Slashdot Mirror
64-Bit Gaming Oversold to Consumers
Ryan Shrout writes "Recently AMD and Atari have both been promoting the game "Shadow Ops: Red Mercury" as the first 64-bit game to hit retail shelves. Even without an operating system ready for it, both companies want us to believe that the 64-bit version of the game adds a large amount of detail and visual quality that the 32-bit version just can't handle. PC Perspective decided to go buy the game and test those claims."
This seems to be just as far fetched as Apples "Worlds first 64bit desktop computer"
except for that whole they were the first large manufacturer to ship computers made for desktop usage that were 64bit
- tristan
The Jaguar was a 16-bit system. It used the same CPU as the Genesis, Motorola 68000. It did have a 64-bit graphic processor, but 64-bit graphic processors have been around in PC's since the 486 days. The N64, while it technically had a 64-bit CPU, was a 32-bit system. The entire motherboard was 32-bit, the CPU was multiplexed. The GC and PS2 are 64-bit (the PS2 uses a CPU that's a close relative of the N64 CPU, just a higher clock speed). The PS2 does, in addition to the main CPU, have two "vector unit" coprocessors, but that doesn't make it a 128-bit system. The PS2 does not have a GPU like the GC does, it's purely a rendering accelerator. It lacks most of the features of modern GPU's, for all intents and purposes it's a really really fast Voodoo1 (it doesn't even support multitexturing, which has been a staple since the Voodoo2). There isn't anything in there that makes it a 128-bit system.
Although things like that do take up substantial computing power, todays 64bit proccessors aren't going to have an easier time doing it then the 32bit counterparts.
You were saying?
Why do I always feel compelled to respond to these trivial bits of misinformation on obsolete consoles?
The Jaguar did indeed contain a Motorola 68000, but it even though it was the only CISC chip in the system it was not the CPU. The system did not have a single CPU, rather any of five processors (two of which were in fact 64 bit devices) could take over the system bus and thus function as CPU. It was this flexible hierarchy that made the Jaguar so difficult to program, resulting in many developers relying on the familair 68000 as the system workhorse (even though it was actually intended originally for housekeeping and to handle controller input), which resulted in the common misconception that the Jag was a 16 bit machine.
The "bitness" of any given system is arguable anyway, and of less significance with each passing generation. NEC first blurred the lines by claiming the TurboGrafx-16 was a 16 bit console based on it's video chip, and the waters have become muddier with each generation. IMHO the Jaguar was the system to finally prove such labels had become worthless. There are three common definitions used to describe a systems "bitness": CPU register width, GPU register width, and system bus width. But more and more it is the overall system efficiency that produces impressive performance, something better measured by standardized benchmarks than the PR hype attached to just one of a system's specifications.
BTW, just for grins, the first console with a 16-bit CPU was the Intellivision. If only George Plimpton had known!
Must... think up... something... clever!
Here is AMD's PR about this game. Here is Firing Squad's review with ATI cards and mentions Athlon 64 briefly.
Ant(Dude) @ Quality Foraged Links (AQFL.net) & The Ant Farm (antfarm.ma.cx / antfarm.home.dhs.org).
Digital Equipment shipped 64-bit desktop computers. (Running Windows even).
Wrong. Some code benifits for the increased register size that 64 bit processors bring. In some cases the speed increase can be several hundred percent due to drasticaly reduced memory access.
You do know the Athlon64/Opteron was out for several months before the G5 hit, don't you?
Perhaps you should get your facts from someone other than the apple marketing dept.
You're sort of correct. But remember, you and I are thinking about this is human terms, to a processor a decimal place doesn't necessairly differentiate value. Consider these three numbers:
10000000000000000000.0
... you can now realistically have large computations of game-world environment *quickly* moved around the system (data with "twice the number of decimal places). At 32-bit system would take multiple "moves" to do this (see note below), with 64 it is done in one swoop.
1.0000000000
0000000000.1
0.0000000001
If I were to offer you one of the three as a quanity of, say, money you'd obviously choose the first. But say to a processor "store this in a register" it doesn't really care - all three numbers have the same relative weight. Not like one takes up more threads or "memory weight."
Now, what 64-bit computing would give us are numbers that look like:
1.00000000000000000000
0.00000000000000000001
The first two don't make much of a bit of difference as far as games are concerned - a texture mapping of 1 is no different than a texture mapping of 1.0000, and the second number is so small it isn't practical for game usage. But look at the third value
NOTE: It is important to understand that doubling the number of decimal places does significantly more than doubling the amount of data to be transfered. Here:
100
10000
Well, 10000 has twice the number of zeros, but to "move" 100 twice would only give you 200. It would take 100 moves to equal 10000 (100 x 100 = 10000). This is the real power of 64-bit computing.
Bzzzt, wrong, thank you for playing, better luck next time, here's the home version.
There was a version of OS/2 made and marketed for a specially made 64-bit DEC-alpha workstation in 1992, mostly to hospitals and a few universities. Not too many used it, but I worked on some at Johns Hopkins and UVA. I bet there were maybe 2,000 of these beasts in existence, but they predated anything Apple did by a longshot. And these weren't servers or anything, just (admittedly specialized) desktop machines. Really pissed me off when I'd catch a nurse playing solitaire on them.
There are several different meanings of "64-bit" and they all have differing impact on making videogames (and computing in general for that matter).
32bit vs 64bit address space: Currently most PCs and all game consoles can handle up to 4 gigabytes of memory. This is getting to be a problem on PC because games are using hundreds of megabytes of textures and because memory-mapped I/O for things like PCI cards eats into that total available memory. Going to 64bit addressing completely solves this problem. This is the "64bit" this article is about. The game in question doesn't really take advantage of this, however.
32bit vs 64bit precision for floating point math: Not really a big deal at all. You can do 64bit math already on all the systems, it's just not done in hardware so it's very, very slow by comparison. There's almost never a need for the extra precision anyway; things that lack precision at 32bit are usually flawed due to positive feedback or a lack of understanding of the math pipeline.
32bit vs 64bit data bus: We've already gone to 64bit data busses and beyond. PlayStation2 uses a 128bit wide data bus. Helps you feed data to the CPU (and other system devices) more quickly. Very useful but old technology these days.
32bit vs 64bit registers: Old news, we went to these with the original Pentium. Basically the same argument as for 64bit data bus.
32bit vs 64bit colour: Going from 8bit integer colour channels (ie. red, green and blue from 0-255 each) to 16bit floating point colour channels. This gives you a huge amount of dynamic range for colour and makes it easier to represent very subtle differences too. You need fairly complex pixel shaders for this to be worthwhile, but if you do have that capability it makes all the difference. The next generation of consoles will use this as will coming PC games - it will make their lighting feel much more realistic.
Graham
Read..
According to these benchmarks, a 64-bit Athlon actually runs games FASTER under the current 32-bit version of Windows XP than under Windows x64 with the latest beta drivers and such. Some games saw as much as a 35% decrease in framerate under the 64bit windows beta.
This just goes to show that we can't really evaluate 64-bit apps on 64-bit platforms (except linux) until we have both an OS and final release drivers.
Looking through my PS2 manuals: "The processor has a 128-bit width data bus and registers. The CPU's general-purpose registers (GPR) and floating-point coprocessor regithers are 128 bits wide. All processors are connected via a 128-bit bus."
Free of Flash! Free of Flash!
You don't know what the HELL you are talking about, and whoever modded you informative is an idiot.
First of all, doubled data width doesn't equal doubled performance. The bottleneck in a computer today is the RAM, and there is no speed increase there from going to 32 to 64-bit.
Also, your babbling about floating point completely wrong. First of all, an increase in bits doesn't mean an increase in speed. You get more precision on your floating point numbers, but that doesn't matter at all because you're still doing the same amount of calculations and crunching the same amount of numbers. Not that it matters at all, since the new 64-bit CPUs still use the same old tried-and-tested 80-bit floating format they've used since the Pentium 1 days (Maybe earlier too, don't know for sure). So there is no increase in precision or speed of floating point math for 64-bit CPUs, which invalidates your whole argument.
Neither does the increased register length give any direct speed advantages. No, the real advantage comes instead from the major architectural changes between 32-bit and 64-bit CPUs, like the increased number of internal registers. And these are not really related to the bitness of the CPU at all.