
A memory element (latch) needs a loop. A Meally/Moore automaton - 2 loops. A circuit that emulates a Turing Machine - 3 loops. Something that's also programmable - 4 loops.
Moreover, the peculiarities that make quantum computing interesting (e.g. the ability to factorize in polynomial time) also make it completely inappropriate for mundane tasks. So please stop the "google in a cube" shit.
K7 processor manage to beat Pentium processors running at the same frequency precisely because of IPC (they get more work done per cycle)
Such claims have to be backed by benchmark runs. The PIV, when released, had a perf improvemnt of only 15->20% when running at 1.5GHz compared to a PIII running at 1 GHz
FYI - WineX uses OpenGL to emulate DirectX! OpenGL is indeed critical.
The Raven
So the Teraflops they're mentioning are just a theoretical upper bound, don't get too aroused when you see it.
The Raven.
In a hundred years nobody will remember you, but I'm pretty sure Poe will still have an important place in the American literature
The Raven.
Furthermore, RAM drives are really meant for servers. Such a server will most likely use a fast internal memory (like RAMBUS) and cheaper, slower & much larger SDRAM 100 for the RAM drive.
The Raven.
The Raven
And how exactly are you gonna keep them in zero g?
The Raven
The Raven.
P.S. The ability to manage people has nothing to do with one's school or even GPA.
So antimatter is a very compact storage of energy, but we just can't produce it efficiently.
The Raven
Anyway, such a comparison is flawed from the start. Bench suites should be developed by independent 3rd parties, or consortiums like SPEC and NOT by vendors.
I actually don't find the results surprising. Microsoft's pet store is heavily optimized for an app server/SQL server; the standard EJB pet store should work with minimal tweaking on any EJB-compliant app server / SQL server pair.
The Raven