Getting something for nothing - yes that does exist

[Borealis7]

how is this picture relevant to CPU performance??

 

[taq8ojh]

Guys it's a troll. The last picture proves that. Let's not feed it.

 

[Idontcare]

But seriously, are you trying to extol the virtue of Win8 by enunciating its native ability to commit "Death by Powerpoint" as a virtue? 😀

I typically judge the ability of a presenter as being inversely proportional to his/her use of PowerPoint in a presentation.

If you want to effectively train techies (which I have done with regard to Enterprise Disaster-Recovery Tech-Support in my capacity as Senior German Engineer) I have found the following to be very efficacious:

1) Do it in a pub - people learn a lot more in a relaxed environment

2) Don't EVER use PowerPoint

3) Don't allow them to bring out any pens and paper to take notes (if someone is writing then they are not listening).

4) Don't tolerate tourists.

5) DO go into as much detail as it takes for the people to comprehend the broader scheme of the information one wishes to impart - even if that means that a one hour talk which starts at 6:30 PM goes on to chucking out time in the aforementioned pub (this was the rule rather than the exception and my boys and girls left not with eyes glazed over with boredom but rather not realising where the time had gone because they found it enjoyable). Remember, context is everything.

6) DO be willing to digress because someone brings up something they have questions about even though it is not overtly part of the original topic. Because things in computing are inter-related, if you are a good techie then you can get the person from that question seamlessly back into the topic by showing him/her how that relates.

For instance an Exchange problem when one is talking about AD by pointing out that Active Directory was first introduced in Exchange 5.5 and it was so much more scalable than what Microsoft was planning for NT5 (which became Win2K) that it abandoned its plans to expand upon SMB which they had been using in their server environment.

It's a kind of "Six Degrees of Kevin Bacon"

There are many different types of learners, you can't legitimately argue that oral learning without visual learning is the sole superior model for communication and information transfer.

The irony here is that you seem to revel in having penalized those members in your audience who were visual learners by starving them of visual communication avenues during your "talks" while simultaneously preventing them from being able to help themselves because you insisted they not be allowed to take notes (yet another visual learning queue)...how tolerant and nurturing of you 🙄

And the ultimate irony here is that you are posting this "pro-oral" teaching methodology, in writing without actually saying a single word orally, on a forum that is utterly devoid of the oral communication pathway.

Powerpoint is a tool that is used to enable learning and communication for visual learners as well as individuals who need to absorb the information on a pace that differs from that of the presenter.

It is rather startling to see someone who claims to know as much as you claim to know, and yet somehow the rather simple concept that some people are visual learners (as opposed to oral learners) seems to have escaped your observation all your life 😕

 

[skipsneeky2]

I think most modern games are smart enough to use the cores of a quad core to its full potential,except for those few games like GTA4 which are just a mess.

Good reason why many games are pushing for quad cores and the days of a simple dual core like a core 2 duo just not enough,the fact we are still rocking quad cores for games nearly 6 years after the first title that benefited them came out is actually kinda disappointing.

 

[sm625]

You will get at least a 60% gaming performance boost from the CPU for free with absolutely no downside, no increased temperatures and no added instability by simply eliminating the Operating System interference with the execution of the game engine - what's not to like?

It is just not true. If you create two folders and put a copy of super-pi in each folder and run them both and start a 1M run on each one, you will get the same score no matter how you set the affinity. Unless of course you set the affinity on both to only one core and the same core. And even then it actually runs both faster than time x 2. I've tried core parking experiments with prime95 too and gotten negligable improvements from core parking.

 

[Nec_V20]

Shut him down someone, please. It's not amusing anymore. He seems to actually believe it.

I actually know it, and you don't know enough to comprehend it.

 

[Nec_V20]

It is just not true. If you create two folders and put a copy of super-pi in each folder and run them both and start a 1M run on each one, you will get the same score no matter how you set the affinity. Unless of course you set the affinity on both to only one core and the same core. And even then it actually runs both faster than time x 2. I've tried core parking experiments with prime95 too and gotten negligable improvements from core parking.

Have you done the same experiment on a low performance multi-core CPU running a gaming engine?

The answer to that would be no.

You are also equating something which makes heavy use of the floating point processor which is also not necessarily applicable to the gaming engine scenario I postulated.

 

[Puppies04]

So we have gone from...

"do this and you will get a 60% performance increase"

too...

"Get an underpowered CPU running windows and all the other random background processes one picks up as the years go by and set all those many processes to one core while leaving the other one free to play a game and you will get a 60% performance increase"

I agree with your idea in theory but a couple of points.

1. This is nothing new, many many people have known about it for years

2. You seem able to take a conversation off on a tangent like this one time when I was 6 and I got a new football for my birthday and my friend Dave wanted to play rugby, but alas that is another story.

 

[Nec_V20]

I think most modern games are smart enough to use the cores of a quad core to its full potential,except for those few games like GTA4 which are just a mess.

Good reason why many games are pushing for quad cores and the days of a simple dual core like a core 2 duo just not enough,the fact we are still rocking quad cores for games nearly 6 years after the first title that benefited them came out is actually kinda disappointing.

If however a "core 2 duo" is all you have and you want to make a game reasonable to play where otherwise it takes seconds to react between mouse clicks, what are you going to do? Buy a new computer?

What if, like my neighbour you just had a new child and the money is tight?

 

[skipsneeky2]

If however a "core 2 duo" is all you have and you want to make a game reasonable to play where otherwise it takes seconds to react between mouse clicks, what are you going to do? Buy a new computer?

What if, like my neighbour you just had a new child and the money is tight?

My first "gaming rig" was in 2006,it had a 2.4ghz 478 pentium 4,a asus p4c800-e deluxe motherboard i got at a swapmeet for $50 new in box,4gb ram and a 6200 le,BF2 was not fun on those specs and it took a year before i saved up and upgraded to a e6750+8800gts 512mb and yeah the old rig was a torture instrument.

If i had a new child and money was tight,my ass would not be here on the forums,that is for sure nor on the computer or thinking about what upgrades my computer would need to play any of the latest games.

 

[Cerb]

You are correct, with the types of games I like to play (C&C, Warhammer 40K, X3, AOE etc.) my CPU has more than enough muscle to cope without resorting to either overclocking or what I mentioned in my original post.

However Windows still goes out of its way to kick you in the nuts if you are trying to run a game engine on a weaker CPU.

What I suggested avoids the vagaries of Non-Uniform Memory Access (NUMA). It assigns the game engine the exclusive use of one core (or more) and its concomitant L2 cache and builds a fence around that core with a big sign saying "Keep Out" to all other processes which would otherwise dirty the cache.

All it does is pin that process to that logical CPU, so that it won't get scheduled on others. It is a known fix for that specific game crashing, often before a game can be started. The problem is not that the OS is giving all this overhead, but that C&C G:ZH was and is broken, but worked barely well enough on single-core CPUs to get out the door. You could have the same problems, or worse, on much newer hardware, as well.

Windows isn't going out of its way to do anything to you. The game's developers are the ones at fault, though at least there's an option to make it work.

The vagaries of NUMA are not applying at all. It's probably a screw-up with locks, an I/O thread getting restarted, or something like that. The worst any laptops ever got, TMK, were early dual-core Atoms, and Core 2 Quads, and both of those had special chip-to-chip links to minimize latency, when external IO was not required. No mobile dual-cores, except 1st-gen Atoms, would have any significant issues with swapping caches and cores. It would take something more, like a programmer assuming thread B couldn't be scheduled until thread A had stopped running, which might always end up 100% correct on a single-core, since there's only 1 time-slice to be had at a time, and a woken thread will typically have to wait on a previously-woken one from the same process.

I have no idea what you mean about Win 8. Like Vista, it's got a lot of good inside, we'll just have to wait a version or two to make use of it with an acceptable non-touch UI.

 

[Nec_V20]

So we have gone from...

"do this and you will get a 60% performance increase"

too...

"Get an underpowered CPU running windows and all the other random background processes one picks up as the years go by and set all those many processes to one core while leaving the other one free to play a game and you will get a 60% performance increase"

I agree with your idea in theory but a couple of points.

1. This is nothing new, many many people have known about it for years

2. You seem able to take a conversation off on a tangent like this one time when I was 6 and I got a new football for my birthday and my friend Dave wanted to play rugby, but alas that is another story.

In my case I have known about it for nearly two decades or 10100 years.

For the sake of brevity I condensed my original thoughts down to the essentials - originally when I wrote the post it was at least three times the length.

Now every attempt at a flame on this thread towards me has been based on the premise that what I posted originally was done in bad faith.

The tangents were not imposed by me but rather on me by posters who are obviously ignorant of or did not comprehend the subject matter which was discussed in that original post - you know, the one at the top of this thread.

I have however answered questions no matter in which bad faith and with which pejorative malice aforethought uttered. BTW did I mention that English is my second language? (Now that really was a non sequitur of the kind you were trying to accuse me of).

So your accusation which you have levelled is a classic case of the logical fallacy known as "post hoc ergo propter hoc".

Any informed reading of my original post would inescapably lead to the conclusion that in the premise of my post I was talking about buttressing a lower performance CPU with regard to increasing its gaming performance. You are then quoting answers I made arising out of questions regarding asking for clarification of that original post to try to discredit the original post per se.

In the one example I cited where the laptop running Command&Conquer Zero Hour went from a reaction to a mouseclick happening after a few seconds even at the lowest graphic settings to the gameplay being flowing with hardly any lag. What kind of percentage would you place on that? Exactly how many hundreds of percent?

Sure, overclocking benchmarks give you bragging rights - to that benchmark; but how long will the system run stably and how much of the lifespan are you subtracting from those components to achieve those benchmarks and how much have you castrated other components in the whole process?

And BTW overclocking is just another name for "voiding the warranty".

All I set out to do in my original post was to open up possibilities to a person which cost nothing and did not inflict any penalties for a higher performance he or she would otherwise never be able to achieve on a lower performance CPU.

 

[ViRGE]

I left this thread open in hopes of seeing some meaningful and insightful discussions. That has not happened. So this thread is closed.

On a side note, Nec_V20, we are (generally speaking) a "proof positive" forum. I.E. as BrightCandle puts it, "He who asserts must prove". These are after all the AnandTech forums. So go be like Anand and do some science. If you had provided rigorous and sound benchmarks alongside your commentary, this thread would have been far better for everyone involved. As it stands however you're making a somewhat wild claim (60%) without the necessary hard data to back it up.

-ViRGE