Dual Core CPU's - not good for gaming...

Status
Not open for further replies.

gimpycow

Baseband Member
Messages
54
Hello all,

I posted the following article that I wrote on a website hosted by old college friends that nobody reads. Anyhow, I am looking for some feedback on it. Let me know your opinions on the topic! Feel free to point out areas that I might have been incorrect.

My article is below:

State of the Gaming World address by Cow

The year 2004 heralded the best looking games ever to grace a monitor. Namely Far Cry, Doom 3 and Half-Life 2 led the way. All three of these engines put DX9 pixel and vertex shaders to good use forcing many gamers to upgrade their systems to enjoy the game in all its glory. All of that is to be expected as this type of upgrading is the way it has been since the dawn of gaming. Approximately every three to five years, a new console or video card would be released taking games to a level that much closer to visual reality. However, as 2005 approaches, hardware manufactures are finally running into the significant roadblocks of heat and energy leakage. As manufacturing processes become smaller and smaller, more transistors can be placed on a chip and greater clock speeds can be attained which is the primary fuel that has driven Moore's Law. However, to make chips run faster and faster is requiring smaller chips being fed more and more energy and thus, increased heat and energy leakage. The heat build up can be such a problem that the system will crash unless sufficient cooling is implemented. While such cooling systems exist on the market today, they are expensive and take a lot of energy to run themselves. There are other factors within chips that cause the heat to increase but that is going much too deep for this article. Instead, this article will focus on the short term solution that hardware manufactures are now implementing and the decisions consumers and game developers will need to make in the not too distant future.

The area where heat is currently hitting the hardest is in the most important piece of the computer, the central processing unit. In general, a computer's performance can only be as fast as its slowest component. Therefore, if a computer's cpu becomes the bottleneck of future systems, advancements in both video cards and memory will only be able to push the limits of graphical boundaries so far. Intel had originally planned to take their Pentium IV processors to 10GHz but have now decided that 3.8GHz is the fastest speed the P4 will achieve. Likewise, AMD's fastest processor is 2.6GHz (the FX55) and their roadmap only shows the chip increasing to 2.8GHz in 2005. While AMD's chip is lower clocked, it outperforms Intel's best processor in almost every benchmark available due to its design. However, AMD is in the same overall performance boat as Intel as the performance ceiling of their chip is getting closer and closer. Now, instead of increasing the core clock speed to yield a performance gain, both companies have found different ways such as increasing the amount of L2 or L3 cache. However, increasing the cache will not come close to yielding the performance gains of Moore's Law that we have become accustomed to in the past. After all, if systems do not get faster, the demand to upgrade or purchase new systems will decrease. Intel and AMD need to keep getting faster if their bottom line is to grow. Since it is mostly games and high-end business applications that need this increased performance, gamers and IT professionals will be the primary consumers interested in this industry change.

What both companies have decided to do is instead of focusing on ways to make single chips faster, they will focus on ways to make a single chip do double or quadruple the amount of work at the same clock speed. In essence, this boils down to placing two or more central processing units onto one chip thus the name, dual-core. While in theory this may work, applications will need to be coded to take advantage of multiple cpu's. Intel has already introduced Hyper Threading to virtually mimic true multi-threading on a single cpu but I don't believe many applications are currently take advantage of it. Likewise, Windows can handle multi-threading but from what I have read, it is far from being optimized for it. I believe LongHorn, a future Windows Operating System, will be enhanced for multi-threading and virtualization but Microsoft is not planning to release this Operating System until sometime in 2006. AMD has already publicly written that current games will benefit much more from one fast processor than two slower ones in the same system.

So what does this mean to the average gamer whose PC is getting old and needs to upgrade soon? In 2005 or 2006, should a consumer choose a slower clocked dual-core cpu or a faster clocked single-core cpu. In addition, game developers will also need to decide how to code their games. Do they continue to code their games as a single-threaded application without the complexity and issues that multi-threading will present? If not, when do they decide to start coding games for multi-processors? After all, if most gamers do not own multi-processor machines, what sense does it make to code games for hardware that has not penetrated the market yet? The old "Catch 22" of new hardware requiring new software coding has occurred many times in the past and it usually takes at least three years for the software code to catch up. When this occurs, it is always key to wait for the optimal buying time else the consumer will either purchase old technology soon to be outdated or new technology which is overpriced and maybe not quite ready for mainstream applications.

Personally, my strategy has always been to upgrade when a "must have" game is released that I cannot play with the graphic details set to a reasonable level. Specifically, I am thinking the tentative release of the graphical monster Unreal 3 in 2006 might truly test my current pc. Since this game engine is already in development, it is probably being targeted for fast single-core cpu systems. If my current system cannot handle the Unreal 3 engine even with a video card upgrade, the consumer's dilemma of what type of system to buy will raise its ugly head. My hope is that by the time I am forced to purchase a new machine, dual-core cpu's will have come down in price and multi-threaded games and operating systems will have already been released. I could be wrong, but the decision to upgrade to a multi-core system versus a faster single-core system might be the hardest decision a pc consumer has had to make in quite sometime. Lets hope that other technologies such as motherboards will make this industry transition a bit less painful by being compatible for both singe and dual-core cpu's. At the end of the day, the industry needs a big performance boost. New rendering techniques in software along with new types of materials used in chips might propel the industry in the long-term, but the short-term is all about the advantages of dual-core chips and multi-threading. Lets hope that developers are already working on multi-threaded engines and working out the kinks else the game industry might hit a wall in the not too distant future.
 
Very nicely writen. I dont have any insight into the accuracy. Its an intresting topic that you dont often hear of.

I think in the coming years nano technology will have an impact on this. I studied nano technology in my material science class and it will have an impact on electronics and computers.
 
Its true developers hardware/software are getting close to this point in mores law .Personaly i believe they will lean toward dual processors like the AGP Sli`s are presenting now only time will tell but thats my opinion
 
Thanks for the feedback!

Nanotechnology is one of things I think may come about in the long-term. However, it is the performance increases (or lack thereof) expected in the short-term that concerns me. I think applications/games are going to need major overhauls for sychronized and optimized multi-threading and I just wonder how long it is going to take.
 
Yea your looking at alot of things having to change . Theres really no mainstream way to get much more out of 1 single processor that will be stable for the everyday user. Thats why i think the future will bring dual and even quad piped processores able to double , triple or even quadruiple todays highspeed standards even more possibly.If i was to estimate the time for us to get to that point i think it would be safe to say at least 8-10 years once the best way has been determined . Right now thats what there contemplating.Its really hard to say unless you live and breath in the high quarters of the developers . gee i wish i had chosen this fieled in highschool :(
 
yeah, devs are going to have to code their games to take advantage of dual processors, but I do think that's the way to go.
 
Does anybody know or can you explain the obstacles that designing apps/games for multi-threading might present?

Also, does anybody have any ideas of future technologies that look promising for greater speed?
 
Nanotechnology is the future. We're using it right now. Applied in general computing, we should see no problems in accelerating computer hardware in the future. I think the transition will go smoothly when the dual-core comes out. Just like the Athlon XP processors and the Athlon64 processors. Despite not a whole lot of applications designed for 64-bit, it can still benefit from an efficient way of processing 32-bit applcations. I believe the same thing will go for the dual-core once out. Perhaps the dual core with supporting chipsets will increase performance in some way, but until written codes in software and other technolgy can take FULL advantage of it, it will have to maximize the efficiency of existing software just as the Athlon64 is doing. Gamers and enthusiast will take interest so it should be ok.
 
That is the problem though, dual cores will probably not run existing software better than a faster single core chip that already exists today. You had mentioned the Athlon 64 but being 64bit has nothing to do with it running games faster. It is just the design of the chip, primarily the memory controller and increased cache, that allow it to play current day 32 bit games faster than other processors. I doubt dual core processors will have any new tech other than being dual core to see a performance jump in single threaded games.

Systems are already built with multiple cpu's just not on the same chip. However, these multipe cpu systems run games no faster than the single cpu systems.

That being the case, the only thing that is going to drive consumers to dual-core chips is software that takes advantage of it. If it takes two or three years for this to happen, why should a consumer upgrade to dual core?

If I was AMD or Intel, I would be paying some cash to a proven development house to work on a killer app/game that is multi-threaded. Even a new 3dmark benchmark demo would suffice. This way people could see the benefits of upgrading now from real benchmarks instead of theoretical hopes that are based off of software yet developed.
 
Status
Not open for further replies.
Back
Top Bottom