 |
| Dark and Sexy, that's how they all look like! |
Hello Everyone!
Today's article focuses on a very HOT topic! GRAPHIC CARDS! Who doesn't want the very best silicon out there to ramp up the effects and smoke the latest titles like Batman: Arkham Origins, Battlefield 4, Splinter Cell, Crysis 3, The Witcher and tons of other games out there! So if you're going to bake in a new graphic card into your Computer chassis, this guide will guide you all you need to know before buying a new Graphic Card! Ride Along!
Starting the article, I'd like to answer a few questions for you to ensure whether this guide is for you or not. Since I'm writing for you to get the answers, I think I'd do well If I'm considerate to your questions :)
Q1) AH WELL! INTEGRATED GRAPHICS CARD OR DEDICATED GRAPHICS CARD?
Now, here! As I started my article, if you are going to play all the latest games with all the latest bells and whistles enabled, you DO need a dedicated graphics card! And if you're one of those, continue further! But...hold on a second!
Q2) I'm a professional doing all kinds of professional CAD and... (blah blah blah). Which Graphic Card should I buy?
Even if you're a game developer, you'd see nVidia pasted on many more of the game screens (see WatchDogs or Batman for example) than AMD. So It'd be beneficial if you're developing what the world (or majority of it) is developing for. Though since my article is not aimed for you guys, I'd likely switch along.
Q3) HEY! ERR...DOES THE PRICE REALLY MATTER?
Well..I knew it'd come to this. This is the big bad ugly truth of Graphic Card. The costlier, the better. There's nothing to go against the fact that the more expensive your graphic card is, the better it'd perform. And to further it up, there are no "BEST IN THE WORLD"s available for less than 30,000 Rupees here ($500 or more) also, with the nVidia Titan, 780 Ti and AMD R9 295X2 rocking the world at the time of writing this (people from future, don’t laugh at these cards :p). So basically, by this time, we can have one thing settled up, buy the graphic card that best comes in your budget, or in other words, "Best Bang for the Buck".
Unless you have a lot of ample cash lying around.
Q4) GREEN TEAM OR RED TEAM? (better said, nVidia or AMD)
Mac vs PC. Microsoft vs Google. Intel vs AMD.
These are some of the best known rivalries in the tech world. And the one we’re gonna see next isn't any lesser. nVidia vs AMD (after AMD bought ATI back in 2006) is the hottest battle in the world of Graphic Cards! Well for most of the times, this question brings out the best of rivalries out of people with the kind of arguments people make, most of it absurd.
Before I tell out the advantages each of them has, I’d like to say it personally that for lower to mid range graphic cards, AMD is the best because of their philosophy of bring the best bang for the buck cards. However, if you tend to move up towards the best single card graphics cards, nVidia I think would be the better choice for you because of all of the technologies they tend to offer.
There was sometime back a notion that AMD Drivers were under-performing as compared to nVidia drivers, but the myth is dead now.
Also, in titles which feature “The way its meant to be played: nVidia”, there have been quite a number of reports that AMD cards have been found to under-perform in those titles (WatchDogs an example). Many of the games feature involvement of nVidia GameWorks that is a specialized unit of nVidia that collaborates with the game developers and helps provide the best visual experience. Nothing we can really do about it, AMD Cards do perform bad in those titles, but just that something I wanted to tell you.
Technology wise, nVidia vs AMD goes as following:
- nVidia has CUDA Architecture (gives impressive parallel computing advantage) while AMD has GCN Architecture (Graphics Core Next, amazing for Coin Miners)
- nVidia has SLI (Scalable Link Interface) which allows multiple cards of the same type to be run simultaneously vs AMD’s Crossfire Tech
- nVidia has PhysX (which enables IMPRESSIVE particle physics) which is unique to nVidia
- Mantle, which is an API that promises better performance as compared to DirectX, unique to AMD.
So with this, lets get started with the actual guide for you to understand those numbers!
To Round Up, I’d discuss the following terms in this guide:
- GPU clock frequency
- GPU shader clock
- Number of shaders/ stream processors/ cores
- Memory clock frequency
- Memory Bus Width
- Memory type
- Memory bandwidth
- Memory size
- Power and Heat Output Constraints
1) GPU Clock Frequency
The only two numbers with MHz or GHz you’d find on a graphic card review would be GPU Clock Frequency and a Memory Clock Frequency. We’re talking about the former here. The general rule is the higher the better. Moreover this is one of the parameters that can be overclocked, so its better to get a graphic card with a high stock core clock.
2) GPU Shader Clock
GPU Shader Clock is the number which tells the speed at which the GPU can render, do geometry calculations regarding polygons and a variety of other things. Shaders are special programs that calculate how the light will reflect off surrounding in a frame. Hence, they perform “shading”. Therefore, the faster they are, the better.
3) Number of Shaders/ Stream Processors/ Cores Shaders
Stream Processors and cores are different words for the same thing. The product lines of AMD and NVidia have a lot of differences in architecture, the most important being:
- nVidia has fewer cores but they are capable of more complex tasks than the AMD cores.
- AMD has more cores that can significantly speed up some calculations but are less flexible.
4) Memory Clock Frequency
Imagine you’re driving on a highway. Then memory clock frequency would determine how fast the traffic is moving on the highway. Memory clock frequency tells us the rate or speed of data transfers that occur in a unit time. Hence surely, more the frequency, the better it is.
5) Memory Bus Width
Keeping up with the same example taken above, the wider the highway the better it’d be and the fewer jams would occur. Size of the Memory Bus refers to the width of our hypothetical highway. The bigger the memory bus would be, more data can pass in a unit time through it.
A general rule-of thumb should be as follows:
A general rule-of thumb should be as follows:
- All low-midrange cards use 128-192bit bus.
- Most decent mid-range cards use a 256bit bus.
- See a number higher than this in your graphic card? You can be sure you’re buying one of the best pieces of equipment out there. More likely a 320bit bus or more.
Being one of the most important factors in any graphic card purchase, let me tell you the recipe of why do we have only these particular sizes of memory buses available. If you ever looked over the PCB of a graphic card, it might look like this
 |
| nVidia GTX 580 PCB |
Now to help you explain best, I’d like to show you this excerpt from enthusiastpc.net
“The large NVidia chip in the middle is the GPU (in this case a GF110). The 12 smaller chips that surround it are the videoram chips. The GTX 580 has a memory bus width of 384 bits and each GDDR5 chip has a bus width of 64 bits. To get to 384 bits (6 x 64) the card thus needs a multiple of 6 GDDR5 chips to get to this bandwidth. This is the reason why the GTX 580 comes in 1.5GB (128MB per chip x 12) or 3GB (256MB per chip x 12). It is also the reason that the GTX 560 comes in either 1GB or 2GB and the GTX 570 in 1.25GB or 2.5GB. This has to do with the physical number of chips needed to get to the bus width of 256 bit (GTX 560) and 320 bit (GTX 570).”
The memory type for all current graphic cards is GDDR5. Older or cheap graphic cards come with DDR3 or GDDR3 but this is no longer the standard. GDDR5 is currently the fastest graphic card memory on the market, so make sure your card comes with the same.
If I could tell you which two things to look for while buying a Graphic Card, I’d say Memory Bandwidth and the GPU. Memory Bandwidth is the highway we talked of while studying the earlier parameters. Just like GPU performance is made from the numbers: GPU Core Clock and Number of cores, Memory Bandwidth is made from the numbers: Memory Bus Width and Memory Clock Speed. Let me explain this to you.
The GTX 580 from nVidia has a memory clock of 1002MHz of GGDR5 memory. Since a GDDR5 memory can do 4 data transfers per clock cycle, this memory delivers a data-rate of 4008MHz. Each of these transfers are of 384 bits. This means that the bandwidth of the memory is 384 x 4008 = 1,539,072 million bits per second. To get from bits to bytes we divide the number by 8 leading to 1,539,072 / 8 = 192,384 million bytes per second or 192.4 GB/s. This final outcome is the real memory performance figure.
My general advice to everyone here would be, dont care what the bus is, dont care what the memory size is, just see the memory bandwidth for comparing the memory aspect of two cards and go with the one with the bigger number.
Memory Size is a common parameter used to forge people into buying cheap graphic cards. For now and for ever, a 2gb Graphic card is NOT necessarily/ always better than a 1Gb card. Manufacturers of cheap cards often provide large memory size coupled with meager bus width size of old memory type. Hence at the end of the day, the card can never use the full amount of memory size its been given. As a general rule of thumb, buy cards with at least 2Gb memory. The more is the better, especially if you’re buying a card for quite a couple of years. Setting up for multi-monitor gaming? You’d need AT LEAST a 4Gb card. The Titans from nVidia are pretty awesome at that. Do check them out.
Basically, you cant buy just any graphic card to put into your cabinet. Another component of your Computer called the PSU (Power Supply Unit) is responsible for powering up the computer. Nowadays the mid-range graphic cards demand a minimum of 500-600W PSU with the highest performing cards needing at least a 1000W PSU. If you’re into SLI/CrossFire then you’d need even more. So you’d need to look out for your PSU before you make your choice into buying a graphic card, unless you want to burn your house down.
Keeping up with this, as a general rule, the more power-hungry a card is, the hotter it is supposed to run. You need to keep in mind if you haven't got a tower for a CPU but a min cabinet, you need to make sure that you have enough ventilation with fans. Your cooling system should be of a good quality, if you're doing SLI/CrossFire, a water cooled system is recommended. Having a sufficiently cooled system will also provide you with overclocking avenues. Hence make sure you double-check the power requirements and heat output of the card before you buy it. Most stock coolers don't cool the card enough for overclocking, though the custom coolers do a very great job at cooling the Graphic Cards, a few of them being MSI, Asus, Sapphire (especially Toxic versions), etc.
And hence I finish with my Graphic Card Purchase Guide for you! I hope you liked it! Please subscribe if you want to remain updated with more of my posts! Thanks for your time! Take care :)
“The large NVidia chip in the middle is the GPU (in this case a GF110). The 12 smaller chips that surround it are the videoram chips. The GTX 580 has a memory bus width of 384 bits and each GDDR5 chip has a bus width of 64 bits. To get to 384 bits (6 x 64) the card thus needs a multiple of 6 GDDR5 chips to get to this bandwidth. This is the reason why the GTX 580 comes in 1.5GB (128MB per chip x 12) or 3GB (256MB per chip x 12). It is also the reason that the GTX 560 comes in either 1GB or 2GB and the GTX 570 in 1.25GB or 2.5GB. This has to do with the physical number of chips needed to get to the bus width of 256 bit (GTX 560) and 320 bit (GTX 570).”
6) Memory Type
The memory type for all current graphic cards is GDDR5. Older or cheap graphic cards come with DDR3 or GDDR3 but this is no longer the standard. GDDR5 is currently the fastest graphic card memory on the market, so make sure your card comes with the same.
7) Memory Bandwidth
If I could tell you which two things to look for while buying a Graphic Card, I’d say Memory Bandwidth and the GPU. Memory Bandwidth is the highway we talked of while studying the earlier parameters. Just like GPU performance is made from the numbers: GPU Core Clock and Number of cores, Memory Bandwidth is made from the numbers: Memory Bus Width and Memory Clock Speed. Let me explain this to you.
The GTX 580 from nVidia has a memory clock of 1002MHz of GGDR5 memory. Since a GDDR5 memory can do 4 data transfers per clock cycle, this memory delivers a data-rate of 4008MHz. Each of these transfers are of 384 bits. This means that the bandwidth of the memory is 384 x 4008 = 1,539,072 million bits per second. To get from bits to bytes we divide the number by 8 leading to 1,539,072 / 8 = 192,384 million bytes per second or 192.4 GB/s. This final outcome is the real memory performance figure.
My general advice to everyone here would be, dont care what the bus is, dont care what the memory size is, just see the memory bandwidth for comparing the memory aspect of two cards and go with the one with the bigger number.
8) Memory Size
Memory Size is a common parameter used to forge people into buying cheap graphic cards. For now and for ever, a 2gb Graphic card is NOT necessarily/ always better than a 1Gb card. Manufacturers of cheap cards often provide large memory size coupled with meager bus width size of old memory type. Hence at the end of the day, the card can never use the full amount of memory size its been given. As a general rule of thumb, buy cards with at least 2Gb memory. The more is the better, especially if you’re buying a card for quite a couple of years. Setting up for multi-monitor gaming? You’d need AT LEAST a 4Gb card. The Titans from nVidia are pretty awesome at that. Do check them out.
9) Power and Heat Output Constraints
Basically, you cant buy just any graphic card to put into your cabinet. Another component of your Computer called the PSU (Power Supply Unit) is responsible for powering up the computer. Nowadays the mid-range graphic cards demand a minimum of 500-600W PSU with the highest performing cards needing at least a 1000W PSU. If you’re into SLI/CrossFire then you’d need even more. So you’d need to look out for your PSU before you make your choice into buying a graphic card, unless you want to burn your house down.
Keeping up with this, as a general rule, the more power-hungry a card is, the hotter it is supposed to run. You need to keep in mind if you haven't got a tower for a CPU but a min cabinet, you need to make sure that you have enough ventilation with fans. Your cooling system should be of a good quality, if you're doing SLI/CrossFire, a water cooled system is recommended. Having a sufficiently cooled system will also provide you with overclocking avenues. Hence make sure you double-check the power requirements and heat output of the card before you buy it. Most stock coolers don't cool the card enough for overclocking, though the custom coolers do a very great job at cooling the Graphic Cards, a few of them being MSI, Asus, Sapphire (especially Toxic versions), etc.
And hence I finish with my Graphic Card Purchase Guide for you! I hope you liked it! Please subscribe if you want to remain updated with more of my posts! Thanks for your time! Take care :)
0 comments:
Post a Comment