Good old times

:grin:

One of my college room mates made a prediction the rest of us didn’t agree. He said by the time we graduate, Pentiums will be sold as scraps. We just saw Pentium 90 with mortal combat in the dorm room one door down the hall. While I was doing something today I suddenly remembered this old game with graphic details rated at low, med, high, and pentium XD

tyrian.jpg

He said by the time we graduate, Pentiums will be sold as scraps

You are both wrong. The Pentiums sold today may become "scrap" but the trademark "Pentium" has been very lucrative for Intel. They will continue selling "Pentium" processors for years to come. The processors just may not be recognizable by today's standards.

What's in a name? that which we call a Pentium By any other name would cook as fast; (with apologies to the The Bard)

[quote author=Coding Badly link=topic=56738.msg407448#msg407448 date=1301267900]

He said by the time we graduate, Pentiums will be sold as scraps

You are both wrong. The Pentiums sold today may become "scrap" but the trademark "Pentium" has been very lucrative for Intel. They will continue selling "Pentium" processors for years to come. The processors just may not be recognizable by today's standards.

[/quote]

Technically he meant Pentium 1, not the name Pentium.

We're tossing computers that have Pentium III processors, so I'd say that's not much of a prediction - it's history.

-j

Technically he meant Pentium 1, not the name Pentium … He said by the time we graduate, Pentiums will be sold as scraps … the rest of us didn’t agree

From EBay…
Intel Pentium 4 2.26 GHz (RK80532PE051512) Processor … $0.99 … 0 bids.

People were willing (eager?) to swap, with you, a working 32 bit processor for a $12 shield…
http://www.arduino.cc/cgi-bin/yabb2/YaBB.pl?num=1290714713/all

Why would you disagree with your roommate? Calling the Pentium I “scrap” is being generous.

OK, this was my original thought (I don't think I expressed it well enough :blush: :blush: :blush:): Using "Pentium" as the ultimate level of graphic detail (in that 90's game) made me laugh. I want to share this laughter with anyone that experienced the same old age of computers (which was not entirely very long ago) :grin: :grin: :grin:

I wasn't denying that the time of Pentium has long come and passed.

I'm collecting them just to cherish and remember the good old days.

liudr:
I don’t think I expressed it well enough

According to my wife: I don’t listen. I need my hearing checked. I don’t pay attention. (and a few other things best not mentioned on a family friendly forum)

After rereading your original post, I believe she’s wrong. I think I’m just a little slow. NOW I get what you were trying to say.

Using “Pentium” as the ultimate level of graphic detail (in that 90’s game) made me laugh.

It’s funny, sad, and a bit pathetic. But not as pathetic as the excitement about “EGA” replacing “CGA”. They BOTH sucked.

I want to share this laughter with anyone that experienced the same old age of computers (which was not entirely very long ago)

I was ELATED when the Pentium was released. Oh, the pain I have endured with segmented-model programming. Large model this, huge model that, and a bit of small model so Windows wouldn’t run out of heap. Good riddance!

Yeah, their experiment with 64-bit, what not, virtual memory management?! I'm more kin to the 32-bit virtual memory management in 80386, milestone from my point of view. 4GB physical ram and 64TB virtual memory space (16 bit segment register * 32 bit offset). 32-bit registers EAX, EBX, ECX, EDX, etc, lavish amount of virtual space to store your task segments when you go multitasking and switching between tasks only takes the CPU microseconds (hard drive swapping not counted). I like the PF# or #PF. If you use a typical debugger and assembler, you had to say a keyword (can't remember but the hex code is probably EB) to use 32-bit operations. Liked "rep movsd", 32-bit data move. If you're in protected mode, you have to use that same keyword to do 16-bit operation. Yeah, the short (16-bit offset jump with IP), the long (32-bit with CS and IP), and maybe 32-bit short (32-bit with IP) and 32-bit long (CS+IP). I remember large model was regular program, small was .com program and huge was the one that can do 32-bit addressing. Assembly made more sense to me back than MSC stuff.

Who would have thought that 32-bit became "not enough". I smiled at the ridiculous generosity of intel's engineers on the memory stuff but now I have a 2-year old medium-quality laptop with 4GB ram. They used to use FAT12 ( probably first on both small hard drives and floppy disks :grin: :grin: :grin:), then FAT16, then FAT32 (made a big deal in win95 or 98, had a tool for conversion). I don't hear from FAT system anymore. That's a reason I collected old and small hard drives. Those are the only ones that made sense to me with sectors, cylinders, platter (plates or else, didn't learn it in English). The last time I practiced my hard drive magic was at least 8 years ago, rescuing some data off a 30GB hard drive. I should go to work.

I was ELATED when the Pentium was released. Oh, the pain I have endured with segmented-model programming. Large model this, huge model that, and a bit of small model so Windows wouldn’t run out of heap. Good riddance!

But that mostly went away with the 386, long before pentium. The 286 was the really painful beast.

We’re “somewhat likely” to end up with such models on the bigger AVRs (anything with more than 64k of flash space.) The architecture is pretty fundamentally 16 bits, but that won’t be enough to address all the possible program locations. (and gcc is having “issues” trying to support this, from what I gather.) Sigh. What goes around, comes around.

So does ATMEGA do any segmentation like 8086 does or does it just have one IP that points to a 16bit address? I remember apple II can't do dynamic program loading. You have to know at which address to load your program in order to run it properly, why? No relative jumps? I've spent my youthful years entirely with x86 assembly, although my brother had a book on 8051 or was it 6502?

So does ATMEGA do any segmentation like 8086 does or does it just have one IP that points to a 16bit address?

Just the one IP. Call and Branch instructions on the large memory AVRs can access any (constant) location in the flash space, but indirect jumps end up having to use an IO register to store the high part of the address (very PIC-like. Eww.) Basically the gnu compiler assumptions break down when a pointer to a data location and a pointer to a code location become different sizes (it was bad enough when they pointed to different memories, but that worked out OK most of the time.)

I remember apple II can't do dynamic program loading. You have to know at which address to load your program in order to run it properly, why? No relative jumps?

Most microprocessors can't do dynamic program loading, and need to be linked to run at a specific address. Even when there are relative jumps, they usually can't access the entire address space, and you need to allocate a separate register to serve as the base for your data access (with some performance penalty), and so on. For instance, on AVR, the relative jump/call instructions only go +/-63 instructions. A Motorola 68000 could generate position independent code, but it was pretty messy and not often used.

The 8086 was actually a lovely architecture to deal with if all your programs and data were 64k or less. Automatic relocation! It was just particularly awkward for data larger than 64k (and especially so for C. Two pointers pointing to the same place that weren't equal!) The 286 managed to make things worse, but starting with the 386 it became possible to have flat 32bit pointers, and everything was much nicer. (For a while. Now, we have some of the same difficulties going to 64bits, and those "multiple models" have re-appeared.) Sigh.

Thanks westfw,

So for beyond 60KB boundary, you can do jmp 1AB000 (direct addressing) but you can't do jmp 1C00?

I really like the CS:IP way of x86, reload your program at 16Byte boundary. Code can start at IP=0000 with any CS. Especially when you have over 1MB memory and still work in real mode. You can access 1MB+64K-16Byte of memory. The extra 64KB was always sweet! All jumps are easier to understand when you read them in assembly, the words like long jumps, wide/huge jumps what not in C or C++ just don't tell the programmer what they're actually doing.

Terminal Velocity has graphic detail that goes, low, medium, 486, Pentium!

I set up a DX4 for a friend year before last and she started playing Quake (1), when I got asked, "How do I get past the first Slasher" I replied, "Upgrade to a P100?"

My DX4-100 system is still in hard daily usage, (but then again so is my Mac Plus and Mac II ci) The only Pentium 100 system I have is this really cool dual processor HP system that has on board SCSI. Both systems are dual boot, DR-DOS 7 and Debian. Both are "pure" SCSI systems that have a seperate 100M swap drive and boy are they fast! I usually boot them in Debian, then run up DOSEMU to run games like Duke Nukem, Warcraft, Terminal Velocity, Overkill. Both have dual monitors, the DX4 has 2 VLB video cards and the dual P100 has dual PCI video cards.

When I want to do something with a 16 bit ISA card I've designed and built, I boot in DR-DOS. Sometimes it's good to not have a whole bunch of operating system getting in the way. :)

cyberteque: Terminal Velocity has graphic detail that goes, low, medium, 486, Pentium!

Oh man, what a game! Might be spending this evening with a copy of dosbox and googling for abandonware :D

CowJam:

cyberteque: Terminal Velocity has graphic detail that goes, low, medium, 486, Pentium!

Oh man, what a game! Might be spending this evening with a copy of dosbox and googling for abandonware :D

On the back of the TV box it says,

"The flight simulator for people who don't want combat as an option!"

lol