What did you think about Digital Storms article concerning hardware and wargaming we published recently?

Found HERE

It was pretty good overall- the only real gap in it was recommending quad core cpu's. To this day a duo core will still outperform a quad core conciderably- and this will always be the case until games start taking advantage of more than a single core. Remember the overall GHZ rating of the CPU is a combo of all cores. So a game that runs on a duo core rated at 3GHZ is getting a core running at 1.5ghz whereas a quad rated at 3ghz is only providing half that at 750 MHZ.

Very few games run on more than one core. Less than 1 percent. And that's not gonna change any time soon.

Same for memory. A jump from 4 to 8 gigs won't be noticable in most games that usually only alot 1 gig at best to begin with. A few games it's possible to tweak an ini. file for more- but for the most part more memory on the GPU will do you better than more system memory if your sitting at 4 gigs already because few games are written towards 64 bits.

A developer will benifit as 3dmax and the like will enjoy 8 gig of ram but few have these expensive programs to justify such a system.

Originally Posted By ghostryder (9 AUG 2012 6:16pm)

Remember the overall GHZ rating of the CPU is a combo of all cores. So a game that runs on a duo core rated at 3GHZ is getting a core running at 1.5ghz whereas a quad rated at 3ghz is only providing half that at 750 MHZ.

Could you post a link that supports this claim please?

And a link that proves your claim that a duo out perfroms a quad core of the same rated speed.

This is pretty old news but let's see what I can find. generally the tech site I find most knowledgable and dependable is Anatech. There was an article on this there so I'll try- but a few more sites have them. Remember overall games are written fast and dirty and It is not a trivial matter to code multi threaded programs. We can't even get them to devote more than a couple dev weeks on A.I.- they are going to be dragged kicking and screaming to code towards multicore systems. Especially in our nich area where most devs are using engines and tools completely uncapable of the tasks.

On the flip side however the MHZ wall was hit long ago- so overall aside from benchmarks no one's going to really notice the differences in most situations. I mention this only because there's a myraid of tech sites that regularly post bogus information and also to hold onto your wallet.

http://www.codinghorror.com/blog/2007/09/choosing-dual-or-quad-core.html

http://answers.yahoo.com/question/index?qid=20100225180351AA7dQMX

It's a faint difference but it does highlight an important point - even the heaviest of multitaskers aren't likely to see a huge difference in system responsiveness with four cores vs. two. It is in the performance of individual threads, especially when you have a lot that are contending for CPU time, that you'll see a much bigger difference between Kentsfield and Conroe.

Architecturally, the two individual die that make up Kentsfield are treated as one. This distinction matters especially when you look at power consumption of the new processor. In the event that only one of the two die is being used, the other cannot simply power down; instead it has to run at the same voltage and frequency of the other active die. Within each die, the two cores have to run at the same frequency and voltage as well, so there's not really much flexibility in operation. The end result is that while Kentsfield doesn't use too much more power than Conroe when running four CPU intensive threads simultaneously given the additional work it is doing, when only running two threads, Kentsfield is quite wasteful with its power consumption.

We have not yet hit the point of diminishing returns with adding more cores, but the two downward trending curves show the current problem with scaling from two to four cores. Very few desktop applications can take good advantage of a dual core CPU and even fewer are geared for quad core chips, and in those applications these quad core processors are actually a step back in terms of efficiency; it's not a fault of the processor, but rather of the software.

Through the benchmarks on the previous page we also highlighted another important point to take away: even in the areas where performance and efficiency were both improved, they came at a very expensive cost: twice the transistor count of Core 2 Duo. As much as Intel may call the next decade the era of multi-core and thread level parallelism, we simply can't lose focus on improving instruction level parallelism and core level efficiency.

http://www.anandtech.com/show/2112/6

Now over the years since efficiency has increased in the cores and the benchmarks narrow but the cost to performance remains- and pretty much the GPU is the bottlneck instead of the CPU when it comes to games. Therefore it is logical and quite good advice to spend money on GPU memory. A 1 gig GPU card in 4g system is going to outperform a system with 512 on the GPU and 8 g of ram.

From anatech this is an honest statment. I just have trouble with sites that make laughable comments- Overclockers and Tom's been doing it forever it seems.

Despite the relatively gung-ho attitude by Remedy, Epic and other ISVs on being able to use more than two cores regularly, the fact of the matter is that neither Alan Wake nor UT2007 are coming out anytime soon. And it's tough to say how long before other titles actually show a noticeable performance or experience difference between two and four core CPUs. Our pragmatic side can't help but point out that if you're building a high end gaming PC today, going quad-core will help in the future, but we can't really say precisely when that future will be.

There are some things that quad core is much better at though. Rendering Videos is one example. Rendering an already captured video is much more CPU intensive than anything and the ability of a Quad Core to render 4 seperate packets of the file at once gives it a large advantage. I have seen a huge increase in performance since upgrading to my rather modest i5 2400 when it comes to rendering videos. If you run a test you will typically see when rendering the CPU's are running at 70%+

...

Oh Hi Wargamers!!! This is my first post!

Yes you get benefits. But remember this is an article aimed at gamers and not home movie producers. Both sides of the coin need to be explained.

If the myth- even to this day some still think, "gee if I have one core at 3 GHZ and go to 2 then I'll be going twice as fast--as I believe both cores are running at 3 GHZ. Gee I'll go 4 times faster with 4 cores!" WRONG! 4 cores at 3 GHZ would produce the same heat as 1 core running at 12GHZ! It would melt through the computer case. Each core in fact only runs at a quarter speed of the rated GHZ. So a quad core at 3GHZ means each individual core is running at 750 MHZ. It's not even getting to 1 GHZ.

So if your game only supports one core--which is true 99 percent of the time---your getting a processor running 1/4 speed of a single core at 3 GHZ.

You'd think this would be crippling to games--and when dual cores first hit it was. But cach sizes, bus handling and controller chips have improved- and the fact other things that are running can be helped by the other cores, it isn't much of a hit anyone would notice- and because the real bottleneck still lies in the GPU the argument is rather mutt overall except to those interested in retro gaming- where going to a quad, or a 64 bit 8 m ram system could kill their ability to get those older games to run at all.

Myself for example. I use development software that would love 8 megs or more of ram. Rendering scenes in 3d MAX, Poser or Daz studio, my audio mixing software all would perform better.

However I'm dev'ing a follow up to Emperor of the Fading Suns. I need to actively play the original to keep it's gameplay fresh in my minmd---and i absolutely cannot get it to run on a quad core or a 64bit Windows OS. I'm a former Microsoft Engineer that knows every trick in trhe book and I can't get it to run on anything but a 32 bit system with 4 megs of ram.

Until I can afford a second computer then I live on 32 bit win 7.

Gamers need to know that before one blindly gives such recommendations to them. If you system is solely a gaming system overall your better served , and your save money- sticking to dual cores and 4 megs of ram. Spend the savings on the GPU.

ghostryder, the only place that your way of measuring core speeds get used is by dodgy dealers on ebay. To be clear, a 3GHz quad core processor has 4 cores running at 3 GHZ each, whilst a a 3GHz dual core processor has 2 cores running at 3 GHz each.

....mmmm... I want to hear more about this. For years GR has been right on the money about most of this stuff. My experiences with Skyrim seem to bear out his position vis-a-vis quad/dual for a given GPU.

ghostryder, the only place that your way of measuring core speeds get used is by dodgy dealers on ebay

or articles on anatech by Electronic computer Engineers---which BTW I am one. Quad cores do not run at full speed on each core-as the anatech article clearly shows. In fact quad cores are poor overclockers because every voltage increase translates to far more than 4x the heat. You can OC a dual core much more- and for games the performance of that single thread is everything. The chips MHZ rating is a combination of all cores. it's always been that way. The promise was never that the cores would run faster but rather that the multi-tasking of the cores would increase speed. this however-much the same as the Pentiums old MMX- is software dependant. Most didn't see it with MMX and most do not see it today in current games.

I believe many people get confused by the way the chips themselves are made--which can be many ways. 4 cores on one die or a core on 4 seperate dies-and to complicate things even more does each core have it's own cache or does it share it with the other cores, ect.

Internally as the anatech article points out in lue of efficiency it notes the main inability to adjust voltage of the individual cores. If they were designed that way then when a single core program ran the other cores could shut down and the voltage could be increased to OC the one core being used.

Physics btw is the sole reason for the MHZ wall. when you move electrons through a transister the positive tempature coiefient magnifies the heat generated in relation to the speed the electron is moving. In simple terms this means as a dumb down explanation say your at 1 ghz. and you up it to 2 ghz. simple math says you'd double the heat generated-but transistors within the CPU have a posative tempature coefficient (hence the the name semi-conductor. there's a resistance that generates heat as electrons move through. It is what is wrong with solar and why that tecnology will never compete. It's semi conductor iniffeciency pared with more loss as you convert DC to AC. Supporters ignore the real physics when they try to sell Solar) that magnifies the heat in relation to speed- so the actual result is nearly 4 times the heat instead of twice the heat. The same thing occurred in the 70s with aluminum wiring. Builder's replaced the copper wiring with aluiminum wiring. But aluminum has a higher posative tempature coefficient and the 60hz cycle produced so much heat houses caught fire and the wiring then was later outlawed.

therefore if you place 2 single core processors side by side the posative tempature coeffienct doesn't result in doubling the heat generated but magnifies it to almost 4. when your going quad core you again doubling the transister count. now the 4 cores are generating 8 times the heat. And this is the reason quads are clocked slower.

The main reason the individual voltages of the cores can't be changed as the article points out is because it is often unclear what program should take presedence. Your game may be single core but you online processes that run WOW or whatever are calling the other cores. There's no real way for the OS nor the processor to monitor multiple processes in such a way that it can then regulate indivual cores independantly.

Another way to look at it in debunking the myth that 4 cores results in 4 times the speed is to assume one horse runs at 30 MPH. 4 horses side by side don't equal 120mph. You simply have 4 horses running at 30 MPH. But in terms of heat generation that equals 8 times the heat of one so the overall cores are slowed to deal with the heat. Or think of the horses then facing a strong wind running down a narrow path fighting each other for room slowing them down. they then add the cores together as a marketing trick to rate the chip - this happens all the time in electronics. That Energy star sticker on that TV is pretty much a joke. Those ratings are when the TV is turned off (ala standby) -- turn that puppy on and it pulls more wattage than an old admiral CRT 32 inch. In the ways they measure them if the cores actually ran at full speed they'd be stamping 12mhz all over the box- and yes I've seen those claims on ebay.

add in the mix of Amdahl's law and you get even more diminishing returns. Nothing is 100 percent parallel. There always a portion of the software that just uses a single core. What the law says is this. Say it takes an hour to compress an audio file or whatever on a single core. even if written for 4 cores the end result isn't 15 minutes on a quad core. Amdahl's graphs were written in the beginning of multple cores so i believe they are inaccurate today but basically it's saying your never get below 30 minutes - this is simplified. the technical jardon gives even me a headache. But overall the performance wall is there so your never see mutitasking getting to 4 times the performance.

So, the basis of your point is that the speed of each individual core is the total speed of the processor divided by the number of cores. So, a dual core 3.0 would run at 1.5 on each core while a quad core 3.0 would run at .75 on each core right?

So, if a program only utilizes 2 cores, the quad core is inherently disadvantaged because it would only run at half the speed, right?

My only problem with this point GhostRyder, is that the article you quote (AnandTech) verbatem was written in 2006 for one, and for two, never states ANYTHING like what you state.

Instead, it says things like the following...

Yes as I stated it was old news. There are more articles related-the trick is finding the right keywords in finding them. But even looking at it as no inherient loss is really wishful thinking as well. You are PAYING for double the transistors--so cost to performance there is a penalty- and the most important aspect of all- the ability to retro game wasn't addressed at all.

We are wargamers. Many retired and on a budget and many that still play Age of rifles, Steel panthers and so on. Any article geared to that demographic needs to point out two things. One. Your be paying more and it may not bring you any benefits and two you may not be able to run any of your older games that do not play on 64 bit setups.

Even after 6 years multicore support hasn't made much progress in terms of games. And going to 64 bit for the recommended ram is probably the worse thing many could do in terms of retrogaming. Games overall probably will be 32 bit single threaded for as long as the developers can hold out. If it cost time and money they just won't do it- and it does cost most of all time. And as mentioned coding toward it is no simple task.

Your only see it from large companies that license AAA engines at high dollar. Most games we play are produced on the barest of budgets given the barest of development time using the barest of IDEs. If there was a course of study i wish these dev's would take it wouldn't be 'programing to multicores' but rather 'How to program A.I.'