Fry: So you're telling me they broadcast commercials into people's dreams?
Leela: Of course.
Fry: But, how is that possible?
Farnsworth: It's very simple. The ad gets into your brain just like this liquid gets into this egg. [He holds up an egg and injects it with liquid. The egg explodes.] Although in reality it's not liquid, but gamma radiation.
Fry: That's awful. It's like brainwashing.
Leela: Didn't you have ads in the 20th century?
Fry: Well sure, but not in our dreams. Only on TV and radio. And in magazines. And movies. And at ball games and on buses and milk cartons and t-shirts and written on the sky. But not in dreams. No siree!
This breaks something major, that'ill be fun. DNS is too low level to be messing around with it in this way. Sure you get IE taking you to some search page but atleast that is at the browser level and you can change your browser.
At most this should only apply to anything starting with www.. will have to check out the pdf.
But having said that there is the "rule" that 10% of code runs 90% of the time, hence why caches work. If the JIT is only optimizing code that gets run 90% of the time and it's not doing a bad job (since the JIT compiler would most likely be in 2nd level cache) with the other 10% then the benchmark results would be relatively indicative.
Because there is no point of emulating eg. the Motorola MPC range of embedded PowerPC or ARM chips because those chips have low power anyway and probably better performance at the clock speed.
The only thing going for them is the ability to use the PC platform for *some* embedded applications but even that is a moot point since you can target linux to lots of embedded processors. Unless I had something that needed MS Windows I would n't even bother looking at them, I would most probably be able to find a Motorola or ARM instruction set CPU that fits my requirements more closely.
In addition the vast majority of embedded CPU's are 8-bit because most of the time thats powerful enough.
Falcon 3.0 rings a bell, but having said that in those days getting enough conventional memory for most games was a pain in the rear.
Re:Watch out for that trick on laptops
on
Memory Activity LEDs
·
· Score: 2, Informative
Eeek all CGA, EGA, and VGA+ adapters use B800-BFFF for text mode. The only problem I had was with TSENG (remember them?) graphics cards in Windows, they would enable that extra 32K for a larger 128K that meant less bank switching.
The only thing I can think of is that your laptop had some kind of ROM mapped into that area.
Whilst QEMM386 was good you could still get some extra goodness out of emm386.exe. I used to work in a pc repair/upgrade shop, you could sometimes see a customers tears of joy when I used to knock out my "signature" EMM386 line in the config.sys after spending they spent the whole pervious evening trying to get the required 614k to get some game working. The trick was (if I remember correctly) adding/I=B000-B7FFF to 32K more "upper" memory since that memory area was only used for monochrome video cards.. that was nearly 9 years ago, man I cant believe I remember all this crap.
As much as I may not like Microsoft I have more of a problem with software patents. Do you think people looking to make a buck would stop at them? If you had thousands of cases like this going on it would be the end of the software industry.
Well the unemployeed can still vote last time I checked.. Vote-in a government that will do something about it and if you can't get in a government to do what the majority wants (if that is the case) then it's time to start questioning your democracy.
Whilst what you have wrote is all true, EISA did n't exactly become prevalent either (except in servers). The problem with VESA bus was that it was really only designed for graphics cards and not as a replacement for ISA. It became popular because it was cheap to implement and as mentioned worked along side the ISA bus.
Linux has had thousands of people contribute to it. Where would Linux be without the army of people contributing device drivers for example? Linus would still be sitting there coding them all instead of co-ordinating and concentrating on the core of the kernel.
My point was that one person can not do it all now, which was not the case in the 80's.
Golden age died along time ago
on
Razor Blade Games?
·
· Score: 3, Informative
The days of the 8-bit micro was the golden era when it came to small developers(otherwise known as geeks in bed-rooms/garages/basements). You'd have the same person doing the coding, graphics and sound and still have something that did n't look amaturish compared to the big guys. You just cant do that these days, and especially so because graphics and sound have much more prominence then playability did back in the day.
I still remember those 1.99 games being available at my local newsagent. Ahh let the nostalgia begin:)
NUMA is n't about utilizing memory space, it's purpose is to reduce the amount of contention on a bus, a 64-bit address space actually helps you. Contention is the the reason adding >2 processors on a board only gets you marginal spreed improvements since there is n't enough bandwidth on the bus for all the processors.
With NUMA instead of sharing the memory and memory bus each processor has it's own local memory and a special bus for accessing other processors memory. Accessing local bus is always at 100% speed whilst the inter-processor communication bus is always slower. This setup means that you get much better performance when you have a large amount of CPU's.
The problem with this setup is that you need an OS that is aware of this. Also some applications would have to be rejigged to get maxiumum performance.
Also remember that the Opteron does n't have a traditional bus as such, it uses HyperTransport for communication between the CPU and external peripherals. From what I can gather on this dual chipset the second CPU has its memory controller turned off and that all memory requests are serviced by the 1st CPU over the HyperTransport bus via the chipset. How optimal this setup is I cant say.
Easily, you have an area of memory that is shared by both processors. When the process requests an area of shared memory for inter-process communication the kernel maps the memory from this area. This is what NUMA is all about.
These numbers for microprocessors etc mean nothing because they are usually referring to operations on data in cache.. you'ill find that real life performance is 10-20x slower because thats how much slower accessing main memory is.
/* * Allocate 'size' units from the given map. * Return the base of the allocated space. * In a map, the addresses are increasing and the * list is terminated by a 0 size. * The core map unit is 64 bytes; the swap map unit * is 512 bytes. * Algorithm is first-fit. */ malloc(mp, size) struct map *mp; { register unsigned int a; register struct map *bp;
The rest is identical.. the only question is what is the legal status of this code? Is it in the public domain?
Slashdot Mirror
Fry: So you're telling me they broadcast commercials into people's dreams?
Leela: Of course.
Fry: But, how is that possible?
Farnsworth: It's very simple. The ad gets into your brain just like this liquid gets into this egg. [He holds up an egg and injects it with liquid. The egg explodes.] Although in reality it's not liquid, but gamma radiation.
Fry: That's awful. It's like brainwashing.
Leela: Didn't you have ads in the 20th century?
Fry: Well sure, but not in our dreams. Only on TV and radio. And in magazines. And movies. And at ball games and on buses and milk cartons and t-shirts and written on the sky. But not in dreams. No siree!
Unless you tell specify otherwise dont all httpd servers report their version in the "server" response header?
This breaks something major, that'ill be fun. DNS is too low level to be messing around with it in this way. Sure you get IE taking you to some search page but atleast that is at the browser level and you can change your browser.
At most this should only apply to anything starting with www.. will have to check out the pdf.
But having said that there is the "rule" that 10% of code runs 90% of the time, hence why caches work. If the JIT is only optimizing code that gets run 90% of the time and it's not doing a bad job (since the JIT compiler would most likely be in 2nd level cache) with the other 10% then the benchmark results would be relatively indicative.
Because there is no point of emulating eg. the Motorola MPC range of embedded PowerPC or ARM chips because those chips have low power anyway and probably better performance at the clock speed.
The only thing going for them is the ability to use the PC platform for *some* embedded applications but even that is a moot point since you can target linux to lots of embedded processors. Unless I had something that needed MS Windows I would n't even bother looking at them, I would most probably be able to find a Motorola or ARM instruction set CPU that fits my requirements more closely.
In addition the vast majority of embedded CPU's are 8-bit because most of the time thats powerful enough.
Falcon 3.0 rings a bell, but having said that in those days getting enough conventional memory for most games was a pain in the rear.
Eeek all CGA, EGA, and VGA+ adapters use B800-BFFF for text mode. The only problem I had was with TSENG (remember them?) graphics cards in Windows, they would enable that extra 32K for a larger 128K that meant less bank switching. The only thing I can think of is that your laptop had some kind of ROM mapped into that area.
Whilst QEMM386 was good you could still get some extra goodness out of emm386.exe. I used to work in a pc repair/upgrade shop, you could sometimes see a customers tears of joy when I used to knock out my "signature" EMM386 line in the config.sys after spending they spent the whole pervious evening trying to get the required 614k to get some game working. The trick was (if I remember correctly) adding /I=B000-B7FFF to 32K more "upper" memory since that memory area was only used for monochrome video cards.. that was nearly 9 years ago, man I cant believe I remember all this crap.
I dont know but I do know that if my PC looked like those machines I'd be irresistable to women!
EMS/XMS memory thats one nightmare I did n't want to be reminded off.. expanded and extended memory I'd almost forgotten.
As much as I may not like Microsoft I have more of a problem with software patents. Do you think people looking to make a buck would stop at them? If you had thousands of cases like this going on it would be the end of the software industry.
Yeah but they are also getting more and more expensive. On the other hand though somoene must be buying them.
Well the unemployeed can still vote last time I checked.. Vote-in a government that will do something about it and if you can't get in a government to do what the majority wants (if that is the case) then it's time to start questioning your democracy.
What difference does it make if the filtering is done at the ISPs'..
Whilst what you have wrote is all true, EISA did n't exactly become prevalent either (except in servers). The problem with VESA bus was that it was really only designed for graphics cards and not as a replacement for ISA. It became popular because it was cheap to implement and as mentioned worked along side the ISA bus.
Implementing replication at the application layer is about as much fun as implementing table locking at that layer.
Linux has had thousands of people contribute to it. Where would Linux be without the army of people contributing device drivers for example? Linus would still be sitting there coding them all instead of co-ordinating and concentrating on the core of the kernel.
My point was that one person can not do it all now, which was not the case in the 80's.
The days of the 8-bit micro was the golden era when it came to small developers(otherwise known as geeks in bed-rooms/garages/basements). You'd have the same person doing the coding, graphics and sound and still have something that did n't look amaturish compared to the big guys. You just cant do that these days, and especially so because graphics and sound have much more prominence then playability did back in the day.
:)
I still remember those 1.99 games being available at my local newsagent. Ahh let the nostalgia begin
NUMA is n't about utilizing memory space, it's purpose is to reduce the amount of contention on a bus, a 64-bit address space actually helps you. Contention is the the reason adding >2 processors on a board only gets you marginal spreed improvements since there is n't enough bandwidth on the bus for all the processors.
With NUMA instead of sharing the memory and memory bus each processor has it's own local memory and a special bus for accessing other processors memory. Accessing local bus is always at 100% speed whilst the inter-processor communication bus is always slower. This setup means that you get much better performance when you have a large amount of CPU's.
The problem with this setup is that you need an OS that is aware of this. Also some applications would have to be rejigged to get maxiumum performance.
Also remember that the Opteron does n't have a traditional bus as such, it uses HyperTransport for communication between the CPU and external peripherals. From what I can gather on this dual chipset the second CPU has its memory controller turned off and that all memory requests are serviced by the 1st CPU over the HyperTransport bus via the chipset. How optimal this setup is I cant say.
Easily, you have an area of memory that is shared by both processors. When the process requests an area of shared memory for inter-process communication the kernel maps the memory from this area. This is what NUMA is all about.
The whole summary sounds like some Matricesque (sp?) movie with little plot twists thrown in there for good measure.
These numbers for microprocessors etc mean nothing because they are usually referring to operations on data in cache.. you'ill find that real life performance is 10-20x slower because thats how much slower accessing main memory is.
Maybe one day he wants to get a PHD and become a lecturer?
For the lazy the URL. http://minnie.tuhs.org/UnixTree/Interdata_v6/usr/s ys/malloc.c.html
Identical code, interesting stuff.
You can download the source tar file from http://public.planetmirror.com/pub/ancient-unix/an cient/sysIII/sys3.tar.gz
The file you want is usr/src/uts/pdp11/os/malloc.c
first few lines are:
#include "sys/param.h"
#include "sys/systm.h"
#include "sys/map.h"
/*
* Allocate 'size' units from the given map.
* Return the base of the allocated space.
* In a map, the addresses are increasing and the
* list is terminated by a 0 size.
* The core map unit is 64 bytes; the swap map unit
* is 512 bytes.
* Algorithm is first-fit.
*/
malloc(mp, size)
struct map *mp;
{
register unsigned int a;
register struct map *bp;
The rest is identical.. the only question is what is the legal status of this code? Is it in the public domain?