I don’t like teraflops, I’m Bits for life

j7vikes

Member
Jan 5, 2020
175
It’s back when I understood things better. I knew that the consoles should be roughly twice as powerful because 16 was double and 32 and 64.

now I read these threads and people mention teraflops but wait how many teraflops can it access and can it do it as fast on a slower SDD and fixed clocks, broken clocks, variable clocks what developers even know how to tell time it will probably be just first party devs who use it.

No one agrees on anything and I feel like I need a doctoral degree in about 5 different fields to even understand the things everyone on here disagrees about.

I didn’t need that shit back then fucking 64much bigger than 32. That was all we needed.
 

Mr_Antimatter

Member
Oct 28, 2017
1,156
Classic console wars were so much fun. Plenty of joyful memories about debating bits during the Classic age, which shifted to a debate about polycounts and jaggies during the early 00’s.

good times.

also on a related note where the heck is Coolspot HD?
 

Dogui

Member
Oct 28, 2017
3,542
Teraflops sounds like some kind of candy tbh. Like strawberry flavored teraflops.
 

Afrikan

Member
Oct 28, 2017
5,726
I'm just trying to figure out when I dozed off for a min, and we switched from polygons to triangles.

Like... I straight up missed it.
 

jett

Community Resettler
Member
Oct 25, 2017
22,447
Whereas both GDDR5 and GDDR5X used a single 32-bit channel per chip, GDDR6 instead uses a pair of 16-bit channels. This means that in a single memory core clock cycle (ed: not to be confused with the memory bus), 32 bytes will be fetched from each channel for a total of 64 bytes.
Next-gen consoles are actually 16-bit consoles, confirmed. We're going back to sprites!

 

Krejlooc

Indie Game Dev
Verified
Oct 27, 2017
21,732
Brain fart.

I'm done for.

Edit- wait... I meant the industry, like bits and teraflops, use to talk about polygons per second... but now I don't see polygons meantioned anymore, more so triangles per second.

I meant in that sense.
triangles per second has always been the "honest" metric used for many graphics chips going back to the mid 90's. The reason being that the term "polygon" is a catch-all for any geometric figure that is 3 or more points big. 3 points is a triangle, 4 is a quad, 5 is a pentagram, and so forth. Each point representing a corner of a polygon is called a vertex and they take space in vram. Triangles are the smallest possible polygon shape, and thus, they are the most fundamental unit of 3D graphics. In theory, if your triangles are small enough, they can create any other geometric shape. A quad (rectangle) for example can be expressed as two triangles.

You might notice that it takes more triangles to create a single, less flexible higher vertex polygon. Thus yields the hidden asterix of certain old "polygons per second" counts. Sometimes certain graphics chips could switch modes to process quads or other primitives as the base unit instead of triangles. This would be done to save vram space, as one quad can be represented by 4 vertexes and 1 draw call, compared to a quad made of triangles which could be made of 6 vertexes and two draw calls. But, again, working with larger primitives like this is less flexible. So sometimes old manufacturers would publish the theoretical number of quads or similar that could be processed in one second instead of triangles, and simply say they were "polygons per second" instead. Outside of weird systems who can't render triangles, only quads, virtually all 3D graphics are done with triangles primarily.
 

Krejlooc

Indie Game Dev
Verified
Oct 27, 2017
21,732
Was the Atari 2600 4Bit if the NES was 8bit?
The Atari 2600 CPU is an 8-bit MOS 6507, a variant of the 6502 family of CPUs that were also used in things like the Commodore 64 and NES, which has access to only 1/8 the ram the variants in the other mentioned computers have access to.
 

Krejlooc

Indie Game Dev
Verified
Oct 27, 2017
21,732
And Pong consoles must have been 2-bit and Tennis for Two was just 1-bit.
There is no such thing as a 1-bit computer, really. A 1-bit computer would be, like, a light bulb.

A 2-bit computer isn't really called a computer, it's a simple circuit called a half-adder, which is the fundamental unit of all computation. Half-adders are a circuit made out of 2 logic gates -- an AND gate and an XOR gate -- that takes two bits as input and outputs two bits that represent the computed summation of both input bits. Half-adders are the building blocks of calculators. From half-adders, all mathematical concepts can be derived -- subtraction, multiplication, division. In the world of computing, addition and subtraction are the exact same thing, the only thing that changes is our perception of the number!
 

sweetmini

Member
Jun 12, 2019
1,665
ST 260 -> 256 KB RAM
ST 520 -> 512 KB RAM
ST 1040 -> 1024 KB RAM
ST 4160 -> 4096 KB RAM

why the hell did they round the numbers like that?!
to me it all comes from the 65XE (and i always thought it was because of the 6502 cpu, not a rounding of ram)
Then for the other computers, they multiplied it.

I am sure i am wrong, but that's the logic i kept in mind ;)
 
OP
OP
JuicyPlayer

JuicyPlayer

Member
Feb 8, 2018
2,134
The Atari 2600 CPU is an 8-bit MOS 6507, a variant of the 6502 family of CPUs that were also used in things like the Commodore 64 and NES, which has access to only 1/8 the ram the variants in the other mentioned computers have access to.
That’s funny how it’s still considered an 8-bit machine. That used to be the talk of elementary school. Friends used to be like, “Atari games are blocky so they don’t have the power to make them look like people like the NES could!”
 
OP
OP
JuicyPlayer

JuicyPlayer

Member
Feb 8, 2018
2,134
I always used to hear that the difference between PS1 and Saturn is that the PS1 did polygons and the Saturn rendered quads. I have no idea which is better , I just know it made Saturn development hard.
 

eXistor

Member
Oct 27, 2017
5,701
It's funny that Bits have effectively been replaced by Tera Flops; neither figure ultimately means that much, but everybody is using it to prove a supposed point.
 

DrHercouet

Member
May 25, 2018
480
France
One day we'll hit the PlayStation 64 and when it happens video games archeologists will have a lot a fun for at least 5 minutes

(if we haven't blown the planet away until then)
 

bionic77

Member
Oct 25, 2017
13,826
It's funny that Bits have effectively been replaced by Tera Flops; neither figure ultimately means that much, but everybody is using it to prove a supposed point.
I think we can all agree that we should ditch the terminology and go back to vague and lame insults about the other console. Its a language we can all understand.

Sony can make fun of how long you will have to wait to load games on the XBOX and MS can make jokes about how weak the PS5 is. Nintendo can say their console is the only one that is rooftop compatible.
 

Krejlooc

Indie Game Dev
Verified
Oct 27, 2017
21,732
I always used to hear that the difference between PS1 and Saturn is that the PS1 did polygons and the Saturn rendered quads. I have no idea which is better , I just know it made Saturn development hard.
the difference between using quads and triangles does not make the saturn harder to develop for, there are other reasons namely related to parallelism that the Saturn is hard to program for. The difference between how the Saturn and PSX render mainly comes down to flexibility and ram management. The PSX using triangles was thus very flexible, but everything it did ate up a ton of RAM. The Saturn's use of quads made it extremely memory efficient for certain types of artwork. On top of that, the Saturn could perform actual perspective correction, the VDP would tessellate a quad as it distorts. The PSX uses affine texture mapping which preserves parallelism resulting in weird textures.

The main drawback of quads is, of course, they can't represent triangles very well. The closest shape you can get to a triangle if quads are your primative is a trapezoid. Unlike the PSX, the Saturn doesn't offer any sort of UV Mapping ability. That is to say, you can't map arbitrary points on a texture to a polygon. Textures are mapped to the entire quad as-is. This is because the saturn really doesn't have polygons as people consider them. They are actually skewed and distorted sprites. The difference between sprite and polygon comes down to the difference in how visual data on the screen is stored in memory in what is called a frame buffer. The Playstation stores every pixel on the screen individually in a large block of memory called a frame buffer, which lets them change any individual pixel on screen with granularity. Addressing every pixel on screen in memory is very expensive in those days because it ate up lots of memory. The Saturn, instead, uses a tilemap, which can draw an entire screen's worth of graphics with a fraction of the memory usage. This compression comes at the expense of granularity -- the saturn can't arbitrarily change any pixel on screen; for backgrounds it can only change groups of 64 pixels at a time (8x8 pixel blocks in a tile). Obviously, if you're drawing 3D characters, you need to be able to draw with more granularity than 8x8 blocks, so it sets aside another form of memory for sprites. Sprites are blocks of pixels that can be placed anywhere on screen. When the Saturn renders "polygons," each polygon is actually just a distorted sprite.

The way the saturn handles full screen graphics makes it terrific for 2D games. The same exact art on the Sega Saturn will take up a fraction of the art's size on the playstation, which is why fighting games from that era could display way more frames of animation.
 

hikarutilmitt

Member
Dec 16, 2017
3,994
That’s funny how it’s still considered an 8-bit machine. That used to be the talk of elementary school. Friends used to be like, “Atari games are blocky so they don’t have the power to make them look like people like the NES could!”
So kind of like Sword and Sworcery vs Shovel Knight.

I think we can all agree that we should ditch the terminology and go back to vague and lame insults about the other console. Its a language we can all understand.

Sony can make fun of how long you will have to wait to load games on the XBOX and MS can make jokes about how weak the PS5 is. Nintendo can say their console is the only one that is rooftop compatible.
Honestly, I would like to go back to this. It would at least be more fun.
 

Naner

Member
Oct 27, 2017
1,148
Numbers are boring, but they don't confuse the average consumer.

Sega Genesis or Sega Saturn, which is the newest one?

Playstation 2 or Playstation 3?

iPhone 4 or iPhone 6?

Galaxy S7 or Galaxy S9?

Easy to understand progression. But yes, I too would prefer actual console names. Yet, if mainline phones ever did that I'd be confused as all hell due to their yearly releases.
iPhone 8 or iPhone 10?

Xbox 360 or Xbox One?
 

bionic77

Member
Oct 25, 2017
13,826
So kind of like Sword and Sworcery vs Shovel Knight.


Honestly, I would like to go back to this. It would at least be more fun.
Me too.

But honestly its only fun if you remember how dumb it all is. This shit gets deadly serious if you believe in it.

I expect some VERY angry posts about flurbs, nurbs, teraflops, fps, resolution load times, etc. Especially right before and after launch. Thats when it is going to be popping. A lot of people are going to say the most absurd things. Accounts will burn. Reputations will rise and fall. Sales numbers will be used.

And it will be glorious...
 

hikarutilmitt

Member
Dec 16, 2017
3,994
Me too.

But honestly its only fun if you remember how dumb it all is. This shit gets deadly serious if you believe in it.

I expect some VERY angry posts about flurbs, nurbs, teraflops, fps, resolution load times, etc. Especially right before and after launch. Thats when it is going to be popping. A lot of people are going to say the most absurd things. Accounts will burn. Reputations will rise and fall. Sales numbers will be used.

And it will be glorious...
And that's the problem, being too steadfast in generic terms like this. TF is so, so generic, just like the bitness was back in the day. I mean, considering the NeoGeo wasn't actually 24-bit...

Hell, this current TF thing is actually somewhat annoying to see in a lot of regards because I get that with servers at work. People would insist on getting (or selling) the 2x8 core server that ran at a little over 2GHz per core instead of the 6-core that ran at 3.4GHz+ just because it had more cores. There's a reason to get one over the other, and doing website hosting, which isn't as multi-threaded as you'd expect, would usually work better with the 6-core. Hell, SSD was always a thing, too. People underestimate the importance of I/O versus straight number crunching.
 
OP
OP
JuicyPlayer

JuicyPlayer

Member
Feb 8, 2018
2,134
the difference between using quads and triangles does not make the saturn harder to develop for, there are other reasons namely related to parallelism that the Saturn is hard to program for. The difference between how the Saturn and PSX render mainly comes down to flexibility and ram management. The PSX using triangles was thus very flexible, but everything it did ate up a ton of RAM. The Saturn's use of quads made it extremely memory efficient for certain types of artwork. On top of that, the Saturn could perform actual perspective correction, the VDP would tessellate a quad as it distorts. The PSX uses affine texture mapping which preserves parallelism resulting in weird textures.

The main drawback of quads is, of course, they can't represent triangles very well. The closest shape you can get to a triangle if quads are your primative is a trapezoid. Unlike the PSX, the Saturn doesn't offer any sort of UV Mapping ability. That is to say, you can't map arbitrary points on a texture to a polygon. Textures are mapped to the entire quad as-is. This is because the saturn really doesn't have polygons as people consider them. They are actually skewed and distorted sprites. The difference between sprite and polygon comes down to the difference in how visual data on the screen is stored in memory in what is called a frame buffer. The Playstation stores every pixel on the screen individually in a large block of memory called a frame buffer, which lets them change any individual pixel on screen with granularity. Addressing every pixel on screen in memory is very expensive in those days because it ate up lots of memory. The Saturn, instead, uses a tilemap, which can draw an entire screen's worth of graphics with a fraction of the memory usage. This compression comes at the expense of granularity -- the saturn can't arbitrarily change any pixel on screen; for backgrounds it can only change groups of 64 pixels at a time (8x8 pixel blocks in a tile). Obviously, if you're drawing 3D characters, you need to be able to draw with more granularity than 8x8 blocks, so it sets aside another form of memory for sprites. Sprites are blocks of pixels that can be placed anywhere on screen. When the Saturn renders "polygons," each polygon is actually just a distorted sprite.

The way the saturn handles full screen graphics makes it terrific for 2D games. The same exact art on the Sega Saturn will take up a fraction of the art's size on the playstation, which is why fighting games from that era could display way more frames of animation.
This is a lot more detailed than what I used to read in gaming magazines back in the day. They always used to say that the PlayStation’s 2D problems were just attributed to RAM.

Another term I used to hear mainly in magazines like gamefan were PlayStation games were lighting and textures were “CD Sourced” which was something the N64 couldn’t do due to its cart sizes.