Inq says quad core useless for games

One of the fundamental problems is the languages used to program in. Regular C/C++ with a thread library is just not very good to express parallism in, because you have to manually set up/define what constitutes a thread. The biggest problem here is the granularity of parallism, it's a bitch to go very fine grain because your code gets swamped by IPC and thread control stuff.

A better starting point would be Occam or CSP (and there are C variants of both), which would allow the programmer to express all the parallism in a program that he would care for. The compiler could then decide what to spawn off in separate threads and what should just be executed sequentially in the same context.

Another (and harder) problem is that of debugging. Reproducibility is a lot harder when you go async processing all the way. You really have to design for program failure when you have big parallel software systems

Cheers

 

Chicken and egg scenario here.

What comes first, games that take advantage of Multithreading without hardware support, or game support without multi-threaded cpus?

luckily the hardware guys are pushing the hardware and the software guys will catch up.

I can only imagine new engines coming out in the next 18-24 will all be able to take advantage of at least 2 cores, but preferably able to scale with as many cores as you throw at it.

 

Erm... maybe I'm misreading you but isn't that egg and egg? Anyway you don't need multi-core in order for your multi-threaded game to work correctly.

It's not chicken and egg in the sense that the hardware isn't waiting for the software. Intel and AMD are going to produce multi-core CPUs regardless of what the software people do, because it's the best way for them to spend their transistors. The software people either take advantage, or they don't.

Personally I have to wonder whether worrying about games and multi-core isn't an exercise in navel gazing on the part of the gaming community. I'd like to see a breakdown in marketshare (in dollar terms, not units) between CPUs which go into desktops which sit in offices running Excel, CPUs which go into gamz0r machines, and CPUs which go into oh-so-fairly expensive x86 servers. Might help educate the debate how important multi-threaded game engines are to the hardware boys.

 

I haven't yet seen any of you mention the Alan Wake presentation at the Intel conference. The guy from Remedy outright said that they were utilizing all four cores on Kentsfield. One entirely for physics alone.

 

Perhaps you should try reading the second and sixth posts...

 

To FUD's defense :

 

i think once multi core (ala SMP) PC's are common place. games will be written to take FULL advantage of the scaling where it can. i can think of many possible abstractions of data/functional independance in game 'threads'. things could get A LOT more sophisticated! imagine (ridiculously) a separate CPU for every NPC! they could have a life!

 

I think multiple cores will be of use in the future just not right now for your pc, look at PS3 it has 9 cores but only 7 do the work and it uses Blue ray and ive heard that they will be using blue ray for desk tops soon anyway, so we will be playing games that are 50gig in size in the future on your desktop and there for need multiple cores.

AMDs quad core is meant to act like one core anyway so its not realy useless, just getting ready for the future id say.

 

Nothing of what you wrote makes any sense at all. Where do I begin ...

Your "look at PS3" comparison doesn't make any sense. It doesn't even support your own statement! And for correctness' sake there are 8 active cores in a PS3's Cell, the ninth is disabled for yield reasons.
Blu-Ray has nothing to do with multicores.
50gig games have nothing to do with multicores, and on top of that "I heard we will soon" isn't very convincing, as even now games are launching on CDs in the US because publishers don't think DVD drives have reached enough market penetration ...

AMD's quad core is a quad core is a quad core. It's not a single core, it doesn't act like one, and it isn't meant to act like one. If AMD had wanted to build a bigger and badder single core, they would have done exactly that!

I'm sorry to be the prick who tells you, but your posting is 99% white noise. I'd suggest lurking a bit around these parts before jumping in with shotgun contributions like that one.

 

LOL omg you have not done your reaserch very well by the sounds of things, let me correct YOU on some of your mistakes, PS3 has 9 cores the 9th core is disabled for yield perposes the 8th core runs and only runs the O/S so there for the other 7 cores do all the work, like i said. I am pretty sure that there will be blue ray for your pc in the future (in fact, I can guarantee it), and if there is and there will be, most, if not all the games that will be made will support multiple cores. Again, in the FUTURE!

Who real cares if DVD drives have not reached enough market penetration, this is the future we are talking about not next week.

AMDs quad core is a quad core well no durr, but if YOU have done any reaserch witch i know YOU have not then YOU would know that AMD are trying to get there quad core chip to run like a single core proccessor.

Please next time get your facts together before you start writing white noise like that again.

 

The Cell has 8 SPE cores and one PPC like core. When people talk about the disabled cores of the Cell, they are generally referring to the SPEs. Hence 8.

Besides, each SPE is basically just a DSP. That is, a number cruncher. You're not going to get great AI from that.



If DVD's have not reached saturation after all this time, having two competing standards for the next-gen is even less likely.



The so-called "Reverse Hyperthreading" rumour has been debunked several times and officially denied by AMD.

 

Good job there...what with further proving zeckensacks' point. He comes in with a clear, concise, grammatically correct non-agressive response to your utter trash of a post, and you go and ridicule him by getting all huff and puff cranky and attempting to turn his own statements against him. And i'm sorry, but if you believe that AMD are in any way attempting to utilise all 4 cores as a single core, you're pretty far off the mark.

Honestly, you've dug the whole 6ft, what's another 6, eh?.

 

I'd be really surprised if they limited the physics to run on one core. Having four cores is nice but if each is dedicated to one task then the slowest one will determine the pace. This is a very fundamental challenge. Not only do they have to identify tasks that can run on a separate thread, these tasks have to be splitted over multiple threads as well when the number of cores scales.

The biggest horror is getting it to run well on a single-core, dual-core, quad-core, multi-core. Spin loops offer the fastest synchronization, but on a single-core it's practically a deadlock... So the optimal code is very different depending on the number of cores.

Eventually I'm hoping on some serious hardware supported thread scheduling. One instruction to put a thread to sleep until an event occurs.