Heres a resolution for ya maybe. Theres 2 things you want to look for when deciding to overclock. First there is the memory clock and then there is stream processes.

I am using this video card for an example.

Newegg.com - MSI N9800GT-T2D512-OC V2 GeForce 9800 GT 512MB 256-bit GDDR3 PCI Express 2.0 x16 HDCP Ready SLI Supported Video Card - Desktop Graphics / Video Cards

This video card is good for what you want and is a fair price you wont have to upgrade it for a few years and it'll play just about every game out so far just by itself. Maybe not on uber graphics but it will suffice for casual gaming.

Ok so this card has a memory clock of 1900 mghz. That explains itself all in one. an average memory clock you wanna look for is about 800 to 1,000 to play standard games and older games. however newer games require higher memory clocks....because why? because their is more graphical teqniques in newer games like bump mapping and triangular mapping or double and even triple pixel rending/shading. Ok so we got the memory clock it can handle the graphics load but there is also the stream processes

Stream Processes tells you how many processes it can do all at once. This card has 112 stream processes which means it can do 112 things at once. some games have more than that but you only see about a handful of things for example if I"m playing an fps I might only use 75 stream processes but if I am playing an rpg where you run around you'll use about 150 stream processes.

All in all the moral of the story is try to stay away from overclocking and I hope my advice will help you out on picking a new card if you chose to do so. Good luck on this.

I'm terribly sorry but the information you're presenting is incorrect.

First of all, although memory clock does have an impact on performance yes, but whether or not having a certain memory clock speed will change if you can/cannot play a game, that is untrue.

There are lots and lots of factors.

First of all you have memory size, the more memory you have, the more graphic data can be stored to be processed and/or stored to be re-used and save on reprocessing.

Second, there's the memory interface bus, which goes from 32bit for older cards up to 512bit on newer cards, the higher the number, the faster the memory interface. If you have 1Gb of video memory but a 64bit interface, and another card has 256Mbs of memory and a 128bit interface, it'll perform better than the 1Gb card.

Third, memory clock, the faster the memory, the faster it can read/write to it, it'll still be limited by the interface and GPU though. However, because the card is advertised at 1,900Mhz, doesn't mean it's really that fast, you have to divide it in half to get the actual memory clock.

Fourth, the GPU, the more stream processors you have, the more bandwidth and processing power you have access to. A card will ALWAYS utilize 100% of it's stream processors at all times unless it's in power savings mode or it's overheating.

Fifth, stream processor clock speed, the faster the speed, the more performance you get obviously.

Sixth, Shader clock speed, self explanatory, just like the stream clock. Shader and stream clock speeds are usually linked.

Seventh, interface, a card running on PCIex4 will have less bandwitdh than a card on PCIex16.


All of these are inter-related and each has their own dependencies, etc. You can look up most of this stuff with Google if you'd like to know more

I don't mean to turn this into an arguement however I know what I am talking about I have built 7 of my own gaming rigs just off those 2 choices and have come out on the top. If you realize when you look at memory clock and stream processes the rest of what you mentioned are usually pretty high. So as you may be right that those other factors are important 98% of the time they are top notch with the 2 things I suggested to look for.

I'm not saying that the more the processors and the higher the memory clock doesn't mean more performance, I agree with that as it's one of the major factors, especially the amount of stream processors. However, the rest of the information you provided is inaccurate, specifically how the stream processors are used.

You can read more about stream processors here if you'd like, it's a very interesting read:

Stream processing - Wikipedia, the free encyclopedia


Not sure if you've read up on how the new nVidia cards now support PhysX but the way it works is it takes a portion of the stream processors and assigns them to physics rendering. So stream processors are necessarily used independently, they can be grouped together to process large amounts of data together, or they can be assigned individually for smaller amounts of data.

well that would be why our info is so different. I use Ati Cards mostly I have only had 2 nvidia cards.

Doesn't change anything really, same processors technology more or less, both work the same way, all processors will be used at all times unless it's in power savings mode or it's overheating.

The only difference is that ATI cards don't have hardware physics acceleration at this time in the form of PhysX.

To all who are reading this, this discussion was begun in another thread, and became a more involved debate that deserved its own thread. If you would like to review the referred thread, you can read that here,
Chicken Nuggets, should I Overclock?

And now, back to the discussion.

TTFN

LGW